btrfs raid56 hard disk broken

[stormwatch]

Hi. I was a happy TT-RSS user running pgsql in /var in its own partition. Then I tried to move the whole root filesystem to a BTRFS raid5 array, paying attention to the right mount options like nocow and following the docs at pgsql homepage regarding heavy i/O.. Nevertheless this setup only lasted a week or so before one of the disks in the array (three SATA drives, Black WD) failed miserably. Like in SMART self-assesment test telling me to dispose it.

Was this just a case of bad luck? Does anyone ever managed to run TT-RSS under BTRFS raid56? If so, what should I pay attention to?

The update daemon (v2) was running continuosly with 200 feeds aproximately for just a single user: Me

Sorry if I am missing something, It is very late and I drank more than I should. Writing this message for the second time after "accidentally" pressing ctrl+w thinking this was emacs.

Cheerio!

[fox]

yes it was tt-rss that broke your hard disk *facepalm*

[cy8aer]

@stormwatch Read https://btrfs.wiki.kernel.org/index.php/RAID56 and make up your mind for the risks.

[miceliux]

btrfs raid56 eats your data. Don't use it at all.

[sleeper_service]

zfs wouldn't eat your data.... just sayin.

also, moved this crap to 'unsupported platforms' since we don't have a 'I took careful aim and put a slug right through the center of my foot, go me" area, though, maybe, comedy bistro...

[etu]

Dude, you... really used btrfs raid5/6 and not like mdadm-raid or LVM or something in the bottom? I wouldn't do that... just yet... if ever...

[virgo]

Dude, you... really used btrfs raid5/6 and not like mdadm-raid or LVM or something in the bottom? I wouldn't do that... just yet... if ever...

Just in case original poster is confused - even then btrfs would not kill actual hard drive. It was going to fail anyway.

[fox]

luckily the drive failed before btrfs did

[stormwatch]

Oh! I'd never blame tt-rss, nor BTRFS, neither pgsql btw... Well, maybe postgre, just kidding.

[sleeper_service]

whereas, you KNOW it's the fault of mysql.

[stormwatch]

I was going to rant about how can there be is no emoji for *facepalm* but it is was recently added to unicode http://emojipedia.org/face-palm/

Besides what is outlined in the FAQ, would you consider good practice to use a dedicated (maybe solid state) drive just for postgresql or /var? I'd also like to run other services like subsonic, ocassionally KODI on the same PC with 16GB RAM.

In the meantime I am reading the articles on http://www.baarf.dk/ They make a good point about disdaining RAID5.

[fox]

unless you need some heavy iops for some reason I would not bother especially for ttrss and xbmc