Backup rsync fails due to index's "vanishing"

[TheDude]

I am using the Backup script from the Wiki and it does an rsync, but, it is only completing about 1/3 of the time.

The times it fails it gives errors about "file has vanished: "/var/opt/scalix/xx/indexes/49494949blahblah/_37j3.cfs"

I think the script was written pre-version 11 before the index service, is there a command I can use to pause this service so my rsync can complete? Also, I need the command to crank it back up when rsync is done!!

Thanks,

Dude

[mito]

Yes, the script was written before Scalix 11, and I'm not sure if the wiki has been updated since.

The script was written based on a thread in the forums, and in that thread someone said he modified it to work with Scalix 11 better.

Here's the post that says he fixed it and how it's fixed:
viewtopic.php?p=24473#24473

I am going live with my Scalix 11 this week sometime, and will definitely be going through the script myself. If it's not updated by then, I'll update it with my changes after they are fully tested.

[TheDude]

Thanks for that but I still need help.

That guy does not like rsync and his re-do specifically exludes it, I however, do want to use rsync I just need to know how to pause indexing for it to finish.

[mito]

I just need to know how to pause indexing for it to finish.

I don't have the code in front of me right now, but if I remember correctly, it stops all the scalix services before doing the rsync, so the indexing should already be stopped?

[TheDude]

Should is the key word :)

The way rysnc works is it compares two folders and changes the second to reflect all the changes from the first.

I would not get this error if things were not changing and happening on first folder (the live one) during the rsync process.

My guess is the index service is not turned off by the legacy command in the old script since my error makes me think the index service is doing stuff.

The Dude

[mito]

Well, I guess we need to get some more knowledgeable person than I to respond then...

The old script calls the init.d scalix service and tells it to stop... it'd be the same as typing

Code: Select all

/sbin/service scalix stop

at the command prompt... something that should stop all processes. Indexing does take a while to do... maybe it stops calling for new indexes, but lets the indexing that is currently running continue until they're finished, and thus sometimes they finish before rsync gets to them, and sometimes it doesn't?

That's a guess... but the only thing I can think of. Either that, or the init.d script didn't get updated by the upgrade?

[florian]

if you want to stop scalix completely, you will also have to stop the scalix-tomcat and scalix-postgres services. the indexer is a java app and runs within the tomcat app server.

another way of doing this would be to combine rsync with LVM snapshots to provide for consistent-state information to be replicated.

cheers,
Florian.