Bug 1319890

Summary: performance collection issues
Product: Red Hat CloudForms Management Engine Reporter: Felix Dewaleyne <fdewaley>
Component: PerformanceAssignee: Keenan Brock <kbrock>
Status: CLOSED CURRENTRELEASE QA Contact: Pradeep Kumar Surisetty <psuriset>
Severity: medium Docs Contact:
Priority: high    
Version: 5.5.0CC: agrare, arcsharm, benglish, cpelland, dajohnso, dmetzger, fdewaley, jfrey, jhardy, kbrock, obarenbo, simaishi
Target Milestone: GAKeywords: TestOnly, ZStream
Target Release: 5.7.0   
Hardware: All   
OS: All   
Whiteboard: perf
Fixed In Version: 5.7.0.0 Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of:
: 1324633 1346999 (view as bug list) Environment:
Last Closed: 2016-12-10 02:05:13 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Bug Depends On:    
Bug Blocks: 1324633, 1346999    

Description Felix Dewaleyne 2016-03-21 18:26:17 UTC 
Description of problem:
performance collection in the customer environment stops abruptly

Version-Release number of selected component (if applicable):
5.5.2


How reproducible:
customer environment only

Steps to Reproduce:
1. set up one appliance to do performance capture for one zone
2. let collection run 
3. access the utilization data of multiple vms

Actual results:
not all vms have performance data available

Expected results:
all performance data is being collected

Additional info:
collections are missing and it seems the holes they form cause the graphs not to be computed
Comment 12 Keenan Brock 2016-04-06 18:11:53 UTC 
I see one culprit:

[hourly] Rollup for VmVmware name: [T1KDB1], id: [12000000000715] for time: [2016-02-06T05:00:00Z]
Looks like that one failed

> note : what would be the impact here of the null values in the
> metric_rollups.intervals_in_rollup value,
I agree that this is odd.

> and why do we have rows that show no vendor or name?
If a vm is no longer available, it will show up as nil. Possibly this VM went away?
Comment 32 Keenan Brock 2016-07-13 15:32:59 UTC 
I believe this is due to either cap&u timer running too long/timing out, or cap&u running too long/timing out.

We are aware of cap&u timer running too long and have a number of fixes in 5.6/5.7 to speed it up.
We have cap&u performance tweaks, but unsure the speed difference
Comment 33 Keenan Brock 2016-09-22 17:04:08 UTC 
This is part of the fixup to perf_capture_timer. It is part of Darga and later
Comment 34 Satoe Imaishi 2016-09-27 13:47:57 UTC 
Keenan, can you add PR link?
Comment 35 Keenan Brock 2016-10-26 19:37:05 UTC 
https://github.com/ManageIQ/manageiq/pull/9766