On Wed, Aug 10, 2011 at 3:40 AM, Ulrich Eckhardt <ulrich.eckha...@dominolaser.com> wrote: > On Wednesday 10 August 2011, Sharma, Gaurav wrote: >> I have a task in hand to setup a master-slave svn repository configuration >> so that master will synchronize its repositories with slave. I searched a >> lot on web and found many links telling about how to set it up. Following >> those links I started using "svnsync" for this purpose. But some how after >> spending many days I am not able it to work as desired. > [...] >> I have two objectives for this setup- >> >> 1- End user should be able to perform both check-out and check-in to >> the slave repository. >> 2- Master should synchronize all its repository with slave > > > There are basically three steps involved with this: > 1. set up a secondary repository > This involves creating it and initialising it with svnsync. After this step, > you should be able to pull changes from the primary repository by invoking > svnsync locally. > 2. provide access to the secondary repository > This then allows you to push (as opposed to pull) changes from the primary to > the secondary repository. Further, it allows you to check out from the > secondary repository. You should disallow checkins and revprop changes except > for the dedicated svnsync user though, as documented. This seems to be set up > correctly already. > 3. connect the repositories > This involves creating a post-commit hook that svnsyncs the changes to the > secondary repository and one for the rev-prop changes, but you seem to have > these set up already. > 4. create a transparent write-proxy > This requires Apache (the rest could be dove with svnserve, too) and should > also be documented. I never made this step though but only the former three, > since the write-proxy wasn't my requirement.
Step 3 and 4 *are not enough*. The synchronization would have to occur at the "commit" stage, not the "post-commit", because "commit" has to be atomic with Subversion. Post-commit is nominally allowed to fail without stopping the commit operation. And the only support I've seen for that sort of thing is in Wandisco's "Multi-Site" tool. If you use the post-commit approach, as soon as people send distinct commits at the same time to the different repositories, and each of them completes before performing a post-commit operation, of if one repository is disconnected for a while for whatever network or downed server reason and gets a commit that is not successfully transmitted ot the other repository, then the other repository receives a local commit before it can perform an svnsync syccessfully, the same revision number will have different contents in each repository and you are *screwed* trying to match them ever again. This way lies madness.
