One of the great features of nnodes is a direct task logging ability while the jobs is running. We didn't know we needed this until we had it. Very simply put, it's a command line tool that let's you keep track of all jobs that are running, done, failed, or to be submitted. The example output for two moment tensor inversions is as follows:
In the backend, nnodes uses a dict to keep track of submission times, execution times etc. with start and endtime attributes for each task. The attributes are simply read and printed after reading the dictionary.
What I imagine is quite similar that could be called like:
nnodes link
One of the great features of
nnodes
is a direct task logging ability while the jobs is running. We didn't know we needed this until we had it. Very simply put, it's a command line tool that let's you keep track of all jobs that are running, done, failed, or to be submitted. The example output for two moment tensor inversions is as follows:where
-
indicates tasks that can be run concurrently1,2,3,..
indicate sequentially run tasks(hh:mm:ss)
stamp(running - hh:mm:ss)
stampIn the backend,
nnodes
uses adict
to keep track of submission times, execution times etc. with start and endtime attributes for each task. The attributes are simply read and printed after reading the dictionary.What I imagine is quite similar that could be called like:
and output a log with
-
for pipelines1,2,3,...
for stages-
again for tasksand similar timestamps.