Skip to content

LendingLimit bias when removing nodes from the cluster #252

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
tardieu opened this issue Oct 8, 2024 · 0 comments · Fixed by #255
Closed

LendingLimit bias when removing nodes from the cluster #252

tardieu opened this issue Oct 8, 2024 · 0 comments · Fixed by #255
Assignees

Comments

@tardieu
Copy link
Member

tardieu commented Oct 8, 2024

If a node is removed from the cluster that was previously flagged as unhealthy, we keep subtracting the unhealthy node resources from the lending limit(s) on the slack quota queue forever.

We need to account for deleted nodes and properly prune the cached node information in the node health monitor.

For now, we can work around the issue by restarting the controller after removing an unhealthy node.

@dgrove-oss dgrove-oss self-assigned this Oct 14, 2024
dgrove-oss added a commit to dgrove-oss/appwrapper that referenced this issue Oct 15, 2024
1. Split node monitoring into two reconcilers, one to monitor Nodes
   and one to monitor and update the designated slack ClusterQueue.
2. Remove entries from in memory caches when a Node is deleted.
3. Watch slack cluster queue to be able to react to changes in
   nominalQuotas and adjust lendingLimits accordingly.

Fixes project-codeflare#252.
dgrove-oss added a commit to dgrove-oss/appwrapper that referenced this issue Oct 15, 2024
1. Split node monitoring into two reconcilers, one to monitor Nodes
   and one to monitor and update the designated slack ClusterQueue.
2. Remove entries from in memory caches when a Node is deleted.
3. Watch slack cluster queue to be able to react to changes in
   nominalQuotas and adjust lendingLimits accordingly.

Fixes project-codeflare#252.
dgrove-oss added a commit to dgrove-oss/appwrapper that referenced this issue Oct 15, 2024
1. Split node monitoring into two reconcilers, one to monitor Nodes
   and one to monitor and update the designated slack ClusterQueue.
2. Remove entries from in memory caches when a Node is deleted.
3. Watch slack cluster queue to be able to react to changes in
   nominalQuotas and adjust lendingLimits accordingly.

Fixes project-codeflare#252.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging a pull request may close this issue.

2 participants