-
Bug
-
Resolution: Unresolved
-
Normal
-
None
-
4.15.0
-
No
-
True
-
Description of problem:
We have a multi-arch CI cluster (build10) which runs 4.15.0-rc5. We are unable to make this cluster to production because the cluster is struggle to build any aarch64 Build. The Build will stuck at Init:0/X, which is extract-image-content. Then the node running this pod will report Node Saturation (>50 system load) and the Disk Pressure alert. The Disk Pressure alert is come and go. The aarch64 nodes (m6g.4xlg) have almost same resources as amd64 (m6a.4xlg) counter parts (64GB mem, 120GB disk, 16 core CPU). We believe there are some issue with extract-image-content binary on aarch64.
Version-Release number of selected component (if applicable):
4.15.0-rc5
How reproducible:
Create Build on aarch64 node
Steps to Reproduce:
1. Create a Build, set node selector to run the build on aarch64 nodes 2. 3.
Actual results:
1. The build pod stuck at Init:0/X 2. The node running the build reports NodeSaturation, and the extract-image-content process utilised all CPU resources 3. The node running the build reports Disk Pressure, and the container running extract-image-content used a lot of disk space
Expected results:
The build completes
Additional info:
Ref: https://redhat-internal.slack.com/archives/CK1AE4ZCK/p1708356005736639 Ref: https://redhat-internal.slack.com/archives/CKJR6200N/p1708442773431539
- is duplicated by
-
OCPNODE-2080 Unable to build images in aarch64 nodes
-
- Closed
-