From d98076c7921b1c3e81854064968a690da21ec8bd Mon Sep 17 00:00:00 2001 From: Thomas Gruber Date: Fri, 25 Feb 2022 14:49:49 +0100 Subject: [PATCH 1/3] Merge current development version into main (#48) * DiskstatCollector: cast part_max_used metric to int * Add uint types to GangliaSink and LibgangliaSink * Use new sink instances to allow multiple of same sink type * Update sink README and SampleSink * Use new receiver instances to allow multiple of same receiver type * Fix metric scope in likwid configuration script * Mention likwid config script in LikwidCollector README * Refactor: Embed Init() into New() function * Refactor: Embed Init() into New() function * Fix: MetricReceiver uses uninitialized values, when initialization fails * Use Ganglia configuration (#44) * Copy all metric configurations from original Ganglia code * Use metric configurations from Ganglia for some metrics * Format value string also for known metrics * Numa-aware memstat collector (#45) * Add samples for collectors, sinks and receivers * Ping InfluxDB server after connecting to recognize faulty connections * Add sink for Prometheus monitoring system (#46) * Add sink for Prometheus monitoring system * Add prometheus sink to README * Add scraper for Prometheus clients (#47) Co-authored-by: Holger Obermaier Co-authored-by: Holger Obermaier <40787752+ho-ob@users.noreply.github.com> --- collectors/diskstatMetric.go | 2 +- collectors/likwidMetric.md | 38 +++- collectors/memstatMetric.go | 176 ++++++++++++------ collectors/sampleMetric.go | 81 +++++++++ collectors/sampleTimerMetric.go | 122 +++++++++++++ receivers/README.md | 22 +-- receivers/metricReceiver.go | 9 +- receivers/natsReceiver.go | 65 ++++--- receivers/prometheusReceiver.go | 122 +++++++++++++ receivers/prometheusReceiver.md | 27 +++ receivers/receiveManager.go | 12 +- receivers/sampleReceiver.go | 77 ++++++++ scripts/likwid_perfgroup_to_cc_config.py | 2 +- sinks/README.md | 20 +- sinks/gangliaCommon.go | 218 +++++++++++++++++++++- sinks/gangliaSink.go | 156 +++++++--------- sinks/httpSink.go | 102 +++++------ sinks/influxAsyncSink.go | 87 +++++---- sinks/influxSink.go | 51 ++++-- sinks/libgangliaSink.go | 221 ++++++++++------------- sinks/metricSink.go | 3 - sinks/natsSink.go | 52 +++--- sinks/prometheusSink.go | 199 ++++++++++++++++++++ sinks/prometheusSink.md | 23 +++ sinks/sampleSink.go | 62 +++++++ sinks/sinkManager.go | 19 +- sinks/stdoutSink.go | 58 +++--- 27 files changed, 1509 insertions(+), 517 deletions(-) create mode 100644 collectors/sampleMetric.go create mode 100644 collectors/sampleTimerMetric.go create mode 100644 receivers/prometheusReceiver.go create mode 100644 receivers/prometheusReceiver.md create mode 100644 receivers/sampleReceiver.go create mode 100644 sinks/prometheusSink.go create mode 100644 sinks/prometheusSink.md create mode 100644 sinks/sampleSink.go diff --git a/collectors/diskstatMetric.go b/collectors/diskstatMetric.go index 819a1ab..16c70ba 100644 --- a/collectors/diskstatMetric.go +++ b/collectors/diskstatMetric.go @@ -102,7 +102,7 @@ func (m *DiskstatCollector) Read(interval time.Duration, output chan lp.CCMetric part_max_used = perc } } - y, err := lp.New("part_max_used", map[string]string{"type": "node"}, m.meta, map[string]interface{}{"value": part_max_used}, time.Now()) + y, err := lp.New("part_max_used", map[string]string{"type": "node"}, m.meta, map[string]interface{}{"value": int(part_max_used)}, time.Now()) if err == nil { y.AddMeta("unit", "percent") output <- y diff --git a/collectors/likwidMetric.md b/collectors/likwidMetric.md index 8b5dee2..3ef51f3 100644 --- a/collectors/likwidMetric.md +++ b/collectors/likwidMetric.md @@ -4,7 +4,7 @@ The `likwid` collector is probably the most complicated collector. The LIKWID library is included as static library with *direct* access mode. The *direct* access mode is suitable if the daemon is executed by a root user. The static library does not contain the performance groups, so all information needs to be provided in the configuration. The `likwid` configuration consists of two parts, the "eventsets" and "globalmetrics": -- An event set list itself has two parts, the "events" and a set of derivable "metrics". Each of the "events" is a counter:event pair in LIKWID's syntax. The "metrics" are a list of formulas to derive the metric value from the measurements of the "events". Each metric has a name, the formula, a scope and a publish flag. A counter names can be used like variables in the formulas, so `PMC0+PMC1` sums the measurements for the both events configured in the counters `PMC0` and `PMC1`. The scope tells the Collector whether it is a metric for each hardware thread (`cpu`) or each CPU socket (`socket`). The last one is the publishing flag. It tells the collector whether a metric should be sent to the router. +- An event set list itself has two parts, the "events" and a set of derivable "metrics". Each of the "events" is a counter:event pair in LIKWID's syntax. The "metrics" are a list of formulas to derive the metric value from the measurements of the "events". Each metric has a name, the formula, a scope and a publish flag. Counter names can be used like variables in the formulas, so `PMC0+PMC1` sums the measurements for the both events configured in the counters `PMC0` and `PMC1`. The scope tells the Collector whether it is a metric for each hardware thread (`cpu`) or each CPU socket (`socket`). The last one is the publishing flag. It tells the collector whether a metric should be sent to the router. - The global metrics are metrics which require data from all event set measurements to be derived. The inputs are the metrics in the event sets. Similar to the metrics in the event sets, the global metrics are defined by a name, a formula, a scope and a publish flag. See event set metrics for details. The only difference is that there is no access to the raw event measurements anymore but only to the metrics. So, the idea is to derive a metric in the "eventsets" section and reuse it in the "globalmetrics" part. If you need a metric only for deriving the global metrics, disable forwarding of the event set metrics. **Be aware** that the combination might be misleading because the "behavior" of a metric changes over time and the multiple measurements might count different computing phases. Additional options: @@ -26,6 +26,42 @@ As a guideline: - All `PWRx` counters have scope `socket`, except `"PWR1" : "RAPL_CORE_ENERGY"` has `cpu` scope - All `DFCx` counters have scope `socket` +### Help with the configuration + +The configuration for the `likwid` collector is quite complicated. Most users don't use LIKWID with the event:counter notation but rely on the performance groups defined by the LIKWID team for each architecture. In order to help with the `likwid` collector configuration, we included a script `scripts/likwid_perfgroup_to_cc_config.py` that creates the configuration of an `eventset` from a performance group (using a LIKWID installation in `$PATH`): +``` +$ likwid-perfctr -i +[...] +short name: ICX +[...] +$ likwid-perfctr -a +[...] +MEM_DP +MEM +FLOPS_SP +CLOCK +[...] +$ scripts/likwid_perfgroup_to_cc_config.py ICX MEM_DP +{ + "events": { + "FIXC0": "INSTR_RETIRED_ANY", + "..." : "..." + }, + "metrics" : [ + { + "calc": "time", + "name": "Runtime (RDTSC) [s]", + "publish": true, + "scope": "hwthread" + }, + { + "..." : "..." + } + ] +} +``` + +You can copy this JSON and add it to the `eventsets` list. If you specify multiple event sets, you can add globally derived metrics in the extra `global_metrics` section with the metric names as variables. ### Example configuration diff --git a/collectors/memstatMetric.go b/collectors/memstatMetric.go index b6ef855..3998537 100644 --- a/collectors/memstatMetric.go +++ b/collectors/memstatMetric.go @@ -1,35 +1,76 @@ package collectors import ( + "bufio" "encoding/json" "errors" "fmt" - "io/ioutil" - "log" + "os" + "path/filepath" + "regexp" "strconv" "strings" "time" + cclog "github.com/ClusterCockpit/cc-metric-collector/internal/ccLogger" lp "github.com/ClusterCockpit/cc-metric-collector/internal/ccMetric" ) -const MEMSTATFILE = `/proc/meminfo` +const MEMSTATFILE = "/proc/meminfo" +const NUMA_MEMSTAT_BASE = "/sys/devices/system/node" type MemstatCollectorConfig struct { ExcludeMetrics []string `json:"exclude_metrics"` + NodeStats bool `json:"node_stats,omitempty"` + NumaStats bool `json:"numa_stats,omitempty"` +} + +type MemstatCollectorNode struct { + file string + tags map[string]string } type MemstatCollector struct { metricCollector - stats map[string]int64 - tags map[string]string - matches map[string]string - config MemstatCollectorConfig + stats map[string]int64 + tags map[string]string + matches map[string]string + config MemstatCollectorConfig + nodefiles map[int]MemstatCollectorNode +} + +func getStats(filename string) map[string]float64 { + stats := make(map[string]float64) + file, err := os.Open(filename) + if err != nil { + cclog.Error(err.Error()) + } + defer file.Close() + + scanner := bufio.NewScanner(file) + for scanner.Scan() { + line := scanner.Text() + linefields := strings.Fields(line) + if len(linefields) == 3 { + v, err := strconv.ParseFloat(linefields[1], 64) + if err == nil { + stats[strings.Trim(linefields[0], ":")] = v + } + } else if len(linefields) == 5 { + v, err := strconv.ParseFloat(linefields[3], 64) + if err == nil { + stats[strings.Trim(linefields[0], ":")] = v + } + } + } + return stats } func (m *MemstatCollector) Init(config json.RawMessage) error { var err error m.name = "MemstatCollector" + m.config.NodeStats = true + m.config.NumaStats = false if len(config) > 0 { err = json.Unmarshal(config, &m.config) if err != nil { @@ -40,7 +81,8 @@ func (m *MemstatCollector) Init(config json.RawMessage) error { m.stats = make(map[string]int64) m.matches = make(map[string]string) m.tags = map[string]string{"type": "node"} - matches := map[string]string{`MemTotal`: "mem_total", + matches := map[string]string{ + "MemTotal": "mem_total", "SwapTotal": "swap_total", "SReclaimable": "mem_sreclaimable", "Slab": "mem_slab", @@ -48,7 +90,9 @@ func (m *MemstatCollector) Init(config json.RawMessage) error { "Buffers": "mem_buffers", "Cached": "mem_cached", "MemAvailable": "mem_available", - "SwapFree": "swap_free"} + "SwapFree": "swap_free", + "MemShared": "mem_shared", + } for k, v := range matches { _, skip := stringArrayContains(m.config.ExcludeMetrics, k) if !skip { @@ -56,13 +100,44 @@ func (m *MemstatCollector) Init(config json.RawMessage) error { } } if len(m.matches) == 0 { - return errors.New("No metrics to collect") + return errors.New("no metrics to collect") } m.setup() - _, err = ioutil.ReadFile(string(MEMSTATFILE)) - if err == nil { - m.init = true + + if m.config.NodeStats { + if stats := getStats(MEMSTATFILE); len(stats) == 0 { + return fmt.Errorf("cannot read data from file %s", MEMSTATFILE) + } } + + if m.config.NumaStats { + globPattern := filepath.Join(NUMA_MEMSTAT_BASE, "node[0-9]*", "meminfo") + regex := regexp.MustCompile(filepath.Join(NUMA_MEMSTAT_BASE, "node(\\d+)", "meminfo")) + files, err := filepath.Glob(globPattern) + if err == nil { + m.nodefiles = make(map[int]MemstatCollectorNode) + for _, f := range files { + if stats := getStats(f); len(stats) == 0 { + return fmt.Errorf("cannot read data from file %s", f) + } + rematch := regex.FindStringSubmatch(f) + if len(rematch) == 2 { + id, err := strconv.Atoi(rematch[1]) + if err == nil { + f := MemstatCollectorNode{ + file: f, + tags: map[string]string{ + "type": "memoryDomain", + "type-id": fmt.Sprintf("%d", id), + }, + } + m.nodefiles[id] = f + } + } + } + } + } + m.init = true return err } @@ -71,56 +146,41 @@ func (m *MemstatCollector) Read(interval time.Duration, output chan lp.CCMetric) return } - buffer, err := ioutil.ReadFile(string(MEMSTATFILE)) - if err != nil { - log.Print(err) - return - } - - ll := strings.Split(string(buffer), "\n") - for _, line := range ll { - ls := strings.Split(line, `:`) - if len(ls) > 1 { - lv := strings.Fields(ls[1]) - m.stats[ls[0]], err = strconv.ParseInt(lv[0], 0, 64) + sendStats := func(stats map[string]float64, tags map[string]string) { + for match, name := range m.matches { + var value float64 = 0 + if v, ok := stats[match]; ok { + value = v + } + y, err := lp.New(name, tags, m.meta, map[string]interface{}{"value": value}, time.Now()) + if err == nil { + output <- y + } } - } - - if _, exists := m.stats[`MemTotal`]; !exists { - err = errors.New("Parse error") - log.Print(err) - return - } - - for match, name := range m.matches { - if _, exists := m.stats[match]; !exists { - err = fmt.Errorf("Parse error for %s : %s", match, name) - log.Print(err) - continue - } - y, err := lp.New(name, m.tags, m.meta, map[string]interface{}{"value": int(float64(m.stats[match]) * 1.0e-3)}, time.Now()) - if err == nil { - output <- y - } - } - - if _, free := m.stats[`MemFree`]; free { - if _, buffers := m.stats[`Buffers`]; buffers { - if _, cached := m.stats[`Cached`]; cached { - memUsed := m.stats[`MemTotal`] - (m.stats[`MemFree`] + m.stats[`Buffers`] + m.stats[`Cached`]) - _, skip := stringArrayContains(m.config.ExcludeMetrics, "mem_used") - y, err := lp.New("mem_used", m.tags, m.meta, map[string]interface{}{"value": int(float64(memUsed) * 1.0e-3)}, time.Now()) - if err == nil && !skip { - output <- y + if _, skip := stringArrayContains(m.config.ExcludeMetrics, "mem_used"); !skip { + if freeVal, free := stats["MemFree"]; free { + if bufVal, buffers := stats["Buffers"]; buffers { + if cacheVal, cached := stats["Cached"]; cached { + memUsed := stats["MemTotal"] - (freeVal + bufVal + cacheVal) + y, err := lp.New("mem_used", tags, m.meta, map[string]interface{}{"value": memUsed}, time.Now()) + if err == nil { + output <- y + } + } } } } } - if _, found := m.stats[`MemShared`]; found { - _, skip := stringArrayContains(m.config.ExcludeMetrics, "mem_shared") - y, err := lp.New("mem_shared", m.tags, m.meta, map[string]interface{}{"value": int(float64(m.stats[`MemShared`]) * 1.0e-3)}, time.Now()) - if err == nil && !skip { - output <- y + + if m.config.NodeStats { + nodestats := getStats(MEMSTATFILE) + sendStats(nodestats, m.tags) + } + + if m.config.NumaStats { + for _, nodeConf := range m.nodefiles { + stats := getStats(nodeConf.file) + sendStats(stats, nodeConf.tags) } } } diff --git a/collectors/sampleMetric.go b/collectors/sampleMetric.go new file mode 100644 index 0000000..cd3b4cc --- /dev/null +++ b/collectors/sampleMetric.go @@ -0,0 +1,81 @@ +package collectors + +import ( + "encoding/json" + "time" + + cclog "github.com/ClusterCockpit/cc-metric-collector/internal/ccLogger" + lp "github.com/ClusterCockpit/cc-metric-collector/internal/ccMetric" +) + +// These are the fields we read from the JSON configuration +type SampleCollectorConfig struct { + Interval string `json:"interval"` +} + +// This contains all variables we need during execution and the variables +// defined by metricCollector (name, init, ...) +type SampleCollector struct { + metricCollector + config SampleTimerCollectorConfig // the configuration structure + meta map[string]string // default meta information + tags map[string]string // default tags +} + +func (m *SampleCollector) Init(config json.RawMessage) error { + var err error = nil + // Always set the name early in Init() to use it in cclog.Component* functions + m.name = "InternalCollector" + // This is for later use, also call it early + m.setup() + // Define meta information sent with each metric + // (Can also be dynamic or this is the basic set with extension through AddMeta()) + m.meta = map[string]string{"source": m.name, "group": "SAMPLE"} + // Define tags sent with each metric + // The 'type' tag is always needed, it defines the granulatity of the metric + // node -> whole system + // socket -> CPU socket (requires socket ID as 'type-id' tag) + // cpu -> single CPU hardware thread (requires cpu ID as 'type-id' tag) + m.tags = map[string]string{"type": "node"} + // Read in the JSON configuration + if len(config) > 0 { + err = json.Unmarshal(config, &m.config) + if err != nil { + cclog.ComponentError(m.name, "Error reading config:", err.Error()) + return err + } + } + + // Set up everything that the collector requires during the Read() execution + // Check files required, test execution of some commands, create data structure + // for all topological entities (sockets, NUMA domains, ...) + // Return some useful error message in case of any failures + + // Set this flag only if everything is initialized properly, all required files exist, ... + m.init = true + return err +} + +func (m *SampleCollector) Read(interval time.Duration, output chan lp.CCMetric) { + // Create a sample metric + timestamp := time.Now() + + value := 1.0 + // If you want to measure something for a specific amout of time, use interval + // start := readState() + // time.Sleep(interval) + // stop := readState() + // value = (stop - start) / interval.Seconds() + + y, err := lp.New("sample_metric", m.tags, m.meta, map[string]interface{}{"value": value}, timestamp) + if err == nil { + // Send it to output channel + output <- y + } + +} + +func (m *SampleCollector) Close() { + // Unset flag + m.init = false +} diff --git a/collectors/sampleTimerMetric.go b/collectors/sampleTimerMetric.go new file mode 100644 index 0000000..aa6807e --- /dev/null +++ b/collectors/sampleTimerMetric.go @@ -0,0 +1,122 @@ +package collectors + +import ( + "encoding/json" + "sync" + "time" + + cclog "github.com/ClusterCockpit/cc-metric-collector/internal/ccLogger" + lp "github.com/ClusterCockpit/cc-metric-collector/internal/ccMetric" +) + +// These are the fields we read from the JSON configuration +type SampleTimerCollectorConfig struct { + Interval string `json:"interval"` +} + +// This contains all variables we need during execution and the variables +// defined by metricCollector (name, init, ...) +type SampleTimerCollector struct { + metricCollector + wg sync.WaitGroup // sync group for management + done chan bool // channel for management + meta map[string]string // default meta information + tags map[string]string // default tags + config SampleTimerCollectorConfig // the configuration structure + interval time.Duration // the interval parsed from configuration + ticker *time.Ticker // own timer + output chan lp.CCMetric // own internal output channel +} + +func (m *SampleTimerCollector) Init(name string, config json.RawMessage) error { + var err error = nil + // Always set the name early in Init() to use it in cclog.Component* functions + m.name = "SampleTimerCollector" + // This is for later use, also call it early + m.setup() + // Define meta information sent with each metric + // (Can also be dynamic or this is the basic set with extension through AddMeta()) + m.meta = map[string]string{"source": m.name, "group": "SAMPLE"} + // Define tags sent with each metric + // The 'type' tag is always needed, it defines the granulatity of the metric + // node -> whole system + // socket -> CPU socket (requires socket ID as 'type-id' tag) + // cpu -> single CPU hardware thread (requires cpu ID as 'type-id' tag) + m.tags = map[string]string{"type": "node"} + // Read in the JSON configuration + if len(config) > 0 { + err = json.Unmarshal(config, &m.config) + if err != nil { + cclog.ComponentError(m.name, "Error reading config:", err.Error()) + return err + } + } + // Parse the read interval duration + m.interval, err = time.ParseDuration(m.config.Interval) + if err != nil { + cclog.ComponentError(m.name, "Error parsing interval:", err.Error()) + return err + } + + // Storage for output channel + m.output = nil + // Mangement channel for the timer function. + m.done = make(chan bool) + // Create the own ticker + m.ticker = time.NewTicker(m.interval) + + // Start the timer loop with return functionality by sending 'true' to the done channel + m.wg.Add(1) + go func() { + select { + case <-m.done: + // Exit the timer loop + cclog.ComponentDebug(m.name, "Closing...") + m.wg.Done() + return + case timestamp := <-m.ticker.C: + // This is executed every timer tick but we have to wait until the first + // Read() to get the output channel + if m.output != nil { + m.ReadMetrics(timestamp) + } + } + }() + + // Set this flag only if everything is initialized properly, all required files exist, ... + m.init = true + return err +} + +// This function is called at each interval timer tick +func (m *SampleTimerCollector) ReadMetrics(timestamp time.Time) { + // Create a sample metric + + value := 1.0 + + // If you want to measure something for a specific amout of time, use interval + // start := readState() + // time.Sleep(interval) + // stop := readState() + // value = (stop - start) / interval.Seconds() + + y, err := lp.New("sample_metric", m.tags, m.meta, map[string]interface{}{"value": value}, timestamp) + if err == nil && m.output != nil { + // Send it to output channel if we have a valid channel + m.output <- y + } +} + +func (m *SampleTimerCollector) Read(interval time.Duration, output chan lp.CCMetric) { + // Capture output channel + m.output = output +} + +func (m *SampleTimerCollector) Close() { + // Send signal to the timer loop to stop it + m.done <- true + // Wait until the timer loop is done + m.wg.Wait() + // Unset flag + m.init = false +} diff --git a/receivers/README.md b/receivers/README.md index 24425f2..808dc74 100644 --- a/receivers/README.md +++ b/receivers/README.md @@ -7,17 +7,13 @@ This folder contains the ReceiveManager and receiver implementations for the cc- The configuration file for the receivers is a list of configurations. The `type` field in each specifies which receiver to initialize. ```json -[ - { - "type": "nats", - "address": "nats://my-url", - "port" : "4222", - "database": "testcluster" +{ + "myreceivername" : { + } -] +} ``` - ## Type `nats` ```json @@ -25,20 +21,20 @@ The configuration file for the receivers is a list of configurations. The `type` "type": "nats", "address": "", "port" : "", - "database": "" + "subject": "" } ``` The `nats` receiver subscribes to the topic `database` and listens on `address` and `port` for metrics in the InfluxDB line protocol. # Contributing own receivers -A receiver contains three functions and is derived from the type `Receiver` (in `metricReceiver.go`): -* `Init(config ReceiverConfig) error` +A receiver contains a few functions and is derived from the type `Receiver` (in `metricReceiver.go`): * `Start() error` * `Close()` * `Name() string` -* `SetSink(sink chan ccMetric.CCMetric)` +* `SetSink(sink chan lp.CCMetric)` +* `New(name string, config json.RawMessage)` The data structures should be set up in `Init()` like opening a file or server connection. The `Start()` function should either start a go routine or issue some other asynchronous mechanism for receiving metrics. The `Close()` function should tear down anything created in `Init()`. -Finally, the receiver needs to be registered in the `receiveManager.go`. There is a list of receivers called `AvailableReceivers` which is a map (`receiver_type_string` -> `pointer to Receiver interface`). Add a new entry with a descriptive name and the new receiver. +Finally, the receiver needs to be registered in the `receiveManager.go`. There is a list of receivers called `AvailableReceivers` which is a map (`receiver_type_string` -> `pointer to NewReceiver function`). Add a new entry with a descriptive name and the new receiver. diff --git a/receivers/metricReceiver.go b/receivers/metricReceiver.go index c712186..e133354 100644 --- a/receivers/metricReceiver.go +++ b/receivers/metricReceiver.go @@ -1,9 +1,6 @@ package receivers import ( - // "time" - "encoding/json" - lp "github.com/ClusterCockpit/cc-metric-collector/internal/ccMetric" ) @@ -20,13 +17,11 @@ type ReceiverConfig struct { } type receiver struct { - typename string - name string - sink chan lp.CCMetric + name string + sink chan lp.CCMetric } type Receiver interface { - Init(name string, config json.RawMessage) error Start() Close() Name() string diff --git a/receivers/natsReceiver.go b/receivers/natsReceiver.go index dc96971..1a5f47b 100644 --- a/receivers/natsReceiver.go +++ b/receivers/natsReceiver.go @@ -32,39 +32,6 @@ var DefaultTime = func() time.Time { return time.Unix(42, 0) } -func (r *NatsReceiver) Init(name string, config json.RawMessage) error { - r.typename = "NatsReceiver" - r.name = name - r.config.Addr = nats.DefaultURL - r.config.Port = "4222" - if len(config) > 0 { - err := json.Unmarshal(config, &r.config) - if err != nil { - cclog.ComponentError(r.name, "Error reading config:", err.Error()) - return err - } - } - if len(r.config.Addr) == 0 || - len(r.config.Port) == 0 || - len(r.config.Subject) == 0 { - return errors.New("not all configuration variables set required by NatsReceiver") - } - r.meta = map[string]string{"source": r.name} - uri := fmt.Sprintf("%s:%s", r.config.Addr, r.config.Port) - cclog.ComponentDebug(r.name, "INIT", uri, "Subject", r.config.Subject) - nc, err := nats.Connect(uri) - if err == nil { - r.nc = nc - } else { - r.nc = nil - return err - } - r.handler = influx.NewMetricHandler() - r.parser = influx.NewParser(r.handler) - r.parser.SetTimeFunc(DefaultTime) - return err -} - func (r *NatsReceiver) Start() { cclog.ComponentDebug(r.name, "START") r.nc.Subscribe(r.config.Subject, r._NatsReceive) @@ -91,3 +58,35 @@ func (r *NatsReceiver) Close() { r.nc.Close() } } + +func NewNatsReceiver(name string, config json.RawMessage) (Receiver, error) { + r := new(NatsReceiver) + r.name = fmt.Sprintf("NatsReceiver(%s)", name) + r.config.Addr = nats.DefaultURL + r.config.Port = "4222" + if len(config) > 0 { + err := json.Unmarshal(config, &r.config) + if err != nil { + cclog.ComponentError(r.name, "Error reading config:", err.Error()) + return nil, err + } + } + if len(r.config.Addr) == 0 || + len(r.config.Port) == 0 || + len(r.config.Subject) == 0 { + return nil, errors.New("not all configuration variables set required by NatsReceiver") + } + r.meta = map[string]string{"source": r.name} + uri := fmt.Sprintf("%s:%s", r.config.Addr, r.config.Port) + cclog.ComponentDebug(r.name, "NewNatsReceiver", uri, "Subject", r.config.Subject) + if nc, err := nats.Connect(uri); err == nil { + r.nc = nc + } else { + r.nc = nil + return nil, err + } + r.handler = influx.NewMetricHandler() + r.parser = influx.NewParser(r.handler) + r.parser.SetTimeFunc(DefaultTime) + return r, nil +} diff --git a/receivers/prometheusReceiver.go b/receivers/prometheusReceiver.go new file mode 100644 index 0000000..c22969d --- /dev/null +++ b/receivers/prometheusReceiver.go @@ -0,0 +1,122 @@ +package receivers + +import ( + "bufio" + "encoding/json" + "errors" + "fmt" + "log" + "net/http" + "strconv" + "strings" + "sync" + "time" + + cclog "github.com/ClusterCockpit/cc-metric-collector/internal/ccLogger" + lp "github.com/ClusterCockpit/cc-metric-collector/internal/ccMetric" +) + +type PrometheusReceiverConfig struct { + defaultReceiverConfig + Addr string `json:"address"` + Port string `json:"port"` + Path string `json:"path"` + Interval string `json:"interval"` + SSL bool `json:"ssl"` +} + +type PrometheusReceiver struct { + receiver + meta map[string]string + config PrometheusReceiverConfig + interval time.Duration + done chan bool + wg sync.WaitGroup + ticker *time.Ticker + uri string +} + +func (r *PrometheusReceiver) Start() { + cclog.ComponentDebug(r.name, "START", r.uri) + r.wg.Add(1) + + r.ticker = time.NewTicker(r.interval) + go func() { + for { + select { + case <-r.done: + r.wg.Done() + return + case t := <-r.ticker.C: + resp, err := http.Get(r.uri) + if err != nil { + log.Fatal(err) + } + defer resp.Body.Close() + + scanner := bufio.NewScanner(resp.Body) + for scanner.Scan() { + line := scanner.Text() + if strings.HasPrefix(line, "#") { + continue + } + lineSplit := strings.Fields(line) + // separate metric name from tags (labels in Prometheus) + tags := map[string]string{} + name := lineSplit[0] + if sindex := strings.Index(name, "{"); sindex >= 0 { + eindex := strings.Index(name, "}") + for _, kv := range strings.Split(name[sindex+1:eindex], ",") { + eq := strings.Index(kv, "=") + tags[kv[0:eq]] = strings.Trim(kv[eq+1:], "\"") + } + name = lineSplit[0][0:sindex] + } + value, err := strconv.ParseFloat(lineSplit[1], 64) + if err == nil { + y, err := lp.New(name, tags, r.meta, map[string]interface{}{"value": value}, t) + if err == nil { + r.sink <- y + } + } + } + } + } + }() +} + +func (r *PrometheusReceiver) Close() { + cclog.ComponentDebug(r.name, "CLOSE") + r.done <- true + r.wg.Wait() +} + +func NewPrometheusReceiver(name string, config json.RawMessage) (Receiver, error) { + r := new(PrometheusReceiver) + r.name = fmt.Sprintf("PrometheusReceiver(%s)", name) + if len(config) > 0 { + err := json.Unmarshal(config, &r.config) + if err != nil { + cclog.ComponentError(r.name, "Error reading config:", err.Error()) + return nil, err + } + } + if len(r.config.Addr) == 0 || + len(r.config.Port) == 0 || + len(r.config.Interval) == 0 { + return nil, errors.New("not all configuration variables set required by PrometheusReceiver (address and port)") + } + if len(r.config.Interval) > 0 { + t, err := time.ParseDuration(r.config.Interval) + if err == nil { + r.interval = t + } + } + r.meta = map[string]string{"source": r.name} + proto := "http" + if r.config.SSL { + proto = "https" + } + r.uri = fmt.Sprintf("%s://%s:%s/%s", proto, r.config.Addr, r.config.Port, r.config.Path) + return r, nil +} diff --git a/receivers/prometheusReceiver.md b/receivers/prometheusReceiver.md new file mode 100644 index 0000000..5fe6e46 --- /dev/null +++ b/receivers/prometheusReceiver.md @@ -0,0 +1,27 @@ +## `prometheus` receiver + +The `prometheus` receiver can be used to scrape the metrics of a single `prometheus` client. It does **not** use any official Golang library but making simple HTTP get requests and parse the response. + +### Configuration structure + +```json +{ + "": { + "type": "prometheus", + "address" : "testpromhost", + "port" : "12345", + "port" : "/prometheus", + "interval": "5s", + "ssl" : true, + } +} +``` + +- `type`: makes the receiver a `prometheus` receiver +- `address`: Hostname or IP of the Prometheus agent +- `port`: Port of Prometheus agent +- `path`: Path to the Prometheus endpoint +- `interval`: Scrape the Prometheus endpoint in this interval (default '5s') +- `ssl`: Use SSL or not + +The receiver requests data from `http(s)://
:/`. \ No newline at end of file diff --git a/receivers/receiveManager.go b/receivers/receiveManager.go index 7141170..1c13026 100644 --- a/receivers/receiveManager.go +++ b/receivers/receiveManager.go @@ -9,8 +9,8 @@ import ( lp "github.com/ClusterCockpit/cc-metric-collector/internal/ccMetric" ) -var AvailableReceivers = map[string]Receiver{ - "nats": &NatsReceiver{}, +var AvailableReceivers = map[string]func(name string, config json.RawMessage) (Receiver, error){ + "nats": NewNatsReceiver, } type receiveManager struct { @@ -30,11 +30,13 @@ type ReceiveManager interface { } func (rm *receiveManager) Init(wg *sync.WaitGroup, receiverConfigFile string) error { + // Initialize struct fields rm.inputs = make([]Receiver, 0) rm.output = nil rm.done = make(chan bool) rm.wg = wg rm.config = make([]json.RawMessage, 0) + configFile, err := os.Open(receiverConfigFile) if err != nil { cclog.ComponentError("ReceiveManager", err.Error()) @@ -51,6 +53,7 @@ func (rm *receiveManager) Init(wg *sync.WaitGroup, receiverConfigFile string) er for name, raw := range rawConfigs { rm.AddInput(name, raw) } + return nil } @@ -75,10 +78,9 @@ func (rm *receiveManager) AddInput(name string, rawConfig json.RawMessage) error cclog.ComponentError("ReceiveManager", "SKIP", config.Type, "unknown receiver:", err.Error()) return err } - r := AvailableReceivers[config.Type] - err = r.Init(name, rawConfig) + r, err := AvailableReceivers[config.Type](name, rawConfig) if err != nil { - cclog.ComponentError("ReceiveManager", "SKIP", r.Name(), "initialization failed:", err.Error()) + cclog.ComponentError("ReceiveManager", "SKIP", name, "initialization failed:", err.Error()) return err } rm.inputs = append(rm.inputs, r) diff --git a/receivers/sampleReceiver.go b/receivers/sampleReceiver.go new file mode 100644 index 0000000..e9edc90 --- /dev/null +++ b/receivers/sampleReceiver.go @@ -0,0 +1,77 @@ +package receivers + +import ( + "encoding/json" + "fmt" + + cclog "github.com/ClusterCockpit/cc-metric-collector/internal/ccLogger" +) + +type SampleReceiverConfig struct { + Type string `json:"type"` + Addr string `json:"address"` + Port string `json:"port"` +} + +type SampleReceiver struct { + receiver + config SampleReceiverConfig + + // Storage for static information + meta map[string]string + // Use in case of own go routine + // done chan bool + // wg sync.WaitGroup +} + +func (r *SampleReceiver) Start() { + cclog.ComponentDebug(r.name, "START") + + // Start server process like http.ListenAndServe() + + // or use own go routine but always make sure it exits + // as soon as it gets the signal of the r.done channel + // r.wg.Add(1) + // go func() { + // for { + // select { + // case <-r.done: + // r.wg.Done() + // return + // } + // } + // r.wg.Done() + // }() +} + +func (r *SampleReceiver) Close() { + cclog.ComponentDebug(r.name, "CLOSE") + + // Close server like http.Shutdown() + + // in case of own go routine, send the signal and wait + // r.done <- true + // r.wg.Wait() +} + +func NewSampleReceiver(name string, config json.RawMessage) (Receiver, error) { + r := new(SampleReceiver) + r.name = fmt.Sprintf("HttpReceiver(%s)", name) + + // Set static information + r.meta = map[string]string{"source": r.name} + + // Read the sample receiver specific JSON config + if len(config) > 0 { + err := json.Unmarshal(config, &r.config) + if err != nil { + cclog.ComponentError(r.name, "Error reading config:", err.Error()) + return nil, err + } + } + + // Check that all required fields in the configuration are set + // Use 'if len(r.config.Option) > 0' for strings + + return r, nil +} diff --git a/scripts/likwid_perfgroup_to_cc_config.py b/scripts/likwid_perfgroup_to_cc_config.py index 52959ed..f1c3451 100755 --- a/scripts/likwid_perfgroup_to_cc_config.py +++ b/scripts/likwid_perfgroup_to_cc_config.py @@ -39,7 +39,7 @@ def group_to_json(groupfile): llist = re.split("\s+", line) calc = llist[-1] metric = " ".join(llist[:-1]) - scope = "hwthread" + scope = "cpu" if "BOX" in calc: scope = "socket" if "PWR" in calc: diff --git a/sinks/README.md b/sinks/README.md index 8ff3743..df532c9 100644 --- a/sinks/README.md +++ b/sinks/README.md @@ -6,9 +6,11 @@ This folder contains the SinkManager and sink implementations for the cc-metric- - [`stdout`](./stdoutSink.md): Print all metrics to `stdout`, `stderr` or a file - [`http`](./httpSink.md): Send metrics to an HTTP server as POST requests - [`influxdb`](./influxSink.md): Send metrics to an [InfluxDB](https://www.influxdata.com/products/influxdb/) database +- [`influxasync`](./influxAsyncSink.md): Send metrics to an [InfluxDB](https://www.influxdata.com/products/influxdb/) database with non-blocking write API - [`nats`](./natsSink.md): Publish metrics to the [NATS](https://nats.io/) network overlay system - [`ganglia`](./gangliaSink.md): Publish metrics in the [Ganglia Monitoring System](http://ganglia.info/) using the `gmetric` CLI tool - [`libganglia`](./libgangliaSink.md): Publish metrics in the [Ganglia Monitoring System](http://ganglia.info/) directly using `libganglia.so` +- [`prometeus`](./prometheusSink.md): Publish metrics for the [Prometheus Monitoring System](https://prometheus.io/) # Configuration @@ -34,11 +36,12 @@ The configuration file for the sinks is a list of configurations. The `type` fie # Contributing own sinks -A sink contains four functions and is derived from the type `sink`: -* `Init(config json.RawMessage) error` +A sink contains five functions and is derived from the type `sink`: +* `Init(name string, config json.RawMessage) error` * `Write(point CCMetric) error` * `Flush() error` * `Close()` +* `New(name string, config json.RawMessage) (Sink, error)` (calls the `Init()` function) The data structures should be set up in `Init()` like opening a file or server connection. The `Write()` function writes/sends the data. For non-blocking sinks, the `Flush()` method tells the sink to drain its internal buffers. The `Close()` function should tear down anything created in `Init()`. @@ -65,8 +68,8 @@ type SampleSink struct { } // Initialize the sink by giving it a name and reading in the config JSON -func (s *SampleSink) Init(config json.RawMessage) error { - s.name = "SampleSink" // Always specify a name here +func (s *SampleSink) Init(name string, config json.RawMessage) error { + s.name = fmt.Sprintf("SampleSink(%s)", name) // Always specify a name here // Read in the config JSON if len(config) > 0 { err := json.Unmarshal(config, &s.config) @@ -91,4 +94,13 @@ func (s *SampleSink) Flush() error { // Close sink: close network connection, close files, close libraries, ... func (s *SampleSink) Close() {} + + +// New function to create a new instance of the sink +func NewSampleSink(name string, config json.RawMessage) (Sink, error) { + s := new(SampleSink) + err := s.Init(name, config) + return s, err +} + ``` \ No newline at end of file diff --git a/sinks/gangliaCommon.go b/sinks/gangliaCommon.go index b939f16..b2a1b2c 100644 --- a/sinks/gangliaCommon.go +++ b/sinks/gangliaCommon.go @@ -1,6 +1,7 @@ package sinks import ( + "fmt" "strings" lp "github.com/ClusterCockpit/cc-metric-collector/internal/ccMetric" @@ -23,11 +24,8 @@ func GangliaMetricName(point lp.CCMetric) string { return name } -func GangliaMetricRename(point lp.CCMetric) string { - name := point.Name() - if name == "mem_total" || name == "swap_total" { - return name - } else if name == "net_bytes_in" { +func GangliaMetricRename(name string) string { + if name == "net_bytes_in" { return "bytes_in" } else if name == "net_bytes_out" { return "bytes_out" @@ -48,3 +46,213 @@ func GangliaSlopeType(point lp.CCMetric) uint { } return 3 } + +const DEFAULT_GANGLIA_METRIC_TMAX = 300 +const DEFAULT_GANGLIA_METRIC_SLOPE = "both" + +type GangliaMetric struct { + Name string + Type string + Slope string + Tmax int + Unit string +} + +type GangliaMetricGroup struct { + Name string + Metrics []GangliaMetric +} + +var CommonGangliaMetrics = []GangliaMetricGroup{ + { + Name: "memory", + Metrics: []GangliaMetric{ + {"mem_total", "float", "zero", 1200, "KB"}, + {"swap_total", "float", "zero", 1200, "KB"}, + {"mem_free", "float", "both", 180, "KB"}, + {"mem_shared", "float", "both", 180, "KB"}, + {"mem_buffers", "float", "both", 180, "KB"}, + {"mem_cached", "float", "both", 180, "KB"}, + {"swap_free", "float", "both", 180, "KB"}, + {"mem_sreclaimable", "float", "both", 180, "KB"}, + {"mem_slab", "float", "both", 180, "KB"}, + }, + }, + { + Name: "cpu", + Metrics: []GangliaMetric{ + {"cpu_num", "uint32", "zero", 1200, "CPUs"}, + {"cpu_speed", "uint32", "zero", 1200, "MHz"}, + {"cpu_user", "float", "both", 90, "%"}, + {"cpu_nice", "float", "both", 90, "%"}, + {"cpu_system", "float", "both", 90, "%"}, + {"cpu_idle", "float", "both", 3800, "%"}, + {"cpu_aidle", "float", "both", 90, "%"}, + {"cpu_wio", "float", "both", 90, "%"}, + {"cpu_intr", "float", "both", 90, "%"}, + {"cpu_sintr", "float", "both", 90, "%"}, + {"cpu_steal", "float", "both", 90, "%"}, + {"cpu_guest", "float", "both", 90, "%"}, + {"cpu_gnice", "float", "both", 90, "%"}, + }, + }, + { + Name: "load", + Metrics: []GangliaMetric{ + {"load_one", "float", "both", 70, ""}, + {"load_five", "float", "both", 325, ""}, + {"load_fifteen", "float", "both", 950, ""}, + }, + }, + { + Name: "disk", + Metrics: []GangliaMetric{ + {"disk_total", "double", "both", 1200, "GB"}, + {"disk_free", "double", "both", 180, "GB"}, + {"part_max_used", "float", "both", 180, "%"}, + }, + }, + { + Name: "network", + Metrics: []GangliaMetric{ + {"bytes_out", "float", "both", 300, "bytes/sec"}, + {"bytes_in", "float", "both", 300, "bytes/sec"}, + {"pkts_in", "float", "both", 300, "packets/sec"}, + {"pkts_out", "float", "both", 300, "packets/sec"}, + }, + }, + { + Name: "process", + Metrics: []GangliaMetric{ + {"proc_run", "uint32", "both", 950, ""}, + {"proc_total", "uint32", "both", 950, ""}, + }, + }, + { + Name: "system", + Metrics: []GangliaMetric{ + {"boottime", "uint32", "zero", 1200, "s"}, + {"sys_clock", "uint32", "zero", 1200, "s"}, + {"machine_type", "string", "zero", 1200, ""}, + {"os_name", "string", "zero", 1200, ""}, + {"os_release", "string", "zero", 1200, ""}, + {"mtu", "uint32", "both", 1200, ""}, + }, + }, +} + +type GangliaMetricConfig struct { + Type string + Slope string + Tmax int + Unit string + Group string + Value string +} + +func GetCommonGangliaConfig(point lp.CCMetric) GangliaMetricConfig { + mname := GangliaMetricRename(point.Name()) + for _, group := range CommonGangliaMetrics { + for _, metric := range group.Metrics { + if metric.Name == mname { + valueStr := "" + value, ok := point.GetField("value") + if ok { + switch real := value.(type) { + case float64: + valueStr = fmt.Sprintf("%f", real) + case float32: + valueStr = fmt.Sprintf("%f", real) + case int64: + valueStr = fmt.Sprintf("%d", real) + case int32: + valueStr = fmt.Sprintf("%d", real) + case int: + valueStr = fmt.Sprintf("%d", real) + case uint64: + valueStr = fmt.Sprintf("%d", real) + case uint32: + valueStr = fmt.Sprintf("%d", real) + case uint: + valueStr = fmt.Sprintf("%d", real) + case string: + valueStr = real + default: + } + } + return GangliaMetricConfig{ + Group: group.Name, + Type: metric.Type, + Slope: metric.Slope, + Tmax: metric.Tmax, + Unit: metric.Unit, + Value: valueStr, + } + } + } + } + return GangliaMetricConfig{ + Group: "", + Type: "", + Slope: "", + Tmax: 0, + Unit: "", + Value: "", + } +} + +func GetGangliaConfig(point lp.CCMetric) GangliaMetricConfig { + group := "" + if g, ok := point.GetMeta("group"); ok { + group = g + } + unit := "" + if u, ok := point.GetMeta("unit"); ok { + unit = u + } + valueType := "double" + valueStr := "" + value, ok := point.GetField("value") + if ok { + switch real := value.(type) { + case float64: + valueStr = fmt.Sprintf("%f", real) + valueType = "double" + case float32: + valueStr = fmt.Sprintf("%f", real) + valueType = "float" + case int64: + valueStr = fmt.Sprintf("%d", real) + valueType = "int32" + case int32: + valueStr = fmt.Sprintf("%d", real) + valueType = "int32" + case int: + valueStr = fmt.Sprintf("%d", real) + valueType = "int32" + case uint64: + valueStr = fmt.Sprintf("%d", real) + valueType = "uint32" + case uint32: + valueStr = fmt.Sprintf("%d", real) + valueType = "uint32" + case uint: + valueStr = fmt.Sprintf("%d", real) + valueType = "uint32" + case string: + valueStr = real + valueType = "string" + default: + valueType = "invalid" + } + } + + return GangliaMetricConfig{ + Group: group, + Type: valueType, + Slope: DEFAULT_GANGLIA_METRIC_SLOPE, + Tmax: DEFAULT_GANGLIA_METRIC_TMAX, + Unit: unit, + Value: valueStr, + } +} diff --git a/sinks/gangliaSink.go b/sinks/gangliaSink.go index fa95f43..22096af 100644 --- a/sinks/gangliaSink.go +++ b/sinks/gangliaSink.go @@ -24,6 +24,7 @@ type GangliaSinkConfig struct { AddTagsAsDesc bool `json:"add_tags_as_desc,omitempty"` ClusterName string `json:"cluster_name,omitempty"` AddTypeToName bool `json:"add_type_to_name,omitempty"` + AddUnits bool `json:"add_units,omitempty"` } type GangliaSink struct { @@ -33,16 +34,73 @@ type GangliaSink struct { config GangliaSinkConfig } -func (s *GangliaSink) Init(config json.RawMessage) error { +func (s *GangliaSink) Write(point lp.CCMetric) error { var err error = nil - s.name = "GangliaSink" + //var tagsstr []string + var argstr []string + + // Get metric name + metricname := GangliaMetricRename(point.Name()) + + // Get metric config (type, value, ... in suitable format) + conf := GetCommonGangliaConfig(point) + if len(conf.Type) == 0 { + conf = GetGangliaConfig(point) + } + if len(conf.Type) == 0 { + return fmt.Errorf("metric %s has no 'value' field", metricname) + } + + if s.config.AddGangliaGroup { + argstr = append(argstr, fmt.Sprintf("--group=%s", conf.Group)) + } + if s.config.AddUnits && len(conf.Unit) > 0 { + argstr = append(argstr, fmt.Sprintf("--units=%s", conf.Unit)) + } + + if len(s.config.ClusterName) > 0 { + argstr = append(argstr, fmt.Sprintf("--cluster=%s", s.config.ClusterName)) + } + // if s.config.AddTagsAsDesc && len(tagsstr) > 0 { + // argstr = append(argstr, fmt.Sprintf("--desc=%q", strings.Join(tagsstr, ","))) + // } + if len(s.gmetric_config) > 0 { + argstr = append(argstr, fmt.Sprintf("--conf=%s", s.gmetric_config)) + } + if s.config.AddTypeToName { + argstr = append(argstr, fmt.Sprintf("--name=%s", GangliaMetricName(point))) + } else { + argstr = append(argstr, fmt.Sprintf("--name=%s", metricname)) + } + argstr = append(argstr, fmt.Sprintf("--slope=%s", conf.Slope)) + argstr = append(argstr, fmt.Sprintf("--value=%s", conf.Value)) + argstr = append(argstr, fmt.Sprintf("--type=%s", conf.Type)) + argstr = append(argstr, fmt.Sprintf("--tmax=%d", conf.Tmax)) + + cclog.ComponentDebug(s.name, s.gmetric_path, strings.Join(argstr, " ")) + command := exec.Command(s.gmetric_path, argstr...) + command.Wait() + _, err = command.Output() + return err +} + +func (s *GangliaSink) Flush() error { + return nil +} + +func (s *GangliaSink) Close() { +} + +func NewGangliaSink(name string, config json.RawMessage) (Sink, error) { + s := new(GangliaSink) + s.name = fmt.Sprintf("GangliaSink(%s)", name) s.config.AddTagsAsDesc = false s.config.AddGangliaGroup = false if len(config) > 0 { err := json.Unmarshal(config, &s.config) if err != nil { cclog.ComponentError(s.name, "Error reading config for", s.name, ":", err.Error()) - return err + return nil, err } } s.gmetric_path = "" @@ -60,98 +118,10 @@ func (s *GangliaSink) Init(config json.RawMessage) error { } } if len(s.gmetric_path) == 0 { - err = errors.New("cannot find executable 'gmetric'") + return nil, errors.New("cannot find executable 'gmetric'") } if len(s.config.GmetricConfig) > 0 { s.gmetric_config = s.config.GmetricConfig } - return err -} - -func (s *GangliaSink) Write(point lp.CCMetric) error { - var err error = nil - var tagsstr []string - var argstr []string - if s.config.AddGangliaGroup { - if point.HasTag("group") { - g, _ := point.GetTag("group") - argstr = append(argstr, fmt.Sprintf("--group=%s", g)) - } else if point.HasMeta("group") { - g, _ := point.GetMeta("group") - argstr = append(argstr, fmt.Sprintf("--group=%s", g)) - } - } - - for key, value := range point.Tags() { - switch key { - case "unit": - argstr = append(argstr, fmt.Sprintf("--units=%s", value)) - default: - tagsstr = append(tagsstr, fmt.Sprintf("%s=%s", key, value)) - } - } - if s.config.MetaAsTags { - for key, value := range point.Meta() { - switch key { - case "unit": - argstr = append(argstr, fmt.Sprintf("--units=%s", value)) - default: - tagsstr = append(tagsstr, fmt.Sprintf("%s=%s", key, value)) - } - } - } - if len(s.config.ClusterName) > 0 { - argstr = append(argstr, fmt.Sprintf("--cluster=%s", s.config.ClusterName)) - } - if s.config.AddTagsAsDesc && len(tagsstr) > 0 { - argstr = append(argstr, fmt.Sprintf("--desc=%q", strings.Join(tagsstr, ","))) - } - if len(s.gmetric_config) > 0 { - argstr = append(argstr, fmt.Sprintf("--conf=%s", s.gmetric_config)) - } - name := GangliaMetricRename(point) - if s.config.AddTypeToName { - argstr = append(argstr, fmt.Sprintf("--name=%s", GangliaMetricName(point))) - } else { - argstr = append(argstr, fmt.Sprintf("--name=%s", name)) - } - slope := GangliaSlopeType(point) - slopeStr := "both" - if slope == 0 { - slopeStr = "zero" - } - argstr = append(argstr, fmt.Sprintf("--slope=%s", slopeStr)) - - for k, v := range point.Fields() { - if k == "value" { - switch value := v.(type) { - case float64: - argstr = append(argstr, - fmt.Sprintf("--value=%v", value), "--type=double") - case float32: - argstr = append(argstr, - fmt.Sprintf("--value=%v", value), "--type=float") - case int: - argstr = append(argstr, - fmt.Sprintf("--value=%d", value), "--type=int32") - case int64: - argstr = append(argstr, - fmt.Sprintf("--value=%d", value), "--type=int32") - case string: - argstr = append(argstr, - fmt.Sprintf("--value=%q", value), "--type=string") - } - } - } - command := exec.Command(s.gmetric_path, argstr...) - command.Wait() - _, err = command.Output() - return err -} - -func (s *GangliaSink) Flush() error { - return nil -} - -func (s *GangliaSink) Close() { + return s, nil } diff --git a/sinks/httpSink.go b/sinks/httpSink.go index ce46bab..c2dd2ea 100644 --- a/sinks/httpSink.go +++ b/sinks/httpSink.go @@ -38,57 +38,6 @@ type HttpSink struct { flushDelay time.Duration } -func (s *HttpSink) Init(config json.RawMessage) error { - // Set default values - s.name = "HttpSink" - s.config.MaxIdleConns = 10 - s.config.IdleConnTimeout = "5s" - s.config.Timeout = "5s" - s.config.FlushDelay = "1s" - - // Read config - if len(config) > 0 { - err := json.Unmarshal(config, &s.config) - if err != nil { - return err - } - } - if len(s.config.URL) == 0 { - return errors.New("`url` config option is required for HTTP sink") - } - if s.config.MaxIdleConns > 0 { - s.maxIdleConns = s.config.MaxIdleConns - } - if len(s.config.IdleConnTimeout) > 0 { - t, err := time.ParseDuration(s.config.IdleConnTimeout) - if err == nil { - s.idleConnTimeout = t - } - } - if len(s.config.Timeout) > 0 { - t, err := time.ParseDuration(s.config.Timeout) - if err == nil { - s.timeout = t - } - } - if len(s.config.FlushDelay) > 0 { - t, err := time.ParseDuration(s.config.FlushDelay) - if err == nil { - s.flushDelay = t - } - } - tr := &http.Transport{ - MaxIdleConns: s.maxIdleConns, - IdleConnTimeout: s.idleConnTimeout, - } - s.client = &http.Client{Transport: tr, Timeout: s.timeout} - s.buffer = &bytes.Buffer{} - s.encoder = influx.NewEncoder(s.buffer) - s.encoder.SetPrecision(time.Second) - - return nil -} - func (s *HttpSink) Write(m lp.CCMetric) error { if s.buffer.Len() == 0 && s.flushDelay != 0 { // This is the first write since the last flush, start the flushTimer! @@ -169,3 +118,54 @@ func (s *HttpSink) Close() { } s.client.CloseIdleConnections() } + +func NewHttpSink(name string, config json.RawMessage) (Sink, error) { + s := new(HttpSink) + // Set default values + s.name = fmt.Sprintf("HttpSink(%s)", name) + s.config.MaxIdleConns = 10 + s.config.IdleConnTimeout = "5s" + s.config.Timeout = "5s" + s.config.FlushDelay = "1s" + + // Read config + if len(config) > 0 { + err := json.Unmarshal(config, &s.config) + if err != nil { + return nil, err + } + } + if len(s.config.URL) == 0 { + return nil, errors.New("`url` config option is required for HTTP sink") + } + if s.config.MaxIdleConns > 0 { + s.maxIdleConns = s.config.MaxIdleConns + } + if len(s.config.IdleConnTimeout) > 0 { + t, err := time.ParseDuration(s.config.IdleConnTimeout) + if err == nil { + s.idleConnTimeout = t + } + } + if len(s.config.Timeout) > 0 { + t, err := time.ParseDuration(s.config.Timeout) + if err == nil { + s.timeout = t + } + } + if len(s.config.FlushDelay) > 0 { + t, err := time.ParseDuration(s.config.FlushDelay) + if err == nil { + s.flushDelay = t + } + } + tr := &http.Transport{ + MaxIdleConns: s.maxIdleConns, + IdleConnTimeout: s.idleConnTimeout, + } + s.client = &http.Client{Transport: tr, Timeout: s.timeout} + s.buffer = &bytes.Buffer{} + s.encoder = influx.NewEncoder(s.buffer) + s.encoder.SetPrecision(time.Second) + return s, nil +} diff --git a/sinks/influxAsyncSink.go b/sinks/influxAsyncSink.go index 20aa60c..7b38873 100644 --- a/sinks/influxAsyncSink.go +++ b/sinks/influxAsyncSink.go @@ -1,6 +1,7 @@ package sinks import ( + "context" "crypto/tls" "encoding/json" "errors" @@ -30,11 +31,10 @@ type InfluxAsyncSinkConfig struct { type InfluxAsyncSink struct { sink - client influxdb2.Client - writeApi influxdb2Api.WriteAPI - retPolicy string - errors <-chan error - config InfluxAsyncSinkConfig + client influxdb2.Client + writeApi influxdb2Api.WriteAPI + errors <-chan error + config InfluxAsyncSinkConfig } func (s *InfluxAsyncSink) connect() error { @@ -65,42 +65,16 @@ func (s *InfluxAsyncSink) connect() error { ) s.client = influxdb2.NewClientWithOptions(uri, auth, clientOptions) s.writeApi = s.client.WriteAPI(s.config.Organization, s.config.Database) + ok, err := s.client.Ping(context.Background()) + if err != nil { + return err + } + if !ok { + return fmt.Errorf("connection to %s not healthy", uri) + } return nil } -func (s *InfluxAsyncSink) Init(config json.RawMessage) error { - s.name = "InfluxSink" - - // Set default for maximum number of points sent to server in single request. - s.config.BatchSize = 100 - - if len(config) > 0 { - err := json.Unmarshal(config, &s.config) - if err != nil { - return err - } - } - if len(s.config.Host) == 0 || - len(s.config.Port) == 0 || - len(s.config.Database) == 0 || - len(s.config.Organization) == 0 || - len(s.config.Password) == 0 { - return errors.New("not all configuration variables set required by InfluxAsyncSink") - } - - // Connect to InfluxDB server - err := s.connect() - - // Start background: Read from error channel - s.errors = s.writeApi.Errors() - go func() { - for err := range s.errors { - cclog.ComponentError(s.name, err.Error()) - } - }() - return err -} - func (s *InfluxAsyncSink) Write(m lp.CCMetric) error { s.writeApi.WritePoint( m.ToPoint(s.config.MetaAsTags), @@ -118,3 +92,40 @@ func (s *InfluxAsyncSink) Close() { s.writeApi.Flush() s.client.Close() } + +func NewInfluxAsyncSink(name string, config json.RawMessage) (Sink, error) { + s := new(InfluxAsyncSink) + s.name = fmt.Sprintf("InfluxSink(%s)", name) + + // Set default for maximum number of points sent to server in single request. + s.config.BatchSize = 100 + + if len(config) > 0 { + err := json.Unmarshal(config, &s.config) + if err != nil { + return nil, err + } + } + if len(s.config.Host) == 0 || + len(s.config.Port) == 0 || + len(s.config.Database) == 0 || + len(s.config.Organization) == 0 || + len(s.config.Password) == 0 { + return nil, errors.New("not all configuration variables set required by InfluxAsyncSink") + } + + // Connect to InfluxDB server + if err := s.connect(); err != nil { + return nil, fmt.Errorf("unable to connect: %v", err) + } + + // Start background: Read from error channel + s.errors = s.writeApi.Errors() + go func() { + for err := range s.errors { + cclog.ComponentError(s.name, err.Error()) + } + }() + + return s, nil +} diff --git a/sinks/influxSink.go b/sinks/influxSink.go index 99304c0..11859b2 100644 --- a/sinks/influxSink.go +++ b/sinks/influxSink.go @@ -54,29 +54,16 @@ func (s *InfluxSink) connect() error { ) s.client = influxdb2.NewClientWithOptions(uri, auth, clientOptions) s.writeApi = s.client.WriteAPIBlocking(s.config.Organization, s.config.Database) + ok, err := s.client.Ping(context.Background()) + if err != nil { + return err + } + if !ok { + return fmt.Errorf("connection to %s not healthy", uri) + } return nil } -func (s *InfluxSink) Init(config json.RawMessage) error { - s.name = "InfluxSink" - if len(config) > 0 { - err := json.Unmarshal(config, &s.config) - if err != nil { - return err - } - } - if len(s.config.Host) == 0 || - len(s.config.Port) == 0 || - len(s.config.Database) == 0 || - len(s.config.Organization) == 0 || - len(s.config.Password) == 0 { - return errors.New("not all configuration variables set required by InfluxSink") - } - - // Connect to InfluxDB server - return s.connect() -} - func (s *InfluxSink) Write(m lp.CCMetric) error { err := s.writeApi.WritePoint( @@ -94,3 +81,27 @@ func (s *InfluxSink) Close() { cclog.ComponentDebug(s.name, "Closing InfluxDB connection") s.client.Close() } + +func NewInfluxSink(name string, config json.RawMessage) (Sink, error) { + s := new(InfluxSink) + s.name = fmt.Sprintf("InfluxSink(%s)", name) + if len(config) > 0 { + err := json.Unmarshal(config, &s.config) + if err != nil { + return nil, err + } + } + if len(s.config.Host) == 0 || + len(s.config.Port) == 0 || + len(s.config.Database) == 0 || + len(s.config.Organization) == 0 || + len(s.config.Password) == 0 { + return nil, errors.New("not all configuration variables set required by InfluxSink") + } + + // Connect to InfluxDB server + if err := s.connect(); err != nil { + return nil, fmt.Errorf("unable to connect: %v", err) + } + return s, nil +} diff --git a/sinks/libgangliaSink.go b/sinks/libgangliaSink.go index 051bd03..1fc7863 100644 --- a/sinks/libgangliaSink.go +++ b/sinks/libgangliaSink.go @@ -82,21 +82,21 @@ const ( GMOND_CONFIG_FILE = `/etc/ganglia/gmond.conf` ) -type LibgangliaSinkSpecialMetric struct { - MetricName string `json:"metric_name,omitempty"` - NewName string `json:"new_name,omitempty"` - Slope string `json:"slope,omitempty"` -} +// type LibgangliaSinkSpecialMetric struct { +// MetricName string `json:"metric_name,omitempty"` +// NewName string `json:"new_name,omitempty"` +// Slope string `json:"slope,omitempty"` +// } type LibgangliaSinkConfig struct { defaultSinkConfig - GangliaLib string `json:"libganglia_path,omitempty"` - GmondConfig string `json:"gmond_config,omitempty"` - AddGangliaGroup bool `json:"add_ganglia_group,omitempty"` - AddTypeToName bool `json:"add_type_to_name,omitempty"` - AddUnits bool `json:"add_units,omitempty"` - ClusterName string `json:"cluster_name,omitempty"` - SpecialMetrics map[string]LibgangliaSinkSpecialMetric `json:"rename_metrics,omitempty"` // Map to rename metric name from key to value + GangliaLib string `json:"libganglia_path,omitempty"` + GmondConfig string `json:"gmond_config,omitempty"` + AddGangliaGroup bool `json:"add_ganglia_group,omitempty"` + AddTypeToName bool `json:"add_type_to_name,omitempty"` + AddUnits bool `json:"add_units,omitempty"` + ClusterName string `json:"cluster_name,omitempty"` + //SpecialMetrics map[string]LibgangliaSinkSpecialMetric `json:"rename_metrics,omitempty"` // Map to rename metric name from key to value //AddTagsAsDesc bool `json:"add_tags_as_desc,omitempty"` } @@ -109,65 +109,6 @@ type LibgangliaSink struct { cstrCache map[string]*C.char } -func (s *LibgangliaSink) Init(config json.RawMessage) error { - var err error = nil - s.name = "LibgangliaSink" - //s.config.AddTagsAsDesc = false - s.config.AddGangliaGroup = false - s.config.AddTypeToName = false - s.config.AddUnits = true - s.config.GmondConfig = string(GMOND_CONFIG_FILE) - s.config.GangliaLib = string(GANGLIA_LIB_NAME) - if len(config) > 0 { - err = json.Unmarshal(config, &s.config) - if err != nil { - cclog.ComponentError(s.name, "Error reading config:", err.Error()) - return err - } - } - lib := dl.New(s.config.GangliaLib, GANGLIA_LIB_DL_FLAGS) - if lib == nil { - return fmt.Errorf("error instantiating DynamicLibrary for %s", s.config.GangliaLib) - } - err = lib.Open() - if err != nil { - return fmt.Errorf("error opening %s: %v", s.config.GangliaLib, err) - } - - // Set up cache for the C strings - s.cstrCache = make(map[string]*C.char) - // s.cstrCache["globals"] = C.CString("globals") - - // s.cstrCache["override_hostname"] = C.CString("override_hostname") - // s.cstrCache["override_ip"] = C.CString("override_ip") - - // Add some constant strings - s.cstrCache["GROUP"] = C.CString("GROUP") - s.cstrCache["CLUSTER"] = C.CString("CLUSTER") - s.cstrCache[""] = C.CString("") - - // Add cluster name for lookup in Write() - if len(s.config.ClusterName) > 0 { - s.cstrCache[s.config.ClusterName] = C.CString(s.config.ClusterName) - } - // Add supported types for later lookup in Write() - s.cstrCache["double"] = C.CString("double") - s.cstrCache["int32"] = C.CString("int32") - s.cstrCache["string"] = C.CString("string") - - // Create Ganglia pool - s.global_context = C.Ganglia_pool_create(nil) - // Load Ganglia configuration - s.cstrCache[s.config.GmondConfig] = C.CString(s.config.GmondConfig) - s.gmond_config = C.Ganglia_gmond_config_create(s.cstrCache[s.config.GmondConfig], 0) - //globals := C.cfg_getsec(gmond_config, s.cstrCache["globals"]) - //override_hostname := C.cfg_getstr(globals, s.cstrCache["override_hostname"]) - //override_ip := C.cfg_getstr(globals, s.cstrCache["override_ip"]) - - s.send_channels = C.Ganglia_udp_send_channels_create(s.global_context, s.gmond_config) - return nil -} - func (s *LibgangliaSink) Write(point lp.CCMetric) error { var err error = nil var c_name *C.char @@ -184,72 +125,48 @@ func (s *LibgangliaSink) Write(point lp.CCMetric) error { } // Get metric name - metricname := GangliaMetricRename(point) - if s.config.AddTypeToName { - c_name = lookup(GangliaMetricName(point)) - } else { - c_name = lookup(metricname) - } + metricname := GangliaMetricRename(point.Name()) - // Get the value C string and lookup the type string in the cache - value, ok := point.GetField("value") - if !ok { + conf := GetCommonGangliaConfig(point) + if len(conf.Type) == 0 { + conf = GetGangliaConfig(point) + } + if len(conf.Type) == 0 { return fmt.Errorf("metric %s has no 'value' field", metricname) } - switch real := value.(type) { - case float64: - c_value = C.CString(fmt.Sprintf("%f", real)) - c_type = lookup("double") - case float32: - c_value = C.CString(fmt.Sprintf("%f", real)) - c_type = lookup("float") - case int64: - c_value = C.CString(fmt.Sprintf("%d", real)) - c_type = lookup("int32") - case int32: - c_value = C.CString(fmt.Sprintf("%d", real)) - c_type = lookup("int32") - case int: - c_value = C.CString(fmt.Sprintf("%d", real)) - c_type = lookup("int32") - case string: - c_value = C.CString(real) - c_type = lookup("string") - default: - return fmt.Errorf("metric %s has invalid 'value' type for %s", point.Name(), s.name) + + if s.config.AddTypeToName { + metricname = GangliaMetricName(point) } + c_value = C.CString(conf.Value) + c_type = lookup(conf.Type) + c_name = lookup(metricname) + // Add unit + unit := "" if s.config.AddUnits { - if tagunit, tagok := point.GetTag("unit"); tagok { - c_unit = lookup(tagunit) - } else if metaunit, metaok := point.GetMeta("unit"); metaok { - c_unit = lookup(metaunit) - } else { - c_unit = lookup("") - } - } else { - c_unit = lookup("") + unit = conf.Unit } + c_unit = lookup(unit) // Determine the slope of the metric. Ganglia's own collector mostly use // 'both' but the mem and swap total uses 'zero'. - slope := GangliaSlopeType(point) slope_type := C.GANGLIA_SLOPE_BOTH - switch slope { - case 0: + switch conf.Slope { + case "zero": slope_type = C.GANGLIA_SLOPE_ZERO + case "both": + slope_type = C.GANGLIA_SLOPE_BOTH } // Create a new Ganglia metric gmetric := C.Ganglia_metric_create(s.global_context) // Set name, value, type and unit in the Ganglia metric - // Since we don't have this information from the collectors, - // we assume that the metric value can go up and down (slope), - // and there is no maximum for 'dmax' and 'tmax'. - // Ganglia's collectors set 'tmax' but not 'dmax' + // The default slope_type is both directions, so up and down. Some metrics want 'zero' slope, probably constant. + // The 'tmax' value is by default 300. rval := C.int(0) - rval = C.Ganglia_metric_set(gmetric, c_name, c_value, c_type, c_unit, C.uint(slope_type), 0, 0) + rval = C.Ganglia_metric_set(gmetric, c_name, c_value, c_type, c_unit, C.uint(slope_type), C.uint(conf.Tmax), 0) switch rval { case 1: C.free(unsafe.Pointer(c_value)) @@ -259,10 +176,10 @@ func (s *LibgangliaSink) Write(point lp.CCMetric) error { return errors.New("one of your parameters has an invalid character '\"'") case 3: C.free(unsafe.Pointer(c_value)) - return fmt.Errorf("the type parameter \"%s\" is not a valid type", C.GoString(c_type)) + return fmt.Errorf("the type parameter \"%s\" is not a valid type", conf.Type) case 4: C.free(unsafe.Pointer(c_value)) - return fmt.Errorf("the value parameter \"%s\" does not represent a number", C.GoString(c_value)) + return fmt.Errorf("the value parameter \"%s\" does not represent a number", conf.Value) default: } @@ -271,8 +188,8 @@ func (s *LibgangliaSink) Write(point lp.CCMetric) error { C.Ganglia_metadata_add(gmetric, lookup("CLUSTER"), lookup(s.config.ClusterName)) } // Set the group metadata in the Ganglia metric if configured - if group, ok := point.GetMeta("group"); ok && s.config.AddGangliaGroup { - c_group := lookup(group) + if s.config.AddGangliaGroup { + c_group := lookup(conf.Group) C.Ganglia_metadata_add(gmetric, lookup("GROUP"), c_group) } @@ -307,3 +224,63 @@ func (s *LibgangliaSink) Close() { C.free(unsafe.Pointer(cstr)) } } + +func NewLibgangliaSink(name string, config json.RawMessage) (Sink, error) { + s := new(LibgangliaSink) + var err error = nil + s.name = fmt.Sprintf("LibgangliaSink(%s)", name) + //s.config.AddTagsAsDesc = false + s.config.AddGangliaGroup = false + s.config.AddTypeToName = false + s.config.AddUnits = true + s.config.GmondConfig = string(GMOND_CONFIG_FILE) + s.config.GangliaLib = string(GANGLIA_LIB_NAME) + if len(config) > 0 { + err = json.Unmarshal(config, &s.config) + if err != nil { + cclog.ComponentError(s.name, "Error reading config:", err.Error()) + return nil, err + } + } + lib := dl.New(s.config.GangliaLib, GANGLIA_LIB_DL_FLAGS) + if lib == nil { + return nil, fmt.Errorf("error instantiating DynamicLibrary for %s", s.config.GangliaLib) + } + err = lib.Open() + if err != nil { + return nil, fmt.Errorf("error opening %s: %v", s.config.GangliaLib, err) + } + + // Set up cache for the C strings + s.cstrCache = make(map[string]*C.char) + // s.cstrCache["globals"] = C.CString("globals") + + // s.cstrCache["override_hostname"] = C.CString("override_hostname") + // s.cstrCache["override_ip"] = C.CString("override_ip") + + // Add some constant strings + s.cstrCache["GROUP"] = C.CString("GROUP") + s.cstrCache["CLUSTER"] = C.CString("CLUSTER") + s.cstrCache[""] = C.CString("") + + // Add cluster name for lookup in Write() + if len(s.config.ClusterName) > 0 { + s.cstrCache[s.config.ClusterName] = C.CString(s.config.ClusterName) + } + // Add supported types for later lookup in Write() + s.cstrCache["double"] = C.CString("double") + s.cstrCache["int32"] = C.CString("int32") + s.cstrCache["string"] = C.CString("string") + + // Create Ganglia pool + s.global_context = C.Ganglia_pool_create(nil) + // Load Ganglia configuration + s.cstrCache[s.config.GmondConfig] = C.CString(s.config.GmondConfig) + s.gmond_config = C.Ganglia_gmond_config_create(s.cstrCache[s.config.GmondConfig], 0) + //globals := C.cfg_getsec(gmond_config, s.cstrCache["globals"]) + //override_hostname := C.cfg_getstr(globals, s.cstrCache["override_hostname"]) + //override_ip := C.cfg_getstr(globals, s.cstrCache["override_ip"]) + + s.send_channels = C.Ganglia_udp_send_channels_create(s.global_context, s.gmond_config) + return s, nil +} diff --git a/sinks/metricSink.go b/sinks/metricSink.go index d76f5f2..8fe02d7 100644 --- a/sinks/metricSink.go +++ b/sinks/metricSink.go @@ -1,8 +1,6 @@ package sinks import ( - "encoding/json" - lp "github.com/ClusterCockpit/cc-metric-collector/internal/ccMetric" ) @@ -17,7 +15,6 @@ type sink struct { } type Sink interface { - Init(config json.RawMessage) error Write(point lp.CCMetric) error Flush() error Close() diff --git a/sinks/natsSink.go b/sinks/natsSink.go index 187157e..0d7987e 100644 --- a/sinks/natsSink.go +++ b/sinks/natsSink.go @@ -53,30 +53,6 @@ func (s *NatsSink) connect() error { return nil } -func (s *NatsSink) Init(config json.RawMessage) error { - s.name = "NatsSink" - if len(config) > 0 { - err := json.Unmarshal(config, &s.config) - if err != nil { - cclog.ComponentError(s.name, "Error reading config for", s.name, ":", err.Error()) - return err - } - } - if len(s.config.Host) == 0 || - len(s.config.Port) == 0 || - len(s.config.Database) == 0 { - return errors.New("not all configuration variables set required by NatsSink") - } - // Setup Influx line protocol - s.buffer = &bytes.Buffer{} - s.buffer.Grow(1025) - s.encoder = influx.NewEncoder(s.buffer) - s.encoder.SetPrecision(time.Second) - s.encoder.SetMaxLineBytes(1024) - // Setup infos for connection - return s.connect() -} - func (s *NatsSink) Write(m lp.CCMetric) error { if s.client != nil { _, err := s.encoder.Encode(m.ToPoint(s.config.MetaAsTags)) @@ -105,3 +81,31 @@ func (s *NatsSink) Close() { s.client.Close() } } + +func NewNatsSink(name string, config json.RawMessage) (Sink, error) { + s := new(NatsSink) + s.name = fmt.Sprintf("NatsSink(%s)", name) + if len(config) > 0 { + err := json.Unmarshal(config, &s.config) + if err != nil { + cclog.ComponentError(s.name, "Error reading config for", s.name, ":", err.Error()) + return nil, err + } + } + if len(s.config.Host) == 0 || + len(s.config.Port) == 0 || + len(s.config.Database) == 0 { + return nil, errors.New("not all configuration variables set required by NatsSink") + } + // Setup Influx line protocol + s.buffer = &bytes.Buffer{} + s.buffer.Grow(1025) + s.encoder = influx.NewEncoder(s.buffer) + s.encoder.SetPrecision(time.Second) + s.encoder.SetMaxLineBytes(1024) + // Setup infos for connection + if err := s.connect(); err != nil { + return nil, fmt.Errorf("Unable to connect: %v", err) + } + return s, nil +} diff --git a/sinks/prometheusSink.go b/sinks/prometheusSink.go new file mode 100644 index 0000000..5011ac0 --- /dev/null +++ b/sinks/prometheusSink.go @@ -0,0 +1,199 @@ +package sinks + +import ( + "context" + "encoding/json" + "errors" + "fmt" + "net/http" + "strings" + "sync" + + cclog "github.com/ClusterCockpit/cc-metric-collector/internal/ccLogger" + lp "github.com/ClusterCockpit/cc-metric-collector/internal/ccMetric" + "github.com/gorilla/mux" + "github.com/prometheus/client_golang/prometheus" + "github.com/prometheus/client_golang/prometheus/promhttp" +) + +type PrometheusSinkConfig struct { + defaultSinkConfig + Host string `json:"host,omitempty"` + Port string `json:"port"` + Path string `json:"path,omitempty"` + GroupAsNameSpace bool `json:"group_as_namespace,omitempty"` + // User string `json:"user,omitempty"` + // Password string `json:"password,omitempty"` + // FlushDelay string `json:"flush_delay,omitempty"` +} + +type PrometheusSink struct { + sink + config PrometheusSinkConfig + labelMetrics map[string]*prometheus.GaugeVec + nodeMetrics map[string]prometheus.Gauge + promWg sync.WaitGroup + promServer *http.Server +} + +func intToFloat64(input interface{}) (float64, error) { + switch value := input.(type) { + case float64: + return value, nil + case float32: + return float64(value), nil + case int: + return float64(value), nil + case int32: + return float64(value), nil + case int64: + return float64(value), nil + } + return 0, errors.New("cannot cast value to float64") +} + +func getLabelValue(metric lp.CCMetric) []string { + labelValues := []string{} + if tid, tidok := metric.GetTag("type-id"); tidok && metric.HasTag("type") { + labelValues = append(labelValues, tid) + } + if d, ok := metric.GetTag("device"); ok { + labelValues = append(labelValues, d) + } else if d, ok := metric.GetMeta("device"); ok { + labelValues = append(labelValues, d) + } + return labelValues +} + +func getLabelNames(metric lp.CCMetric) []string { + labelNames := []string{} + if t, tok := metric.GetTag("type"); tok && metric.HasTag("type-id") { + labelNames = append(labelNames, t) + } + if _, ok := metric.GetTag("device"); ok { + labelNames = append(labelNames, "device") + } else if _, ok := metric.GetMeta("device"); ok { + labelNames = append(labelNames, "device") + } + return labelNames +} + +func (s *PrometheusSink) newMetric(metric lp.CCMetric) error { + var value float64 = 0 + name := metric.Name() + opts := prometheus.GaugeOpts{ + Name: name, + } + labels := getLabelNames(metric) + labelValues := getLabelValue(metric) + if len(labels) > 0 && len(labels) != len(labelValues) { + return fmt.Errorf("cannot detect metric labels for metric %s", name) + } + + if metricValue, ok := metric.GetField("value"); ok { + if floatValue, err := intToFloat64(metricValue); err == nil { + value = floatValue + } else { + return fmt.Errorf("metric %s with value '%v' cannot be casted to float64", name, metricValue) + } + } + if s.config.GroupAsNameSpace && metric.HasMeta("group") { + g, _ := metric.GetMeta("group") + opts.Namespace = strings.ToLower(g) + } + + if len(labels) > 0 { + new := prometheus.NewGaugeVec(opts, labels) + new.WithLabelValues(labelValues...).Set(value) + s.labelMetrics[name] = new + prometheus.Register(new) + } else { + new := prometheus.NewGauge(opts) + new.Set(value) + s.nodeMetrics[name] = new + prometheus.Register(new) + } + return nil +} + +func (s *PrometheusSink) updateMetric(metric lp.CCMetric) error { + var value float64 = 0.0 + name := metric.Name() + labelValues := getLabelValue(metric) + + if metricValue, ok := metric.GetField("value"); ok { + if floatValue, err := intToFloat64(metricValue); err == nil { + value = floatValue + } else { + return fmt.Errorf("metric %s with value '%v' cannot be casted to float64", name, metricValue) + } + } + + if len(labelValues) > 0 { + if _, ok := s.labelMetrics[name]; !ok { + err := s.newMetric(metric) + if err != nil { + return err + } + } + s.labelMetrics[name].WithLabelValues(labelValues...).Set(value) + } else { + if _, ok := s.labelMetrics[name]; !ok { + err := s.newMetric(metric) + if err != nil { + return err + } + } + s.nodeMetrics[name].Set(value) + } + return nil +} + +func (s *PrometheusSink) Write(m lp.CCMetric) error { + return s.updateMetric(m) +} + +func (s *PrometheusSink) Flush() error { + return nil +} + +func (s *PrometheusSink) Close() { + cclog.ComponentDebug(s.name, "CLOSE") + s.promServer.Shutdown(context.Background()) + s.promWg.Wait() +} + +func NewPrometheusSink(name string, config json.RawMessage) (Sink, error) { + s := new(PrometheusSink) + s.name = "PrometheusSink" + if len(config) > 0 { + err := json.Unmarshal(config, &s.config) + if err != nil { + cclog.ComponentError(s.name, "Error reading config for", s.name, ":", err.Error()) + return nil, err + } + } + if len(s.config.Port) == 0 { + err := errors.New("not all configuration variables set required by PrometheusSink") + cclog.ComponentError(s.name, err.Error()) + return nil, err + } + s.labelMetrics = make(map[string]*prometheus.GaugeVec) + s.nodeMetrics = make(map[string]prometheus.Gauge) + s.promWg.Add(1) + go func() { + router := mux.NewRouter() + // Prometheus endpoint + router.Path("/" + s.config.Path).Handler(promhttp.Handler()) + + url := fmt.Sprintf("%s:%s", s.config.Host, s.config.Port) + cclog.ComponentDebug(s.name, "Serving Prometheus metrics at", fmt.Sprintf("%s:%s/%s", s.config.Host, s.config.Port, s.config.Path)) + s.promServer = &http.Server{Addr: url, Handler: router} + err := s.promServer.ListenAndServe() + if err != nil && err.Error() != "http: Server closed" { + cclog.ComponentError(s.name, err.Error()) + } + s.promWg.Done() + }() + return s, nil +} diff --git a/sinks/prometheusSink.md b/sinks/prometheusSink.md new file mode 100644 index 0000000..0996c43 --- /dev/null +++ b/sinks/prometheusSink.md @@ -0,0 +1,23 @@ +## `prometheus` sink + +The `prometheus` sink publishes all metrics via an HTTP server ready to be scraped by a [Prometheus](https://prometheus.io) server. It creates gauge metrics for all node metrics and gauge vectors for all metrics with a subtype like 'device', 'cpu' or 'socket'. + + +### Configuration structure + +```json +{ + "": { + "type": "prometheus", + "host": "localhost", + "port": "8080", + "path": "metrics" + } +} +``` + +- `type`: makes the sink an `prometheus` sink +- `host`: The HTTP server gets bound to that IP/hostname +- `port`: Portnumber (as string) for the HTTP server +- `path`: Path where the metrics should be servered. The metrics will be published at `host`:`port`/`path` +- `group_as_namespace`: Most metrics contain a group as meta information like 'memory', 'load'. With this the metric names are extended to `group`_`name` if possible. diff --git a/sinks/sampleSink.go b/sinks/sampleSink.go new file mode 100644 index 0000000..be0196a --- /dev/null +++ b/sinks/sampleSink.go @@ -0,0 +1,62 @@ +package sinks + +import ( + "encoding/json" + "fmt" + "log" + + cclog "github.com/ClusterCockpit/cc-metric-collector/internal/ccLogger" + lp "github.com/ClusterCockpit/cc-metric-collector/internal/ccMetric" +) + +type SampleSinkConfig struct { + defaultSinkConfig // defines JSON tags for 'type' and 'meta_as_tags' + // Add additional options +} + +type SampleSink struct { + sink // declarate 'name' and 'meta_as_tags' + config SampleSinkConfig // entry point to the SampleSinkConfig +} + +// Code to submit a single CCMetric to the sink +func (s *SampleSink) Write(point lp.CCMetric) error { + log.Print(point) + return nil +} + +// If the sink uses batched sends internally, you can tell to flush its buffers +func (s *SampleSink) Flush() error { + return nil +} + +// Close sink: close network connection, close files, close libraries, ... +func (s *SampleSink) Close() { + cclog.ComponentDebug(s.name, "CLOSE") +} + +// New function to create a new instance of the sink +// Initialize the sink by giving it a name and reading in the config JSON +func NewSampleSink(name string, config json.RawMessage) (Sink, error) { + s := new(SampleSink) + s.name = fmt.Sprintf("SampleSink(%s)", name) // Always specify a name here + + // Set defaults in s.config + + // Read in the config JSON + if len(config) > 0 { + err := json.Unmarshal(config, &s.config) + if err != nil { + return nil, err + } + } + + // Check if all required fields in the config are set + // Use 'len(s.config.Option) > 0' for string settings + + // Establish connection to the server, library, ... + // Check required files exist and lookup path(s) of executable(s) + + // Return (nil, meaningful error message) in case of errors + return s, nil +} diff --git a/sinks/sinkManager.go b/sinks/sinkManager.go index 487e7ca..f531f5d 100644 --- a/sinks/sinkManager.go +++ b/sinks/sinkManager.go @@ -13,14 +13,14 @@ import ( const SINK_MAX_FORWARD = 50 // Map of all available sinks -var AvailableSinks = map[string]Sink{ - "influxdb": new(InfluxSink), - "stdout": new(StdoutSink), - "nats": new(NatsSink), - "http": new(HttpSink), - "ganglia": new(GangliaSink), - "influxasync": new(InfluxAsyncSink), - "libganglia": new(LibgangliaSink), +var AvailableSinks = map[string]func(name string, config json.RawMessage) (Sink, error){ + "ganglia": NewGangliaSink, + "libganglia": NewLibgangliaSink, + "stdout": NewStdoutSink, + "nats": NewNatsSink, + "influxdb": NewInfluxSink, + "influxasync": NewInfluxAsyncSink, + "http": NewHttpSink, } // Metric collector manager data structure @@ -149,8 +149,7 @@ func (sm *sinkManager) AddOutput(name string, rawConfig json.RawMessage) error { cclog.ComponentError("SinkManager", "SKIP", name, "unknown sink:", sinkConfig.Type) return err } - s := AvailableSinks[sinkConfig.Type] - err = s.Init(rawConfig) + s, err := AvailableSinks[sinkConfig.Type](name, rawConfig) if err != nil { cclog.ComponentError("SinkManager", "SKIP", s.Name(), "initialization failed:", err.Error()) return err diff --git a/sinks/stdoutSink.go b/sinks/stdoutSink.go index 5d0761a..acf2621 100644 --- a/sinks/stdoutSink.go +++ b/sinks/stdoutSink.go @@ -19,34 +19,6 @@ type StdoutSink struct { } } -func (s *StdoutSink) Init(config json.RawMessage) error { - s.name = "StdoutSink" - if len(config) > 0 { - err := json.Unmarshal(config, &s.config) - if err != nil { - return err - } - } - - s.output = os.Stdout - if len(s.config.Output) > 0 { - switch strings.ToLower(s.config.Output) { - case "stdout": - s.output = os.Stdout - case "stderr": - s.output = os.Stderr - default: - f, err := os.OpenFile(s.config.Output, os.O_CREATE|os.O_WRONLY, os.FileMode(0600)) - if err != nil { - return err - } - s.output = f - } - } - s.meta_as_tags = s.config.MetaAsTags - return nil -} - func (s *StdoutSink) Write(m lp.CCMetric) error { fmt.Fprint( s.output, @@ -65,3 +37,33 @@ func (s *StdoutSink) Close() { s.output.Close() } } + +func NewStdoutSink(name string, config json.RawMessage) (Sink, error) { + s := new(StdoutSink) + s.name = fmt.Sprintf("StdoutSink(%s)", name) + if len(config) > 0 { + err := json.Unmarshal(config, &s.config) + if err != nil { + return nil, err + } + } + + s.output = os.Stdout + if len(s.config.Output) > 0 { + switch strings.ToLower(s.config.Output) { + case "stdout": + s.output = os.Stdout + case "stderr": + s.output = os.Stderr + default: + f, err := os.OpenFile(s.config.Output, os.O_CREATE|os.O_WRONLY, os.FileMode(0600)) + if err != nil { + return nil, err + } + s.output = f + } + } + s.meta_as_tags = s.config.MetaAsTags + + return s, nil +} From 9ece27eec6038dd8cd4d3c567eb6798f9db9ec00 Mon Sep 17 00:00:00 2001 From: Thomas Roehl Date: Thu, 3 Mar 2022 17:26:32 +0100 Subject: [PATCH 2/3] Use right systemd macro to create the user --- scripts/cc-metric-collector.spec | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/scripts/cc-metric-collector.spec b/scripts/cc-metric-collector.spec index fef1bb8..00a4aa4 100644 --- a/scripts/cc-metric-collector.spec +++ b/scripts/cc-metric-collector.spec @@ -42,7 +42,7 @@ install -Dpm 0644 scripts/%{name}.sysusers %{buildroot}%{_sysusersdir}/%{name}.c # go test should be here... :) %pre -%sysusers_create_compat scripts/%{name}.sysusers +%sysusers_create_package scripts/%{name}.sysusers %post %systemd_post %{name}.service From 31f10b3163deb7929d8613d312627d4a7039ad1b Mon Sep 17 00:00:00 2001 From: Thomas Roehl Date: Thu, 3 Mar 2022 17:31:15 +0100 Subject: [PATCH 3/3] Fix user creation type --- scripts/cc-metric-collector.sysusers | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/scripts/cc-metric-collector.sysusers b/scripts/cc-metric-collector.sysusers index 65646af..5c17ec7 100644 --- a/scripts/cc-metric-collector.sysusers +++ b/scripts/cc-metric-collector.sysusers @@ -1,2 +1,2 @@ #Type Name ID GECOS Home directory Shell -s hpcop - "User for ClusterCockpit" /run/cc-metric-collector /sbin/nologin +u hpcop - "User for ClusterCockpit" /run/cc-metric-collector /sbin/nologin