missing perf.cache_references will prevent getting meter of perf.instructions

Bug #1670948 reported by Lianhao Lu
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Ceilometer
Fix Released
Undecided
Ren Qiaowei

Bug Description

when running the perf.instructions pollster on compute node, the missing of perf.cache_references in the inspector data will prevent getting the meter of perf.instructions which resides in the data got from inspector.

2017-03-07 13:38:01.221 106045 INFO ceilometer.agent.manager [-] Polling pollster perf.instructions in the context of all_pollsters
2017-03-07 13:38:01.223 106045 WARNING ceilometer.compute.pollsters.perf [-] Cannot inspect data of PerfEventsInstructionsPollster for 01a40df5-4a1b-4f02-b623-df429291aa1a, non-fatal reason: Perf is not supported by current version of libvirt, and failed to inspect perf events of 01a40df5-4a1b-4f02-b623-df429291aa1a, can not get info from libvirt: 'perf.cache_references'
2017-03-07 13:38:01.223 106045 ERROR ceilometer.agent.manager [-] Prevent pollster perf.instructions from polling [<NovaLikeServer: vm4>, <NovaLikeServer: vm3>] on source all_pollsters anymore!

Changed in ceilometer:
assignee: nobody → Ren Qiaowei (qiaowei-ren)
Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix proposed to ceilometer (master)

Fix proposed to branch: master
Review: https://review.openstack.org/443452

Changed in ceilometer:
status: New → In Progress
Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Related fix merged to ceilometer (master)

Reviewed: https://review.openstack.org/445976
Committed: https://git.openstack.org/cgit/openstack/ceilometer/commit/?id=3304cf26421d77b40225a886782d46c73439c0e9
Submitter: Jenkins
Branch: master

commit 3304cf26421d77b40225a886782d46c73439c0e9
Author: Mehdi Abaakouk <email address hidden>
Date: Wed Mar 15 14:53:29 2017 +0100

    compute: create one pollster to rule them all

    Currently each compute pollster polls the backend for each
    metric of each instance while almost all backends can retrieve all stats
    at once (all except xenapi).

    This change reworks the compute pollsters to make one inspector call to
    retrieve data for an instance.

    The cache structure is simplified a lot by removing the useless intermediate
    object.

    We now cache the cpu stats, since putting the polling of an instance
    twice in polling.yaml is a non sense. We don't care if the second poll
    have a wrong timestamp due to the caching.

    Related-Bug: 1670948
    Change-Id: Id90f701cef261e165b63f0ba92a35cecfed659e7

Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Change abandoned on ceilometer (master)

Change abandoned by gordon chung (<email address hidden>) on branch: master
Review: https://review.openstack.org/443452
Reason: https://review.openstack.org/#/c/445976/ merged. please reopen if you think bug still exists.

gordon chung (chungg)
Changed in ceilometer:
status: In Progress → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.