Reiserfs + Loadsheeding = :(

Kasyx

Expert Member
Joined
Jun 6, 2006
Messages
2,565
Reaction score
1
Location
127.0.0.1
Just thought I would make mention of this to anyone setting up any new machines for themselves or for clients, I do not suggest using ReiserFS. This is not because I have anything against it, I think it is an excellent filesystem. However, if one does not have the machine attached to a UPS and/or backup generator which will gaurantee it non-fluctuating power, I suggest going for ext3 or even XFS. The reason being that if one is unlucky enough to get loadshed whilst ReiserFS is syncing, failarity will ensue:

Code:
ReiserFS: md0: warning: vs-5150: search_by_key: invalid format found in block 1966228. Fsck?
ReiserFS: md0: warning: vs-13070: reiserfs_read_locked_inode: i/o failure occurred trying to find stat data of [515 104654 0x0 SD]
ReiserFS: warning: is_leaf: free space seems wrong: level=1, nr_items=39, free_space=65388 rdkey
ReiserFS: md0: warning: vs-5150: search_by_key: invalid format found in block 1966228. Fsck?
ReiserFS: md0: warning: vs-13070: reiserfs_read_locked_inode: i/o failure occurred trying to find stat data of [515 104654 0x0 SD]
ReiserFS: warning: is_leaf: free space seems wrong: level=1, nr_items=39, free_space=65388 rdkey
ReiserFS: md0: warning: vs-5150: search_by_key: invalid format found in block 1966228. Fsck?

Fsck indeed.

My suggestion, thus, from personal experience, is that simply due to this "power crisis" (read as: fsck up), running reiserfs on machines that are liable to go off with the power is pretty much a bad idea.
 
my reiserfs F* install has been load shed many a time without any problems for me. In fact the only time I have ever had a partition fail was an ext3 partition soon after power went down. make of that what you will.
 
I thought reiserfs is a journalized fs and therefore immune to such things.:confused:
 
Just thought I would make mention of this to anyone setting up any new machines for themselves or for clients, I do not suggest using ReiserFS. This is not because I have anything against it, I think it is an excellent filesystem. However, if one does not have the machine attached to a UPS and/or backup generator which will gaurantee it non-fluctuating power, I suggest going for ext3 or even XFS. The reason being that if one is unlucky enough to get loadshed whilst ReiserFS is syncing, failarity will ensue:

Code:
ReiserFS: md0: warning: vs-5150: search_by_key: invalid format found in block 1966228. Fsck?
ReiserFS: md0: warning: vs-13070: reiserfs_read_locked_inode: i/o failure occurred trying to find stat data of [515 104654 0x0 SD]
ReiserFS: warning: is_leaf: free space seems wrong: level=1, nr_items=39, free_space=65388 rdkey
ReiserFS: md0: warning: vs-5150: search_by_key: invalid format found in block 1966228. Fsck?
ReiserFS: md0: warning: vs-13070: reiserfs_read_locked_inode: i/o failure occurred trying to find stat data of [515 104654 0x0 SD]
ReiserFS: warning: is_leaf: free space seems wrong: level=1, nr_items=39, free_space=65388 rdkey
ReiserFS: md0: warning: vs-5150: search_by_key: invalid format found in block 1966228. Fsck?

Fsck indeed.

My suggestion, thus, from personal experience, is that simply due to this "power crisis" (read as: fsck up), running reiserfs on machines that are liable to go off with the power is pretty much a bad idea.

I only use ReiserFS on many boxes I've only had to do a tree rebuild twice, (both bad shutdowns). XFS only beats Reiser with large files.

Code:
reiserfsck --rebuild-tree -S /dev/xxxx

I thought reiserfs is a journalized fs and therefore immune to such things.:confused:

Whether you cache the filesystem or just the changes to the filesystem, the cache still has to be flushed.
 
Yeah, there is much tree rebuilding going on, I feel like a reverse lumberjack :(
 
Just thought I would make mention of this to anyone setting up any new machines for themselves or for clients, I do not suggest using ReiserFS.

In my current job, taking care of 15,000 odd servers, I'm forced to use ext3 because Red Hat doesn't support anything else. I seriously long for Reiser.

I suggest going for ext3 or even XFS.

Neither have the resilience in powerfail situations that Reiser has. Don't let the messages and the occasional fsck scare you. When you actually need to recover something you'll be glad you're on Reiser.
 
In my current job, taking care of 15,000 odd servers, I'm forced to use ext3 because Red Hat doesn't support anything else. I seriously long for Reiser.
Why? Why do you long for ReiserFS, what makes it better than any other journaling file system like ext3, JFS or even XFS?
 
Don't let the messages and the occasional fsck scare you.

Taking this out of the thread's context... :D

but, yes, all modern filesystems do have safeguards against accidental power loss.

Only when using FAT/FAT32 will one get crosslinked files, and that's fun...
 
Why? Why do you long for ReiserFS, what makes it better than any other journaling file system like ext3, JFS or even XFS?

I've no experience with XFS, but our redhat servers at work use ext3 and I use reiserfs on my work and home PCs. I have had to deal with numerous unplanned reboots both with the servers and with my PCs and the only OS to give me grief every time was ext3. Losing power on a PC running ext3 ALWAYS causes issues for me. I have yet to have a problem on reiserfs.

I also find reiserfs to be much faster w.r.t write operations (especially when deleting thousands of files at a time).
 
Why? Why do you long for ReiserFS, what makes it better than any other journaling file system like ext3, JFS or even XFS?

I didn't say I would choose reiserfs over JFS or XFS. I would, simply because I'm more familiar with it, but I know they are good under various circumstances. But that's of academic interest, since neither are supported by RHEL/RHAS. Only EXT2, EXT3 and vfat is.

Reason for preferring reiser over EXT3?

1. Much better successrate at recovering in the event of catastrophic failure.
2. Performance
 
I've no experience with XFS, but our redhat servers at work use ext3 and I use reiserfs on my work and home PCs. I have had to deal with numerous unplanned reboots both with the servers and with my PCs and the only OS to give me grief every time was ext3. Losing power on a PC running ext3 ALWAYS causes issues for me. I have yet to have a problem on reiserfs.

I also find reiserfs to be much faster w.r.t write operations (especially when deleting thousands of files at a time).

Ext3 is a file system, not an OS... :p
 
You might want to rethink your reasons...
http://linuxgazette.net/122/TWDT.html#piszcz
http://www.debian-administration.org/articles/388

The second one gives a bit of explanation on the test results, the first is only the results with no explanation or conclusion.

Anyway, each to his own!
:P

Indeed, each to his own. Can't say those benchmarks are compelling in the absolute, just confirms that some filesystems are better at some things than others.

Interesting that they show the command-line arguments to make the filesystems, and then leave out something crucial like the mounting options (notail, noatime etc.).
I use the noatime mount option, (all Linux filesystems appear to support noatime). I expect it would have quite an effect on those find benchmarks.
AFAIK notail is ReiserFS specific, but this implies less efficient usage of space because this type of packing counters fragmentation.

The "Remove 10,000 directories" test doesn't surprise me, Reiserfs is known to have a weakness with sparse files:
http://www.funtoo.org/en/articles/linux/ffg/2/

Performance issues

While ReiserFS generally blows the socks off ext2, ReiserFS does have a few special-case performance weaknesses. The first is sparse file performance. ReiserFS sparse file performance will be significantly worse than ext2. This will change at some point, when the Namesys developers get around to optimizing that part of ReiserFS for ReiserFS 4. Until then, ext2 is a better solution for applications that place heavy demands on sparse files.

The expected improvement isn't reflected in your quoted benchmarks for ReiserFS 4. One could speculate on the reasons for that but I'd rather not.

I can't say the mount time bothers me, and yes I have experienced the mount time for a 2TB RAID array. It's not like I have to reboot boxes like that very often.

Since that time, I have converted entirely to XFS for every Linux machine I use, so I may be a bit bias regarding the XFS filesystem.

ReiserFS takes a VERY long time to mount the filesystem. I included this test because I found it actually takes minutes to hours mounting a ReiserFS filesystem on a large RAID volume.

My experience is minutes, certainly not hours :rolleyes:.

XFS appears to scale better, though, will have to have a look at some point and run bonnie++ myself.

This webpage on FS tweaking might be of interest:
http://www.ubuntugeek.com/how-to-increase-ext3-and-reiserfs-filesystems-performance.html
 
Top
Sign up to the MyBroadband newsletter
X