<DanC__>
"With Kiwix, you can easily set up your own copy of Wikipedia, Project Gutenberg, Stack Overflow, and many other free content sources. Perfect for your local network
<gharris>
Trying to get a decent backup with an automated backup solution. From the docs that I've seen, it seems that the MongoDB is the part of the backup that's unreliable. The Admin Docs say to snapshot the whole dir, which isn't practical. The dir is huge compared to the DB at this point, and the backup time to get a full backup means hours of downti
<gharris>
systemctl start sandstorm
<gharris>
tar -cf /root/sandstorm-mongo-snapshot.tar /opt/sandstorm/var/mongo
<gharris>
me. I created a cron job to stop Sandstorm, back up the DB locally, then start the DB. I realize I may not get EVERYTHING, but it should be pretty close. Here's what I've got: systemctl stop sandstorm
<gharris>
Again, I'm looking for a solid, reliable, backup that gets nearly everything with great uptime, rather than a perfect backup.
<gharris>
Does this look reasonable, in conjunction with a backup that will grab the whole system?
bodisiw has quit [Quit: This computer has gone to sleep]
bodisiw has joined #sandstorm
ocdtr_web has joined #sandstorm
<ocdtr_web>
gharris: Probably a good discussion for the discussion group. A. So it doesn't get lost in the IRC logs, and B. Because if you get something better than the current docs suggest, it'd be good to put in the docs later.
<gharris>
Where would I find said discussion group? I've dug around the website and must be missing where all the action is. Sorry.
<gharris>
Thanks TimMc! I'll post what I've got over there too.
<TimMc>
You had previously mentioned not wanting ZFS, right?
<gharris>
Perhaps. But yes, reformatting the server just to get ZFS, and the flakiness it seems to still be incurring, wasn't really something I was looking forwards to.
<TimMc>
Whether it's safe to back up grains while the server is running probably depends entirely on each grain's software.
<TimMc>
You might want to look at incremental backups of the grains, which you could do during the downtime.
<gharris>
?! That's an interesting idea. Is there an example on what would make a restore unstable?
<TimMc>
Perhaps an app has its own Mongo or other DB that doesn't handle live backups gracefully.
<gharris>
I've got a network backup solution that's doing incremental and full backups on a rotation. Incremental backups are a reasonable 10 min deal, but full takes a few hours at this point.
<ocdtr_web>
Wekan at the very least also uses Mongo. Any of the Meteor apps, most likely.
<gharris>
Hrm. So, some apps would attempt to run their own MongoDB separate from the primary MongoDB? That seems like a huge resource hog.
Lionel_Debroux has joined #sandstorm
<TimMc>
I wouldn't trust Mongo to be safely multi-tenant. :-)
<ocdtr_web>
Every app handles it's own data storage.
<ocdtr_web>
There's no central data. Sandstorm's Mongo is for running Sandstorm.
<ocdtr_web>
Grain storage may contain Mongo, MySQL, SQLite... whatever.
<gharris>
Ugh! So, what I've done actually creates more problems than it solved.
<TimMc>
Those incremental backups will still work, if you run them during the downtime!
<TimMc>
You just don't get clean full backups.
<gharris>
Hrm. There's an interesting idea. Run full backups, but run an incremental immediately afterwards that runs the cron to shut down the service.
<TimMc>
Oh, that might do it!
<TimMc>
If your backup solution is flexible enough to arrange that, a "cleanup phase" might work.
<gharris>
It will do Pre and Post processing, it's just a matter of setup at that point.
<gharris>
Thanks for the brainstorm. Off to see what I can break.
<TimMc>
^_^
ocdtr_web has quit [Quit: Page closed]
isd has joined #sandstorm
Telesight has joined #sandstorm
ogres has quit [Quit: Connection closed for inactivity]