osd/scheduler: add mclock queue length perfcounter#58302
Conversation
|
@sseshasa Please REVIEW here |
|
Test results with this PR |
|
Old PR Closed: #55295 |
61d7d56 to
c951f55
Compare
|
@athanatos review thanks |
509b86b to
fa06b83
Compare
7d72bbf to
7910725
Compare
sseshasa
left a comment
There was a problem hiding this comment.
@jianwei1216 The changes look good to me. Thanks! I will tag the PR for it to be picked up for CI testing.
7910725 to
f204ee5
Compare
f204ee5 to
16e2ec5
Compare
|
jenkins test API |
|
needs-qa @sseshasa |
16e2ec5 to
fad64e4
Compare
Easy to visualize the number of ops in each subqueue of mclock.
Add request statistics for each op_scheduler_class
daemon or tell command:
ceph daemon osd.0 perf dump mclock-0
{
"mclock-shard-queue-0": {
"mclock_immediate_queue_len": 0,
"mclock_client_queue_len": 0,
"mclock_recovery_queue_len": 0,
"mclock_best_effort_queue_len": 0
"mclock_all_type_queue_len": 0
}
}
Fixes: https://tracker.ceph.com/issues/66714
Signed-off-by: zhangjianwei2 <zhangjianwei2@cmss.chinamobile.com>
fad64e4 to
6db763e
Compare
|
https://jenkins.ceph.com/job/ceph-pull-requests/139327/console test host: Could not resolve host: github.com @sseshasa |
6db763e to
0c51699
Compare
when unittest_mclock_scheduler, dead loop in PerfCountersCollectionImpl::add() Fixes: https://tracker.ceph.com/issues/66758 Signed-off-by: zhangjianwei2 <zhangjianwei2@cmss.chinamobile.com>
0c51699 to
16b4a89
Compare
|
jenkins test make check |
3 similar comments
|
jenkins test make check |
|
jenkins test make check |
|
jenkins test make check |
|
@sseshasa: is this ready for QA? |
Yes, the latest push fixed the issue with logger removal. I will add the 'needs-qa' label. |
|
@jianwei1216: do we need backports for this? |
need backports this PR to reef and squid |
|
@rzarzynski @jianwei1216 Yes, this can be backported to Reef and Squid. @jianwei1216 I can take this up. |



feature: https://tracker.ceph.com/issues/66714