Conversation
nturinski
left a comment
There was a problem hiding this comment.
Is there a way for us to parse the error that's coming from the logs? I remember that there was a specific error about the url being incorrect that Anthony was running into.
I'm think that we may want to check for that error specifically because otherwise the retry logic might be annoying if it's failing for totally different reasonsl
Yeah that's a good idea. I know we talked briefly about seeing if we could parse errors from the logs for other reasons as well. Would it be worth it for us to reach out to Julius' team to get a better understanding about how the log messages are formatted under different error conditions? |
|
Have we done a test run on how long it takes to retry 3 times? while retrying, do we show any UI indications? |
I believe Megan is showing some UI indications for each attempt in the activity log. I've tried testing it with a bad URL a few times and it seems to complete all retries in around or a little bit less than 10 seconds. We may want to have Anthony test it later to see if we need to adjust the number of retries.
I was looking into this issue with Megan, and I think we might not even need to parse the logs after all. The error seems to throw specifically when the url is incorrect/not ready yet. If, say, the run fails due to something else like the Dockerfile, it will not throw right away in the run step because the run merely starts there, and until we poll its status again during |
| ext.outputChannel.appendLog(message); | ||
|
|
||
| context.run = await context.client.registries.beginScheduleRunAndWait(context.resourceGroupName, context.registryName, runRequest); | ||
| runRequest.sourceLocation = 'a' |
There was a problem hiding this comment.
That is an oopsie by me. I was using this to make the run fail.


Fixes #562.
I arbitrarily chose 3 retries since that should cover the timing issue and building the image takes a bit of time. Let me know if anyone disagrees with that choice.
I also tried a couple of times to test the retries but I could only get it to work on the first try. Guess our code is just too good 🤪