A task was cancelled and other errors

[kenkendk]

@ts678

Personally, I wish there had been over 9 days (as little as 2 due to notifier) before Stable.

True, but this had been in Canary for a while, so I assumed it was working as expected.

@handyguy

I changed to --read-write-timeout=5m and that seems to have cleared the problem. Is there a disposition on whether this is a bug or a feature, and if this will be addressed in a later release?

We have removed the timeout in the current canary. I am thinking of making a new beta/stable release without the timeout. We are working to add back the timeout, but in a much more controlled way.

@srudin

No offense - but if I need to wait and check feedback for some months first before installing a stable release to make sure it is really stable than the basic concept of the different channels is useless imo.

I agree. That is not the intention for the stable release channel. It was tested in the canary channel without being flagged. The problem here is that the timeouts seem to trigger mostly for larger backups, and apparently we did not hit enough large backups during canary testing.

@srudin

The error message then changed from “A task was cancelled” to “Unexpected EOF or 0 bytes from transport stream” but the backup showed a failure or an error nonetheless.

I think this is caused by having different points in the response pipeline where the the timeout occurs. If the timeout occurs before the server sends data, you will likely get “A task was cancelled” (terrible error message btw), but if you are uploading/downloading data when the timeout occurs, you can get “Unexpected EOF”.