Add an error message to help debug issue when accessing secrets in CI #5441
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/5441
Note: Links to docs will display an error until the docs builds have been completed. ❗ 1 Active SEVsThere are 1 currently active SEVs. If your PR is affected, please view them below: ✅ No FailuresAs of commit 5830c88 with merge base 06c0fa3 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
facebook-github-bot
added
the
CLA Signed
|
@guangy10 has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
|
@huydhn Interesting that we already increased the timeout to 120min but the job still timed out after 60min: https://github.com/pytorch/executorch/actions/runs/10908226772/job/30282356192. |
Good catch, it turns out that there is another "hidden" timeout there when the AWS credential expires after 1 hour. I have the fix here pytorch/test-infra#5685 to set it correcly. |
As reported by @guangy10 in pytorch/executorch#5441 (comment), it turns out that the job timeout can be set to a higher value (120 minutes), but the AWS credential still expires after 1 hour https://github.com/aws-actions/configure-aws-credentials. The fix here is to set the role duration correctly. However, the max that it can go is 18000 seconds. I believe that's plenty of room already, but if a higher is needed, we will need to update the server side https://github.com/pytorch-labs/pytorch-gha-infra/blob/main/runners/gha_roles.tf#L1425 ### Testing The role duration is set correctly to 7200 seconds (2 hours) https://github.com/pytorch/test-infra/actions/runs/10916800466/job/30298898748#step:4:4
As reported by @guangy10 in pytorch/executorch#5441 (comment), it turns out that the job timeout can be set to a higher value (120 minutes), but the AWS credential still expires after 1 hour https://github.com/aws-actions/configure-aws-credentials. The fix here is to set the role duration correctly. However, the max that it can go is 18000 seconds. I believe that's plenty of room already, but if a higher is needed, we will need to update the server side https://github.com/pytorch-labs/pytorch-gha-infra/blob/main/runners/gha_roles.tf#L1425 ### Testing The role duration is set correctly to 7200 seconds (2 hours) https://github.com/pytorch/test-infra/actions/runs/10916800466/job/30298898748#step:4:4
As reported by @guangy10 in pytorch/executorch#5441 (comment), it turns out that the job timeout can be set to a higher value (120 minutes), but the AWS credential still expires after 1 hour https://github.com/aws-actions/configure-aws-credentials. The fix here is to set the role duration correctly. However, the max that it can go is 18000 seconds. I believe that's plenty of room already, but if a higher is needed, we will need to update the server side https://github.com/pytorch-labs/pytorch-gha-infra/blob/main/runners/gha_roles.tf#L1425 ### Testing The role duration is set correctly to 7200 seconds (2 hours) https://github.com/pytorch/test-infra/actions/runs/10916800466/job/30298898748#step:4:4
4 participants
What shows here (#5428 (comment)) seems to be a common issue especially for people who are exporting the Diff to GitHub from Phabricator. Added an error message when secret is empty.