Conversation
|
Download PR build artifacts for linux_amd64.tar.gz, darwin_arm64.tar.gz, and windows_amd64.zip. 📦 Click here to get additional PR build artifactsArtifact URLs |
| } | ||
| return err | ||
| } | ||
| c.log.Tracef(" Sending %d batches took %.3fms...", len(batches), time.Since(start).Seconds()*1000) |
There was a problem hiding this comment.
[nit] This stops the clock before the split-and-resend loop below, so when a 413 makes us split batches, the extra round trips don't show up in any stage timing. That's one of the slow paths we'd want to see. Can we move the trace after the split loop, or add one for it?
| if errors.As(err, &werr) || errors.Is(err, internal.ErrSizeLimitReached) { | ||
| return err | ||
| } | ||
| i.Log.Tracef(" Writing to %q failed after %.3fms, trying other client...", client.url, time.Since(start).Seconds()*1000) |
There was a problem hiding this comment.
[nit] On the last client this still says "trying other client", but there's none left and we fall through to the final error. Could be misleading when reading a trace for a total failure.
Summary
This PR adds some more tracing information to debug slow writes more efficiently.
Checklist
Related issues