Ad Widget

Collapse

Issues with VMWare Datastore history data since upgrade to V4.0

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • jacotec
    Junior Member
    • Oct 2018
    • 3

    #1

    Issues with VMWare Datastore history data since upgrade to V4.0

    Hi,

    I'm experiencing strange issues with my items:

    VMWare Hypervisor Datastore_xxx free%
    VMWare Hypervisor Datastore_xxx latency

    I'm using the ESXi template with 60d history and 730d trends (Housekeeping is done via database partitioning). That works pretty well for all other data, but the datastore data behavior is strange!

    On one host, the data for these two events is cleared every hour in minute 7. So, at 12:05 I see the graph only for the last 58 minutes and the graph history data is cleared at 12:07 ... so at 12:08 I only see one minute of data in the graph.

    For three other hosts the data of these two items is deleted once per day in the night, but every host at a different time.

    Another host has all its history available.

    All hosts are using exactly the same template. I'm fully puzzled why this happens and it's really annoying as the datastore latency is quit an important parameter for me.

    The logs do not show any hints. Housekeeping does not handle trends and history as this is done by SQL partitioning, and the unwanted graph deletion times do not match the housekeeper runs.

    Any ideas are highly appreciated!
  • jacotec
    Junior Member
    • Oct 2018
    • 3

    #2
    Just FYI: The issue is still present with Zabbix 4.01

    Comment

    • jacotec
      Junior Member
      • Oct 2018
      • 3

      #3
      Hi,

      I want to come back to this issue which is still unresolved (currently running V4.03). I found a relation of a full datastore history loss of a host - whenever the data is lost on any host, I see a related entry in the zabbix_server.log:

      Cannot get Datastore info: Timeout was reached.

      I normally have no issues getting any performance data from any hosts, no history data is lost - except from the datastore.

      I have raised the number of the VMWare collectors from 10 to 30 for a test and have increased the timeout for the VMWare collections from 10s to 30s - but that does not help. What really worries me: Even in case Zabbix can't get the performance value of the datastore of any host with a polling - it will not have this minute's values - but why does it in that case delete all historical data and trends of this datastore by just missing a query? I'm tending to treat this as a bug.

      EDIT: I've filed Bug ZBX-15417 according to this: https://support.zabbix.com/browse/ZBX-15417
      Last edited by jacotec; 07-01-2019, 18:18.

      Comment

      Working...