Details
-
Suggestion
-
Resolution: Unresolved
-
None
-
None
-
5
-
Description
"My understanding from the online KB is that a mirror will occasionally "phone home" to see if it needs to fetch, in the event the triggers were missed or couldn't be acted on. That appears to happen, but has some form of "stop trying" logic built in.
Due to our own issues, the mirror was out of sync, and had failed to correct enough times that numerous repos were in a "stop trying" state, with no obvious way to get them out of it.
Because we didn't get a reply, we eventually discovered that if a remote user just did a push to the primary, that caused the expected trigger, and subsequent pull, and the repos self-repaired.
None of that was obvious anywhere. Remote users weren't pushing, and their pulls were out of date. So it was in limbo until we did a bunch of pushes to kick things.
It's worth mentioning that NONE of this had anything to do with networking. The IT person who built the mirror provisioned the disk incorrectly, and we ran out of space on the mirror, preventing the pulls, and putting the repos into the stop trying state.
The primary should support a sync status for each repo and offer a "trigger a sync" option in the event things get stuck like mine did."