| cyberanger | chris4585: I bit the bullet and did it from scratch with debootstrap | 01:54 |
|---|---|---|
| vychune | o/ | 02:05 |
| chris4585 | cyberanger, yeah? | 04:04 |
| chris4585 | I like it from debootstrap better | 04:04 |
| wrst | morning cyberanger | 13:08 |
| kd4zay | morning all | 16:26 |
| wrst | kd4zay: morning | 16:46 |
| vychune | o/ | 16:52 |
| === vychune is now known as vychune_bath | ||
| === vychune_bath is now known as vychune | ||
| === vychune is now known as vychune_afk | ||
| kd4zay | cyberanger: how do you guys handle your linux backups? I looking for a way to take snapshots of the system while mounted. deshavu! | 18:24 |
| Juzzy | lvm can | 18:37 |
| Juzzy | how many linux boxes? | 18:38 |
| chris4585 | kd4zay, my favorite method of backing up http://paste.ubuntu.com/609655/ | 19:34 |
| Xpistos | what are you backing up with that> | 20:00 |
| Juzzy | heh | 20:01 |
| Juzzy | everything apperently | 20:01 |
| kd4zay | I have about 8 linux boxes | 20:30 |
| kd4zay | chris4585: so if i was to need to do a full metal restore i would need to: manually create the partitions ,restore the backup, chroot from the bootup media mount dev and proc then run grub to reinstall the mbr ? | 20:35 |
| kd4zay | juzzy: lvm was not install on these server originally | 20:41 |
| cyberanger | hey wrst | 21:05 |
| cyberanger | kd4zay: we've not done frozen snapshots, becuase the system was mounted | 21:06 |
| cyberanger | in the past, we've umounted /boot | 21:06 |
| cyberanger | then rsync'd it | 21:06 |
| cyberanger | then do the same, but without unmounting it | 21:06 |
| cyberanger | ignoring a directory here and there | 21:07 |
| kd4zay | i see | 21:07 |
| cyberanger | not a full backup, exactly, but the directories we ignored are the stuff that is created on the fly | 21:07 |
| cyberanger | /proc for example | 21:08 |
| kd4zay | dev proc | 21:08 |
| cyberanger | /dev too, yep | 21:08 |
| kd4zay | sys | 21:08 |
| cyberanger | um, not sure on that one | 21:09 |
| kd4zay | though rsync would only keep one copy synced correct? | 21:09 |
| cyberanger | there was little that one machine did, /mirror and /etc were about all that was tweaked from stock | 21:10 |
| cyberanger | apache and ssh added | 21:10 |
| cyberanger | that is true, if you keep using the same destnation or don't back that up | 21:10 |
| cyberanger | once rsync got the data off the mirror, we could tarball it, we did tarball /boot and /etc for sure | 21:11 |
| cyberanger | /home was pretty useless in this case | 21:11 |
| cyberanger | thing about the mirror, only the most recent data was worth anything | 21:12 |
| kd4zay | i have a hylafax server / a mailarchiva box / snort / and a few others that im trying to come up with a way to manage all the backups for all or anyone but be able to do baremetal recovery | 21:13 |
| cyberanger | so "rsync -avPe ssh exampleuser@exampledomain:/mirror /backup/mirror1 && cp /backup/mirror1 /backup/mirror2" | 21:13 |
| cyberanger | then we allways have a fairly recent bit of /mirror, and if that had an issue, it'd be on one of two sets, if both sets failed, a bigger issue just happened | 21:15 |
| cyberanger | and since there are no donations, no backers, besides my wallet | 21:15 |
| cyberanger | this isn't corprate, that's what I gotta settle for | 21:15 |
| kd4zay | sure that could work with a automated script that could create x amount backups or just a daily cron for each day of the week | 21:16 |
| cyberanger | yep, and toss in something checking sha256sums | 21:16 |
| cyberanger | make sure the backups are good | 21:16 |
| cyberanger | now, there is other options, but for a true snapshot, you've gotta stop read and write access | 21:17 |
| kd4zay | the only issue i see would be any database being open of course you can just stop mysql before the backups | 21:18 |
| cyberanger | yeah, and anything reading the database has downtime too | 21:19 |
| kd4zay | so if i was doing a bare metal restore , i would need to create the partitions, restore the backup and use grub to install the mbr back ? | 21:23 |
| kd4zay | cause really thats what i am looking for is a way to recover from a full failure at any point with a week | 21:24 |
| cyberanger | yeah, that would work | 21:27 |
| Juzzy | for big deployments I use arkeia (commercial) b/c it has point in time, etc | 21:27 |
| cyberanger | xfs has some nifty utilities that might be better | 21:27 |
| Juzzy | its cheap, and has wicked nice .deb support | 21:27 |
| cyberanger | for going from nearest backup to new box | 21:28 |
| cyberanger | I've gotten used to doing by hand in under an hour (less with the backup box already setup from one backup, then it's just an update) | 21:29 |
| kd4zay | the company here uses symantec back exec system recover which take snapshot through out the day, very expense! now i would not need that for the nixs but I will need to come up with a process | 21:36 |
| kd4zay | cyberanger: your solution seem like a valid one | 21:37 |
| kd4zay | arkeia has a free linux version ! | 21:38 |
| kd4zay | but only supports 2 server 8p | 21:45 |
| Juzzy | hehe | 21:54 |
| Juzzy | there's always bakula | 21:55 |
| kd4zay | bakula ? i havent looked at that yet | 22:23 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!