Closed milroy closed 2 weeks ago
After many attempts I was unsuccessful at getting the error to propagate to flux dmesg
. I updated the testsuite test to use resource-query
instead. Let me know if I should keep poking at the resource module-based test.
I think the PR is ready for review.
I updated the testsuite test to use resource-query instead.
(The updated test based on resource-query
fails without the change in this PR and succeeds with it.)
Thanks for the reviews. Setting MWP.
All modified and coverable lines are covered by tests :white_check_mark:
Project coverage is 75.2%. Comparing base (
1c1f8d6
) to head (ee57490
). Report is 4 commits behind head on master.
Issue #1260 and logs from production clusters have reported match errors such as
by_subplan: planner_multi_avail_during returned -1.
for JGFs withnode
vertices with an edge to thecluster
vertex. The error occurs because PR #1248 prevents creation of subplan filters on leaf vertices, causing subsequent checks against the planner to fail.This PR adds a check to
by_subplan
to bypass pruning based on null subplans.Currently, the test case succeeds without the fix because the error isn't logged in
flux dmesg
for some reason.