Taking a look at the source I see it’s not so simple to restart after calling stop().
Perhaps two new methods could be added to ContentNode, pauseSync() and resumeSync() that simply acquire and release the sync lock? I realize this doesn’t stop all possible index update because the local node could do its own writes but we could be aware of that in our implementation and find other ways of avoiding writes to the JCR on this particular node while doing the index backup. What do you think? I could submit a PR. Tom -- Tom Duffey Crescendo Collective C: +1 414-915-3915 On May 5, 2025 at 6:12:02 PM, Tom Duffey <tduf...@crescendocollective.com> wrote: > Hi All, > > I recently posted to the user mailing list inquiring about ways to do a > hot backup of the index. Seems like this has been asked a few times with no > real solution so I wanted to run another potentially dumb idea by those of > you on the dev list: Could we add the ability to “pause” a cluster node to > make it temporarily stop syncing? This way the instance could stay live, > backup taken and then resume sync without worrying about any in-between > index updates. > > Maybe as simple as calling stop() on the node, doing the backup and then > start()ing it up again? > > Is this dumb? > > Tom > > -- > Tom Duffey > Crescendo Collective > C: +1 414-915-3915 >
