weird error at PR initialization #1453
Comments
|
Need more information here to debug as the latest build works for me. |
|
This appeared only once. It has not happened again ever since. I'll reopen the issue if it reappears with some proper debug info. |
|
Hi folks, getting this issue on several PRs (intermittently) using the latest master (ca976d8). We have dozens of PRs a day, and it's not happening on all PRs. |
|
Is this consistently happening on only one repo? Or a variety of repos When this happens, is there a Github incident going on? You can check that online What is the typical number of files in the PRs that these happen on? Is there any chance there are network blips occurring with your setup? Debug logs for the whole command would be useful here. |
|
We are experiencing this periodically as well. For a separate issue, I've noticed that GitHub APIs seem to be eventually consistent, and this impacts services that consume webhooks and follow up with API calls. To be more specific, the other use-case I've seen is a review-event webhook that triggers a ListReviews call; the call response is missing the exact review triggering the webhook event. Retrying just a bit later returns the expected result. I wouldn't be surprised if something similar is happening here. Based on that thinking, it seems like adding retries with backoff on 404 would be a good solution. Does that sound reasonable? My colleague @nrmitchi is thinking of taking a pass at this on our local fork. |
|
I have experienced this again one more time, so based on other accounts here, I am reopening the issue. |
* Improve github pull request call retries Retry with fixed 1 second backoff up to 3 retries was added by #1131 to address #1019, but the issue continued to show up (#1453). Increase max attempts to 5 and use exponential backoff for a maximum total retry time of (2^n - n - 1) seconds, which is roughly 30 seconds for current max attempts n = 5. Also move the sleep to the top of the loop so that we never sleep without sending the request again on the last iteration. * Fix style with gofmt -s
* Improve github pull request call retries Retry with fixed 1 second backoff up to 3 retries was added by runatlantis#1131 to address runatlantis#1019, but the issue continued to show up (runatlantis#1453). Increase max attempts to 5 and use exponential backoff for a maximum total retry time of (2^n - n - 1) seconds, which is roughly 30 seconds for current max attempts n = 5. Also move the sleep to the top of the loop so that we never sleep without sending the request again on the last iteration. * Fix style with gofmt -s
Hey there,
I just switched to the latest build (8ca6e164749a4fa6509202124c775769fa17c103)and getting this error when:
The text was updated successfully, but these errors were encountered: