| Issue 482: | replicate doesn't retry on failure | |
| 4 people starred this issue and may be notified of changes. | Back to list |
What steps will reproduce the problem? 1. Configure replication to a remote system. 2. Turn off the remote system. 3. Push a change. 4. Turn on the remote system. It never gets the change. Upon getting a replication failure to a URL, Gerrit should block all future replication to that URL and use a background timer to continuously retry with a reasonable delay between attempts, until the remote comes back and starts to answer us again.
Aug 20, 2010
#1
lincoln....@sonyericsson.com
Aug 20, 2010
Nope, nobody is working on it. I'll assign Ulrik as the owner for now to mark it as "maybe being worked on". :-)
Owner:
ulrik.sj...@sonyericsson.com
Aug 23, 2010
Hello Shawn. About the fourth step, do you mean it never gets the change only if a full sync is not performed? Thanks.
Aug 24, 2010
In the 4th step, the remote system doesn't get the change because Gerrit didn't restart replication. If we added another step: 5. Restart Gerrit Code Review (or run ssh gerrit replicate --all) The remote system would be correctly updated. The bug here is we don't want to restart Gerrit, or do a manual SSH invocation to make the remote system become current again. We want Gerrit to realize it failed to connect to the remote system, and reschedule the replication job to try again in a few minutes (or some other reasonable delay).
Aug 24, 2010
Ok, Shawn. Thanks a lot for clarifying it to me.
Oct 29, 2010
Fixed in If46862787a4f9a60d555e96634e25c96efa3332e
Status:
Submitted
Labels: FixedIn-2.1.6
Dec 15, 2010
(No comment was entered for this change.)
Status:
Released
|
|
| ► Sign in to add a comment |