Reiserfs no more...

[hokatichenci]

Well, I've decided to bite the bullet and switch from reiserfs to XFS. I needed full uptime from it, and it croaked horribly this morning. So I've spent my morning so far trying to get LiveCDs to work, and eventually discovered that there was enough life left in reiser to copy over critical files (/bin, /dev, /usr, /opt etc) so that I could get my xfs working as my main partition. Now I'm praying fsck works well enough so that I can copy all the data back and switch it to XFS.

I should have known reiser was a bad idea when it took a minute to mount!

 

[Bones]

Reiserfs hasn't bombed on me in at least four years; the current code seems stable to me. What were the circumstances that caused the failure? Anyway, I've heard good things about xfs, but haven't tried it out myself. Let us know how your recovery fares.

 

[hokatichenci]

I'm running AMD64, doing a lot of dvd ripping/copying (I've got a large collection, I've been doing batch rips for a while). I was doing some ripping/copying at the same time and somehow reiser borked. I'm not sure how it happened, since the entire system croaked (only things in memory ran, and I couldn't start anything new or fully quit anything), but it definitely croaked hardcore. A fsck.reiserfs --check revealed critical errors in the tree (something about multiple files having the same numbers, inode#'s i assume). I'm still copying stuff over because I don't want to risk rebuilding the internal tree and messing something up in the first place.

Next step is to buy a battery module for my raid card to go with my UPS incase of complete and total systems failure. Then my XFS and raid write-back should be safe enough for me to sleep happily.

 

[hokatichenci]

Update-

I was able to save all the data before running a fsck on it, the fsck took about ~ 6 hours. It mounts cleanly now but since all the data got copied off it its fairly useless. I'm going to run a few days at least without it to make sure that I got everything important of fit and then move to XFS. Let this be a warning to people with large partitions and want to run reiser... since my feeling is that was the cause of the problem.

 

[FastCorner]

I was *this* close to using ReiserFS for my RAID5 array, but I backed out used EXT3 (with data+metadata journalling) instead.

What do you mean when you said it took over a minute to mount? According to docs and my own experience, it takes more than that to mount large file systems because it tries to cache bitmaps and stuff to RAM.

 

[Dew]

I was *this* close to using ReiserFS for my RAID5 array, but I backed out used EXT3 (with data+metadata journalling) instead.

What do you mean when you said it took over a minute to mount? According to docs and my own experience, it takes more than that to mount large file systems because it tries to cache bitmaps and stuff to RAM.

This is what I love about XFS, mounts take less than one second. That's on a 1.2TB array.

 

[tdg]

I ran ReiserFS on a workstation several years back, when it was still well known to be buggy, sure enough the filesystem took a dump, two times. Ever since I just stick with EXT3 or UFS, though I've heard good things about XFS.

 

[hokatichenci]

What do you mean when you said it took over a minute to mount? According to docs and my own experience, it takes more than that to mount large file systems because it tries to cache bitmaps and stuff to RAM.

It took a minute to mount, not sure what it's doing, but cpu usage jumps around like 4-10%, ram usage is pretty much the same, and i've got 2gb of system ram. There might be a fix out there but I was looking for an out of the box solution.

 

[FastCorner]

It took a minute to mount, not sure what it's doing, but cpu usage jumps around like 4-10%, ram usage is pretty much the same, and i've got 2gb of system ram. There might be a fix out there but I was looking for an out of the box solution.

There's a recent patch for that, but I'm not sure it's made it its way into a released kernel yet.