Page MenuHomePhabricator

Diffusion update frequency gets large on update error
Closed, ResolvedPublic

Description

Hi,

We recently upgrade our Phabricator instance (https://reviews.llvm.org) from 9eda21451a4455ba0c11bf675b334c028e29b0c9 (Jun 30, 2015) to 27006fedccc2f487dcaf7177f6d127ccd0e6df7d (Fri Sep 23, 2016) .

In the new version, we are experiencing issues with importing commits from our svn repository:

Our svn server has always been unstable, which causes update errors from time to time. But this can often be fixed by retrying update once or twice. In the older version, the retry (scheduled by Phabricator iteself) works fine. However, in the newer version, the update frequency becomes a few days when there is an update error, and I didn't see any leased or queued task in the /daemon page. So I guess Phabricator does not retry in this case (and it would not update unless manually scheduled).

Is this behavior expected? Is there a way to force phabricator to retry after a certain amount of time?

Thanks!

Event Timeline

This is a screenshot of the status page. This can usually be fixed by manually rescheduling an update by either clicking "Update now" or calling "./bin/repository update" in command line.

Screenshot from 2016-10-12 14:13:06.png (747×1 px, 92 KB)

epriestley claimed this task.
epriestley added a subscriber: epriestley.

Update past D16623 (Sep 28). See also T11705.

File a new report if you're still seeing issues after updating.

Thanks for the quick response!