Ishwarya

Cadet
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
04-11-2021
11:46 AM
- 2,664 Views
Issue while moving to Infra node
Hi Everyone,
I’ve installed elastic search and trying to run it on infra node. Already specified the nodeselector but only one pod is running in the infra node. Remaining (2) are running on worker nodes onl.
Shared the log of the pod running on the infra node.
1. Taints applied
2. Nodeselector specified in openshift-logging in namespace too.
3. Labels are given on corresponding Infra nodes
Not sure where I’m missing. Any suggestions are welcome. Kindly help me out.
2 Replies


Flight Engineer
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
08-10-2021
08:29 PM
- 2,490 Views
Hey,
it kinda reads, like you only have one infra node. If that is the case, check if there are any pod anti-affinity rules configured, which might prevent the other two pods from being scheduled on the same infra nide.
Also, keep in mind, that your nodeselector configuration might only work during scheduling and not during execution, meaning a restart of the other two pods might schedule them correctly afterwards.
Lastly, check node resources, if there is enough available for the other two pods.
Those are the general considerations regarding pod placement.
it kinda reads, like you only have one infra node. If that is the case, check if there are any pod anti-affinity rules configured, which might prevent the other two pods from being scheduled on the same infra nide.
Also, keep in mind, that your nodeselector configuration might only work during scheduling and not during execution, meaning a restart of the other two pods might schedule them correctly afterwards.
Lastly, check node resources, if there is enough available for the other two pods.
Those are the general considerations regarding pod placement.
best regards,
Alexander
Alexander


Flight Engineer
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
08-10-2021
08:31 PM
- 2,489 Views
additionally to debug:
taint the worker in a way that the other two pods are not possible to start there and then check the logs of the deployment pods or the events to see the reasons the pods can't be placed anywhere.
taint the worker in a way that the other two pods are not possible to start there and then check the logs of the deployment pods or the events to see the reasons the pods can't be placed anywhere.
best regards,
Alexander
Alexander
Join the discussion
You must log in to join this conversation.