openconfig / featureprofiles / 40426879105

7%

push

github

Modified supervisor_failure_test.go (#5871)

* Modified supervisor_failure_test for b/456351949

* Added time.Sleep(5 * time.Second) while waiting for the gRIBI server to come up

* TE-8.2: flush both supervisors before test to avoid stale-state false positives

Only the active supervisor can be flushed through gRIBI, so the supervisor that was active at the start of a run kept its TE-8.2.1 PRESERVE entries after the switchover. On the next run it became the new active supervisor and the post-switchover AFT check passed on those stale entries (seen on Juniper PTX10008, which does not synchronize gRIBI state to the backup RE).

Add a pre-test cleanup that flushes the active supervisor, switches over, flushes the other supervisor and waits up to 30 minutes for the standby to become switchover-ready again (same timeout as other switchover tests; Arista needs about 22 minutes). Also factor the gRIBI reconnect retry loop into startGRIBIClientWithRetry and move the dual-controller check to the start of the test.

---------

Co-authored-by: jaya kumar <101202728+jayakumarramisetty@users.noreply.github.com>

0 of 36 new or added lines in 1 file covered. (0.0%)

2339 of 32208 relevant lines covered (7.26%)

0.08 hits per line