]> git.apps.os.sepia.ceph.com Git - ceph.git/commit
qa/standalone/scrub: Force a subset of scrub tests to use "wpq" scheduler
authorSridhar Seshasayee <sseshasa@redhat.com>
Mon, 21 Jun 2021 12:47:32 +0000 (18:17 +0530)
committerSridhar Seshasayee <sseshasa@redhat.com>
Fri, 30 Jul 2021 12:46:00 +0000 (18:16 +0530)
commit33d2a2c93b533838f8efe1329181fb8990b4a511
tree9b98ababc7ea7ffa8526da60b059b842c5fec7a4
parentf658ff3511237d689e82a7f42188c5de9e3d1089
qa/standalone/scrub: Force a subset of scrub tests to use "wpq" scheduler

The following tests in the test files mentioned below use the
"osd_scrub_sleep" option to introduce delays during scrubbing to help
determine scrubbing states, validate reservations during scrubbing etc..
This works when using the "wpq" scheduler.

But when the "mclock_scheduler" is enabled, the "osd_scrub_sleep" is
disabled and overridden to 0. This is done to delegate the scheduling of
the background scrubs to the "mclock_scheduler" based on the set QoS
parameters. Due to this, the checks to verify the scrub states,
reservations etc. fail since the window to check them is very short
due to scrubs completing very quickly. This affects a small subset of
scrub tests mentioned below,

1. osd-scrub-dump.sh -> TEST_recover_unexpected()
2. osd-scrub-repair.sh -> TEST_auto_repair_bluestore_tag()
3. osd-scrub-test.sh -> TEST_scrub_abort(), TEST_deep_scrub_abort()

Only for the above tests, until there's a reliable way to query scrub
states with "--osd-scrub-sleep" set to 0, the "osd_op_queue" config
option is set to "wpq".

Signed-off-by: Sridhar Seshasayee <sseshasa@redhat.com>
qa/standalone/scrub/osd-scrub-dump.sh
qa/standalone/scrub/osd-scrub-repair.sh
qa/standalone/scrub/osd-scrub-test.sh