You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Problem: it's difficult to know what's going on inside the job manager. For example, a pending alloc request is "retracted" (flag indicating that it's outstanding cleared) when the scheduler is unloaded, and then the alloc request is resent when the scheduler is loaded again.
One idea is to add a debug flag at job submission. If set, the job manager could emit verbose debug eventlog entries for the job. The key events that drive the job state machine are already in there. This would surround it with some further context about what's happening to use as a trace for testing, or a debugging tool.
The text was updated successfully, but these errors were encountered:
garlick
changed the title
add flux job submit --debug flag
job-manager: add flux job submit --debug flag
Feb 25, 2019
garlick
added a commit
to garlick/flux-core
that referenced
this issue
Feb 27, 2019
Add a new flag to job.h for flux_job_submit(),
FLUX_JOB_DEBUG, to enable eventlog debugging.
Then add flux job submit [--debug] option to
allow it to be set on the command line at submit time.
Fixesflux-framework#2033.
Add a new flag to job.h for flux_job_submit(),
FLUX_JOB_DEBUG, to enable eventlog debugging.
Then add flux job submit [--debug] option to
allow it to be set on the command line at submit time.
Fixesflux-framework#2033.
garlick
added a commit
to garlick/flux-core
that referenced
this issue
Feb 27, 2019
Add a new flag to job.h for flux_job_submit(),
FLUX_JOB_DEBUG, to enable eventlog debugging.
Then add flux job submit [--debug] option to
allow it to be set on the command line at submit time.
Fixesflux-framework#2033.
Problem: it's difficult to know what's going on inside the job manager. For example, a pending alloc request is "retracted" (flag indicating that it's outstanding cleared) when the scheduler is unloaded, and then the alloc request is resent when the scheduler is loaded again.
One idea is to add a debug flag at job submission. If set, the job manager could emit verbose
debug
eventlog entries for the job. The key events that drive the job state machine are already in there. This would surround it with some further context about what's happening to use as a trace for testing, or a debugging tool.The text was updated successfully, but these errors were encountered: