You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When a job is cancelled (in our case because a new commit was added to the PR, triggering another build), the temp files generated by the job are not being cleaned up, causing our agent to fail.
Checkmarx's behaviour of zipping up the .git folders means that the zip generated by one of our codebases is very large (~3GB) so our /tmp/ space has to be enormous to handle this. As we're running inside containers, having persistent volumes or large temporary volumes is very inefficient for the cluster. Our widespread usage of Checkmarx within the company also means that this problem scales out very quickly.
Expected behavior
When a job terminates for any reason (success, failure, cancelled etc) the temp files generated by the task should always be processed.
Actual behavior
The files remain in place, causing our self-hosted ADO agents to run out of disk space and error.
Steps to reproduce
Create/select a very large SCM repo
Run a Checkmarx scan on an agent with a small /tmp reserve
After the job creates the zip file, cancel the job
Repeat as necessary until the agent fails.
Environment
Checkmarx AST@2
Ubuntu 20.04 LTS
Additional comments
Add any other context about the problem here.
Logs
Screenshot of the files left in the temp folder. The top is from a cancelled job, the second from a job that succeeded
2024/04/17 11:14:34 Scan status: Running
##[error]The Operation will be canceled. The next steps may not contain expected logs.
##[error]The operation was canceled.
Finishing: CheckmarxAST
##[error]Failed to create CoreCLR, HRESULT: 0x80004005
,##[error]We stopped hearing from agent ci-ado-agent-dev-43cee. Verify the agent machine is running and has a healthy network connection. Anything that terminates an agent process, starves it for CPU, or blocks its network access can cause this error. For more information, see: https://go.microsoft.com/fwlink/?linkid=846610
The text was updated successfully, but these errors were encountered:
Describe the bug
When a job is cancelled (in our case because a new commit was added to the PR, triggering another build), the temp files generated by the job are not being cleaned up, causing our agent to fail.
Checkmarx's behaviour of zipping up the .git folders means that the zip generated by one of our codebases is very large (~3GB) so our /tmp/ space has to be enormous to handle this. As we're running inside containers, having persistent volumes or large temporary volumes is very inefficient for the cluster. Our widespread usage of Checkmarx within the company also means that this problem scales out very quickly.
Expected behavior
When a job terminates for any reason (success, failure, cancelled etc) the temp files generated by the task should always be processed.
Actual behavior
The files remain in place, causing our self-hosted ADO agents to run out of disk space and error.
Steps to reproduce
Environment
Additional comments
Add any other context about the problem here.
Logs
Screenshot of the files left in the temp folder. The top is from a cancelled job, the second from a job that succeeded
The text was updated successfully, but these errors were encountered: