ClusterCockpit/cc-backend
1
0
Fork 0
mirror of https://github.com/ClusterCockpit/cc-backend synced 2026-02-11 21:41:46 +01:00
Code Issues Packages Releases Activity

fix: Transfer always to main job table before archiving

Browse Source
This commit is contained in:
Jan Eitzinger
2026-02-09 19:57:46 +01:00
parent 51e9d33f9f
commit ac7eb93141
4 changed files with 91 additions and 58 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
internal/api/job.go
View File

@@ -754,6 +754,7 @@ func (api *RestAPI) stopJobByRequest(rw http.ResponseWriter, r *http.Request) {
return
}
isCached := false
job, err = api.JobRepository.Find(req.JobID, req.Cluster, req.StartTime)
if err != nil {
// Try cached jobs if not found in main repository
@@ -764,9 +765,10 @@ func (api *RestAPI) stopJobByRequest(rw http.ResponseWriter, r *http.Request) {
return
}
job = cachedJob
isCached = true
}
api.checkAndHandleStopJob(rw, job, req)
api.checkAndHandleStopJob(rw, job, req, isCached)
}
// deleteJobByID godoc
@@ -923,7 +925,7 @@ func (api *RestAPI) deleteJobBefore(rw http.ResponseWriter, r *http.Request) {
}
}
func (api *RestAPI) checkAndHandleStopJob(rw http.ResponseWriter, job *schema.Job, req StopJobAPIRequest) {
func (api *RestAPI) checkAndHandleStopJob(rw http.ResponseWriter, job *schema.Job, req StopJobAPIRequest, isCached bool) {
// Sanity checks
if job.State != schema.JobStateRunning {
handleError(fmt.Errorf("jobId %d (id %d) on %s : job has already been stopped (state is: %s)", job.JobID, *job.ID, job.Cluster, job.State), http.StatusUnprocessableEntity, rw)
@@ -948,11 +950,21 @@ func (api *RestAPI) checkAndHandleStopJob(rw http.ResponseWriter, job *schema.Jo
api.JobRepository.Mutex.Lock()
defer api.JobRepository.Mutex.Unlock()
if err := api.JobRepository.Stop(*job.ID, job.Duration, job.State, job.MonitoringStatus); err != nil {
if err := api.JobRepository.StopCached(*job.ID, job.Duration, job.State, job.MonitoringStatus); err != nil {
handleError(fmt.Errorf("jobId %d (id %d) on %s : marking job as '%s' (duration: %d) in DB failed: %w", job.JobID, *job.ID, job.Cluster, job.State, job.Duration, err), http.StatusInternalServerError, rw)
// If the job is still in job_cache, transfer it to the job table first
// so that job.ID always points to the job table for downstream code
if isCached {
newID, err := api.JobRepository.TransferCachedJobToMain(*job.ID)
if err != nil {
handleError(fmt.Errorf("jobId %d (id %d) on %s : transferring cached job failed: %w", job.JobID, *job.ID, job.Cluster, err), http.StatusInternalServerError, rw)
return
}
cclog.Infof("transferred cached job to main table: old id %d -> new id %d (jobId=%d)", *job.ID, newID, job.JobID)
job.ID = &newID
}
if err := api.JobRepository.Stop(*job.ID, job.Duration, job.State, job.MonitoringStatus); err != nil {
handleError(fmt.Errorf("jobId %d (id %d) on %s : marking job as '%s' (duration: %d) in DB failed: %w", job.JobID, *job.ID, job.Cluster, job.State, job.Duration, err), http.StatusInternalServerError, rw)
return
}
cclog.Infof("archiving job... (dbid: %d): cluster=%s, jobId=%d, user=%s, startTime=%d, duration=%d, state=%s", *job.ID, job.Cluster, job.JobID, job.User, job.StartTime, job.Duration, job.State)

22
internal/api/nats.go
View File

@@ -251,6 +251,7 @@ func (api *NatsAPI) handleStopJob(payload string) {
return
}
isCached := false
job, err := api.JobRepository.Find(req.JobID, req.Cluster, req.StartTime)
if err != nil {
cachedJob, cachedErr := api.JobRepository.FindCached(req.JobID, req.Cluster, req.StartTime)
@@ -260,6 +261,7 @@ func (api *NatsAPI) handleStopJob(payload string) {
return
}
job = cachedJob
isCached = true
}
if job.State != schema.JobStateRunning {
@@ -287,16 +289,26 @@ func (api *NatsAPI) handleStopJob(payload string) {
api.JobRepository.Mutex.Lock()
defer api.JobRepository.Mutex.Unlock()
if err := api.JobRepository.Stop(*job.ID, job.Duration, job.State, job.MonitoringStatus); err != nil {
if err := api.JobRepository.StopCached(*job.ID, job.Duration, job.State, job.MonitoringStatus); err != nil {
cclog.Errorf("NATS job stop: jobId %d (id %d) on %s: marking job as '%s' failed: %v",
job.JobID, job.ID, job.Cluster, job.State, err)
// If the job is still in job_cache, transfer it to the job table first
if isCached {
newID, err := api.JobRepository.TransferCachedJobToMain(*job.ID)
if err != nil {
cclog.Errorf("NATS job stop: jobId %d (id %d) on %s: transferring cached job failed: %v",
job.JobID, *job.ID, job.Cluster, err)
return
}
cclog.Infof("NATS: transferred cached job to main table: old id %d -> new id %d (jobId=%d)", *job.ID, newID, job.JobID)
job.ID = &newID
}
if err := api.JobRepository.Stop(*job.ID, job.Duration, job.State, job.MonitoringStatus); err != nil {
cclog.Errorf("NATS job stop: jobId %d (id %d) on %s: marking job as '%s' failed: %v",
job.JobID, *job.ID, job.Cluster, job.State, err)
return
}
cclog.Infof("NATS: archiving job (dbid: %d): cluster=%s, jobId=%d, user=%s, startTime=%d, duration=%d, state=%s",
job.ID, job.Cluster, job.JobID, job.User, job.StartTime, job.Duration, job.State)
*job.ID, job.Cluster, job.JobID, job.User, job.StartTime, job.Duration, job.State)
if job.MonitoringStatus == schema.MonitoringStatusDisabled {
return