[ZBX-14677] wmvare monitoring (datastore,NIC,performace counter...) failed after 3.4.8 Created: 2018 Aug 02  Updated: 2018 Aug 28  Resolved: 2018 Aug 24

Status: Closed
Project: ZABBIX BUGS AND ISSUES
Component/s: Proxy (P), Server (S)
Affects Version/s: 3.4.11, 3.4.12
Fix Version/s: 3.0.21rc1, 3.4.13rc1, 4.0.0beta1, 4.0 (plan)

Type: Problem report Priority: Critical
Reporter: Alexander Kuznetsov Assignee: Andris Zeila
Resolution: Fixed Votes: 0
Labels: vmware
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified
Environment:

Docker container. Ubuntu image16.04


Attachments: Zip Archive 3.4.12.vmware.txt.zip     PNG File Performance counter data is not available..png     PNG File VMware Cache usage.png     File vmware-instance-30.patch     File vmware-instance-34.patch     PNG File vmware.hv.perfcounter.png     Zip Archive vpxd-8049.log.zip    
Issue Links:
Causes
causes ZBX-14765 vmware incorrect memory release of 'h... Closed
causes ZBX-14775 Get vpxd.stats.maxquerymetrics parame... Closed
caused by ZBX-12990 VMware Datastore usage mismatch Closed
Duplicate
duplicates ZBX-14680 Monitoring itens Datastore VMware wit... Closed
is duplicated by ZBX-14545 Issue on VMware retrieve for 1 of 2 c... Closed
Team: Team A
Sprint: Sprint 39, Sprint 40, Sprint 41
Story Points: 1

 Description   

We use proxy for monitoring vmware

After update zabbix server and proxy to 3.4.12 (from 3.4.8) some stock item stoped collect with error "Performance counter data is not available.".

Downgrade proxy to 3.4.8 fixed it.

Steps to reproduce:

  1. Update\Install zabbix server or proxy to 3.4.11(12)
  2. Setup vmware (VMware vCenter Server 6.0.0 build-4541947) monitoring by stock templates throught server or proxy.
  3. Check datastore`s\NIC data on ESX or create item to collect perfcounter from vcenter. Ex: vmware.hv.perfcounter[{$URL},{HOST.HOST},"cpu/usage[average]"]

 ** 



 Comments   
Comment by Vladislavs Sokurenko [ 2018 Aug 02 ]

This could be regression that is caused by ZBX-12990

Comment by Alexander Kuznetsov [ 2018 Aug 02 ]

I think - not. Because on older version zabbix server\proxy all work fine. And it also applies to performance counters for CPU or NICs

Comment by Alexey Pustovalov [ 2018 Aug 02 ]

This one ZBX-14548 ?

vso It is reported that is reproducible in both 3.4.11, 3.4.12 so it's not fixed and not caused by ZBX-14548

Comment by Alexander Kuznetsov [ 2018 Aug 02 ]

Looks same (update + Vmware + datastore) but detail - others.
In ZBX-14548 some items working (latency) in this case - not (Also performance counters NIC \ CPU \ Memory do not work). I'm not sure that the reason is the same, although it seems.

Comment by Michael Veksler [ 2018 Aug 07 ]

Logging of QueryPerf request has been added. Increased the log level for the QueryPerf response and the list's of available counters. Improvement of perfCounter instance name processing.
Please, apply the patch vmware-instance-34.patch from the attachments and provide additional information from zabbix_server.log. It would be nice if you could catch the list of available counters in log file, that would be rotate relatively fast. Server with the patch can run with "DebugLevel=3" option.

Comment by Alexander Kuznetsov [ 2018 Aug 08 ]

Which version to apply the patch? For the current 3.4.8 or 3.4.12

MVekslers patch for 3.4.12

Comment by Alexander Kuznetsov [ 2018 Aug 08 ]

3.4.12.vmware.txt.zip  - log from patched proxy.

 

Most likely the problem is described here: <soapenv: Fault> <faultcode> ServerFaultCode </ faultcode> <faultstring> This operation is restricted by the administrator - & apos; vpxd.stats.maxQueryMetrics & apos ;. Contact your system administrator. </ Faultstring> <detail> <RuntimeFaultFault xmlns = "urn: vim25" xsi: type = "RuntimeFault"> </ RuntimeFaultFault> </ detail> </ soapenv: Fault>
MVekslers As a temporary workaround, you can try to use vmware KB2107096

In addition, you can open the vpxd log file and look for the message that displays the number of metrics that exceed the limit.

What value will be display in your vpxd.log file ?

kuzmi4

From log

2018-08-09T10:32:43.251+03:00 error vpxd[07104] [Originator@6876 sub=MoPerfManager opID=63026c51] The query size of 486 metrics exceeded the vpxd.stats.maxQueryMetrics limit of 64 metrics. Dropping.

Added log file: vpxd-8049.log.zip

MVekslers An interesting point: the parameter config.vpxd.stats.maxQueryMetrics limit the count of perfMetric for Datastore only and does not limit perfMetric count for VirtualMachine and HostSystem. In your case: 162(DS) x 3(perfCounter) = 486

Comment by Andris Zeila [ 2018 Aug 21 ]

Released in:

  • pre-3.0.21rc1 r83961
  • pre-3.4.13rc1 r83963
  • pre-4.0.0alpha10 r83965
Generated at Fri Mar 29 12:12:57 EET 2024 using Jira 9.12.4#9120004-sha1:625303b708afdb767e17cb2838290c41888e9ff0.