diff options
| author | Arun Isaac | 2024-10-27 02:54:14 +0000 |
|---|---|---|
| committer | Arun Isaac | 2024-11-06 00:37:10 +0000 |
| commit | dcdfae66509895ec6513b1d0ada365e89abf869f (patch) | |
| tree | b90b66cccf7b80c2073dcffb5773f43e0b821071 | |
| parent | 2f1fb017a9330fe5c11719a297140ec6896c3c6c (diff) | |
| download | ravanan-dcdfae66509895ec6513b1d0ada365e89abf869f.tar.gz ravanan-dcdfae66509895ec6513b1d0ada365e89abf869f.tar.lz ravanan-dcdfae66509895ec6513b1d0ada365e89abf869f.zip | |
slurm-api: Find state of purged jobs.
* ravanan/slurm-api.scm (job-state): Query slurmdb for jobs that have been purged from slurmctld's active memory.
| -rw-r--r-- | ravanan/slurm-api.scm | 59 |
1 files changed, 45 insertions, 14 deletions
diff --git a/ravanan/slurm-api.scm b/ravanan/slurm-api.scm index 8e7abd7..ea56775 100644 --- a/ravanan/slurm-api.scm +++ b/ravanan/slurm-api.scm @@ -108,17 +108,48 @@ FAQ}) to request for the job." "Query the state of slurm @var{job-id} via @var{api-endpoint} authenticating using @var{jwt}. Return value is one of the symbols @code{pending}, @code{failed} and @code{completed}." - ;; TODO: What if job has been "archived"? Then, look up archived - ;; jobs too. - (let ((response (check-api-error - (slurm-http-get api-endpoint - jwt - (string-append "/slurm/v0.0.41/job/" - (number->string job-id)))))) - (match (json-ref (find (lambda (job) - (= (json-ref job "job_id") - job-id)) - (vector->list (json-ref response "jobs"))) - "job_state") - (#(job-state) - (string->symbol (string-downcase job-state)))))) + (let ((response (slurm-http-get api-endpoint + jwt + (string-append "/slurm/v0.0.41/job/" + (number->string job-id))))) + (match (json-ref response "errors") + (#() + (match (json-ref (find (lambda (job) + (= (json-ref job "job_id") + job-id)) + (vector->list (json-ref response "jobs"))) + "job_state") + (#(job-state) + (string->symbol (string-downcase job-state))))) + (#(errors ...) + ;; Check in slurmdbd if job has been completed and purged from + ;; slurmctld's active memory. + (match (find (lambda (error) + (= (json-ref error "error_number") + ;; Error number 2017 (Invalid job id specified) may + ;; have occurred because the job has completed, has + ;; exceeded MinJobAge (as set in slurm.conf) and has + ;; therefore been purged from slurmctld's active + ;; memory. + 2017)) + errors) + (error-2017 + (let ((response + (check-api-error + (slurm-http-get api-endpoint + jwt + (string-append "/slurmdb/v0.0.41/job/" + (number->string job-id)))))) + (match (json-ref (find (lambda (job) + (= (json-ref job "job_id") + job-id)) + (vector->list (json-ref response "jobs"))) + "exit_code" "status") + (#(job-state) + ;; job-state is either "SUCCESS" or "ERROR". + (if (eq? (string->symbol (string-downcase job-state)) + 'success) + 'success + 'failed))))) + (#f + (check-api-error response))))))) |