Note: This bug is displayed in read-only format because the product is no longer active in Red Hat Bugzilla.

Bug 1626362

Summary: Metrics network graph empty intermittently
Product: OpenShift Container Platform Reporter: Min Woo Park <mpark>
Component: HawkularAssignee: Ruben Vargas Palma <rvargasp>
Status: CLOSED WORKSFORME QA Contact: Junqi Zhao <juzhao>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: 3.4.1CC: aos-bugs, mpark
Target Milestone: ---   
Target Release: 3.4.z   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2019-04-23 18:16:51 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:
Embargoed:
Attachments:
Description Flags
pod metrics graph screen shot 1
none
pod metrics graph screen shot 2
none
heapster pod log
none
hawkular-metrics pod log
none
hawkular-cassandra pod log none

Description Min Woo Park 2018-09-07 07:28:45 UTC
Created attachment 1481502 [details]
pod metrics graph screen shot 1

Description of problem:

When the customer monitor from pod metrics, they could see emtpy graph only for network intermittently. (attached screenshot)
CPU and Memory graph is no problem at the same time.

Does this mean Pod network got any issue at that time?
How can we troubleshooting for this issue?

Attached hawkular-metrics, hawkular-cassandra, heapster pod log

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


How reproducible:


Steps to Reproduce:
1.
2.
3.

Actual results:


Expected results:


Additional info:

Comment 1 Min Woo Park 2018-09-07 07:29:22 UTC
Created attachment 1481503 [details]
pod metrics graph screen shot 2

Comment 2 Min Woo Park 2018-09-07 07:29:44 UTC
Created attachment 1481504 [details]
heapster pod log

Comment 3 Min Woo Park 2018-09-07 07:30:06 UTC
Created attachment 1481505 [details]
hawkular-metrics pod log

Comment 4 Min Woo Park 2018-09-07 07:30:28 UTC
Created attachment 1481506 [details]
hawkular-cassandra pod log

Comment 5 John Sanda 2018-09-10 22:14:36 UTC
Gaps in the graphs means that there are missing data points. Has the customer looked at other graphs for other pods/applications to see if there are any gaps?

Unless there is an issue on the heapster side, I doubt the issue is specific to network metrics.

The heapster log has very little info in it, and there are no errors in the hawkular-metrics log. The cassandra log reports lots of ParNew GCs. ParNew is a stop-the-world garbage collection which means all application threads are paused while it runs. Excessive GC could cause Cassandra to fall behind with and then drop some requests.

Can we get the output of:

$ oc -n openshift-infra exec <cassandra pod> nodetool tpstats