Merge branch 'import-data-sanitation' of https://github.com/ClusterCockpit/cc-backend into import-data-sanitation

This commit is contained in:
Christoph Kluge 2023-03-31 17:18:49 +02:00
commit 2401a2b940
12 changed files with 150 additions and 65 deletions

4
.gitignore vendored
View File

@ -9,6 +9,6 @@
/web/frontend/public/build
/web/frontend/node_modules
.vscode/settings.json
/.vscode/*
/archive-migration
.vscode/launch.json
/archive-manager

View File

@ -15,6 +15,7 @@
"kind": "file",
"path": "./var/job-archive"
},
"validate": true,
"clusters": [
{
"name": "test",
@ -24,9 +25,18 @@
"token": "eyJhbGciOiJF-E-pQBQ"
},
"filterRanges": {
"numNodes": { "from": 1, "to": 64 },
"duration": { "from": 0, "to": 86400 },
"startTime": { "from": "2022-01-01T00:00:00Z", "to": null }
"numNodes": {
"from": 1,
"to": 64
},
"duration": {
"from": 0,
"to": 86400
},
"startTime": {
"from": "2022-01-01T00:00:00Z",
"to": null
}
}
}
]

View File

@ -228,7 +228,9 @@ func InitDB() error {
i := 0
errorOccured := 0
for jobMeta := range ar.Iter() {
for jobContainer := range ar.Iter(false) {
jobMeta := jobContainer.Meta
// // Bundle 100 inserts into one transaction for better performance:
if i%10 == 0 {

View File

@ -30,7 +30,12 @@ type ArchiveBackend interface {
GetClusters() []string
Iter() <-chan *schema.JobMeta
Iter(loadMetricData bool) <-chan JobContainer
}
type JobContainer struct {
Meta *schema.JobMeta
Data *schema.JobData
}
var cache *lrucache.Cache = lrucache.New(128 * 1024 * 1024)

View File

@ -21,6 +21,7 @@ import (
"github.com/ClusterCockpit/cc-backend/internal/config"
"github.com/ClusterCockpit/cc-backend/pkg/log"
"github.com/ClusterCockpit/cc-backend/pkg/schema"
"github.com/santhosh-tekuri/jsonschema/v5"
)
type FsArchiveConfig struct {
@ -52,14 +53,52 @@ func getPath(
func loadJobMeta(filename string) (*schema.JobMeta, error) {
f, err := os.Open(filename)
b, err := os.ReadFile(filename)
if err != nil {
log.Errorf("fsBackend loadJobMeta()- %v", err)
return &schema.JobMeta{}, err
}
defer f.Close()
if config.Keys.Validate {
if err := schema.Validate(schema.Meta, bytes.NewReader(b)); err != nil {
return &schema.JobMeta{}, fmt.Errorf("validate job meta: %v", err)
}
}
return DecodeJobMeta(bufio.NewReader(f))
return DecodeJobMeta(bytes.NewReader(b))
}
func loadJobData(filename string, isCompressed bool) (schema.JobData, error) {
f, err := os.Open(filename)
if err != nil {
log.Errorf("fsBackend LoadJobData()- %v", err)
return nil, err
}
if isCompressed {
r, err := gzip.NewReader(f)
if err != nil {
log.Errorf(" %v", err)
return nil, err
}
defer r.Close()
if config.Keys.Validate {
if err := schema.Validate(schema.Data, r); err != nil {
return schema.JobData{}, fmt.Errorf("validate job data: %v", err)
}
}
return DecodeJobData(r, filename)
} else {
defer f.Close()
if config.Keys.Validate {
if err := schema.Validate(schema.Data, bufio.NewReader(f)); err != nil {
return schema.JobData{}, fmt.Errorf("validate job data: %v", err)
}
}
return DecodeJobData(bufio.NewReader(f), filename)
}
}
func (fsa *FsArchive) Init(rawConfig json.RawMessage) (int, error) {
@ -116,24 +155,8 @@ func (fsa *FsArchive) LoadJobData(job *schema.Job) (schema.JobData, error) {
filename = getPath(job, fsa.path, "data.json")
isCompressed = false
}
f, err := os.Open(filename)
if err != nil {
log.Errorf("fsBackend LoadJobData()- %v", err)
return nil, err
}
defer f.Close()
if isCompressed {
r, err := gzip.NewReader(f)
if err != nil {
log.Errorf(" %v", err)
return nil, err
}
return DecodeJobData(r, filename)
} else {
return DecodeJobData(bufio.NewReader(f), filename)
}
return loadJobData(filename, isCompressed)
}
func (fsa *FsArchive) LoadJobMeta(job *schema.Job) (*schema.JobMeta, error) {
@ -157,9 +180,9 @@ func (fsa *FsArchive) LoadClusterCfg(name string) (*schema.Cluster, error) {
return DecodeCluster(bytes.NewReader(b))
}
func (fsa *FsArchive) Iter() <-chan *schema.JobMeta {
func (fsa *FsArchive) Iter(loadMetricData bool) <-chan JobContainer {
ch := make(chan *schema.JobMeta)
ch := make(chan JobContainer)
go func() {
clustersDir, err := os.ReadDir(fsa.path)
if err != nil {
@ -196,10 +219,26 @@ func (fsa *FsArchive) Iter() <-chan *schema.JobMeta {
for _, startTimeDir := range startTimeDirs {
if startTimeDir.IsDir() {
job, err := loadJobMeta(filepath.Join(dirpath, startTimeDir.Name(), "meta.json"))
if err != nil {
if err != nil && !errors.Is(err, &jsonschema.ValidationError{}) {
log.Errorf("in %s: %s", filepath.Join(dirpath, startTimeDir.Name()), err.Error())
}
if loadMetricData {
var isCompressed bool = true
filename := filepath.Join(dirpath, startTimeDir.Name(), "data.json.gz")
if !checkFileExists(filename) {
filename = filepath.Join(dirpath, startTimeDir.Name(), "data.json")
isCompressed = false
}
data, err := loadJobData(filename, isCompressed)
if err != nil && !errors.Is(err, &jsonschema.ValidationError{}) {
log.Errorf("in %s: %s", filepath.Join(dirpath, startTimeDir.Name()), err.Error())
}
ch <- JobContainer{Meta: job, Data: &data}
} else {
ch <- job
ch <- JobContainer{Meta: job, Data: nil}
}
}
}

View File

@ -154,10 +154,10 @@ func TestIter(t *testing.T) {
t.Fatal(err)
}
for job := range fsa.Iter() {
fmt.Printf("Job %d\n", job.JobID)
for job := range fsa.Iter(false) {
fmt.Printf("Job %d\n", job.Meta.JobID)
if job.Cluster != "emmy" {
if job.Meta.Cluster != "emmy" {
t.Fail()
}
}

View File

@ -338,7 +338,7 @@
"user",
"project",
"cluster",
"subcluster",
"subCluster",
"numNodes",
"exclusive",
"startTime",

View File

@ -193,7 +193,7 @@
},
"data": {
"type": "array",
"items": {
"contains": {
"type": "number",
"minimum": 0
},

View File

@ -182,6 +182,7 @@
"walltime": 10,
"jobState": "completed",
"cluster": "emmy",
"subCluster": "haswell",
"stopTime": 1609009562,
"user": "emmyUser6",
"startTime": 1608923076,

View File

@ -100,6 +100,7 @@
],
"walltime": 10,
"cluster": "emmy",
"subCluster": "haswell",
"jobState": "completed",
"statistics": {
"clock": {

View File

@ -4,6 +4,32 @@
// license that can be found in the LICENSE file.
package main
func main() {
import (
"encoding/json"
"flag"
"fmt"
"log"
"github.com/ClusterCockpit/cc-backend/internal/config"
"github.com/ClusterCockpit/cc-backend/pkg/archive"
)
func main() {
var srcPath, flagConfigFile string
flag.StringVar(&srcPath, "s", "./var/job-archive", "Specify the source job archive path. Default is ./var/job-archive")
flag.StringVar(&flagConfigFile, "config", "./config.json", "Specify alternative path to `config.json`")
flag.Parse()
archiveCfg := fmt.Sprintf("{\"kind\": \"file\",\"path\": \"%s\"}", srcPath)
config.Init(flagConfigFile)
if err := archive.Init(json.RawMessage(archiveCfg)); err != nil {
log.Fatal(err)
}
ar := archive.GetHandle()
for job := range ar.Iter(true) {
log.Printf("Validate %s - %d\n", job.Meta.Cluster, job.Meta.JobID)
}
}

View File

@ -180,6 +180,7 @@ func main() {
flag.StringVar(&srcPath, "s", "./var/job-archive", "Specify the source job archive path. Default is ./var/job-archive")
flag.StringVar(&dstPath, "d", "./var/job-archive-new", "Specify the destination job archive path. Default is ./var/job-archive-new")
flag.Parse()
if _, err := os.Stat(filepath.Join(srcPath, "version.txt")); !errors.Is(err, os.ErrNotExist) {
log.Fatal("Archive version exists!")