Add retry logic for image copying #63
Conversation
| if strings.Contains(err.Error(), "blob unknown to registry") { | ||
| log.Warn(fmt.Sprintf("encountered `blob unknown to registry error` for image %s", src)) | ||
| } | ||
| log.Info(fmt.Sprintf("attempt #%v failed, waiting %vs and then retrying", i, retryWait)) |
There was a problem hiding this comment.
We're retrying on any error, not just "blob unknown to registry". Is that by design? This means we'll retry if the registry is down, among other things. If it's an error that takes a timeout to reach, we'll hit 5 timeouts.
There was a problem hiding this comment.
I agree this isn't ideal but figured it was the easiest approach. I took this approach since I noticed the Docker CLI does the same and containers/image has a conversation to handle this specifically around the 429 HTTP code: containers/image#703.
They do an exponential backoff for 5 retries it looks like.
I would prefer to dig more into the specific errors this can return and make determinations on it but would need more time to do that properly.
There was a problem hiding this comment.
Added a TODO to come back to this
* Add retry logic for image copying * Remove unnecessary break * Change to 5 retries * Use proper format for prints * Return empty byte array * Add todo (cherry picked from commit b7c9bbd)
No description provided.