Which filesystem for AMD64

secondshadow · Guru Joined: 23 Jun 2003 Posts: 362

If you go back and read your previous statements, you make more comparisions to ReiserFS vs ext3 than you do XFS.

amax · Posted: Tue Feb 24, 2004 6:04 pm Post subject:

OK

secondshadow · Guru Joined: 23 Jun 2003 Posts: 362

If you'd bothered to pay attention I put that comment in a section specifically with that warning RANT, and besides that I didn't call you stupid, I warned that you were at the risk of looking stupid. Not only that, but you just admitted to not knowing what you were talking about to begin with:

amax · Posted: Tue Feb 24, 2004 6:41 pm Post subject:

also for reiserfs fanatics

my personal latest problem on my /home was

Fatal corruptions were found, Semantic pass skipped
There were found 17 corruptions which can be fixed only during --rebuild-tree

and fixing was _5_HOURS_ on my 160gb drive

next. if I working with MIXED fies many small and (like wideo in *.jpeg)
and some BIG video files.. then... IF MY BIG FILE is very high fragmented, ALL disk partition working slow slow slow.. even on EXT3 work with very fragmented files will be faster.

next. reiser used fixed block size 4k ? or now it is flexible?

next. tailing option increase fragmentation and decrease recoverable. if notail, it will more compatible with lilo etc, but key perfomance feature "integration tailed file into B-Tree" will be off.

Reiser use only metadata journaling - not full journaling - with ext3 it is possible to use full journaling.

NEXT. on RAID level 5

try check perfomance with reiserfs ?! it will equal with ext3 even on READ. on write it will be slowly (only my IMHO, I tested it)

next. recovery level. on ext3 ANY even it is very small change in data ANY action performing writing of full journalig block with metadata. power interrupt give me minimum surprises on ext3 instead...

it is only my IMHO. for servers where is used ext3 on / and /usr with -j it maybe much stable, recoverable.

for specific things reiserfs is much better.

for disks with bad blocks XFS and ReiserFS is not acceptable.

for stupid users "like me" ext3 is good because they are belive it

on amd64 env it is not tested for me..

at home workstation I am preferred to use XFS

amazing etc # mount
/dev/hda6 on / type xfs (rw,sync,noatime)
none on /dev type devfs (rw)
none on /proc type proc (rw)
none on /sys type sysfs (rw)
none on /dev/pts type devpts (rw,mode=0620)
/dev/hda7 on /var type xfs (rw,nosuid,nodev,noatime)
/dev/hda8 on /usr type xfs (rw,sync,noatime)
/dev/hda9 on /var/log type ext3 (rw,noatime)
/dev/hda10 on /tmp type xfs (rw,noatime)
/dev/hda13 on /home type xfs (rw,nosuid,nodev,noatime)
/dev/hdb1 on /mnt/hdb1 type vfat (rw,umask=000,codepage=866,iocharset=iso8859-1)
none on /mnt/cdrom type supermount (ro,dev=/dev/sr0,fs=iso9660,--,unhide)
none on /mnt/floppy type supermount (rw,sync,dev=/dev/fd0,fs=auto,--,iocharset=koi8-r,umask=000)
none on /mnt/flash type supermount (rw,sync,dev=/dev/sdb1,fs=auto,--,iocharset=koi8-r,umask=000)
none on /dev/shm type tmpfs (rw)
none on /proc/sys/fs/binfmt_misc type binfmt_misc (rw)
none on /proc/bus/usb type usbfs (rw)

amax · Posted: Tue Feb 24, 2004 6:47 pm Post subject:

info from http://www.namesys.com/

Reiser4 is in final testing, and will ship soon!
etc..

good news!

Soon I will try reiserfs again

)

secondshadow · Guru Joined: 23 Jun 2003 Posts: 362

amax: once again I never called you stupid. As far as you finally reading up on the FS, good for you. I'm proud. As far as bad blocks on the disk...well...if your disk has bad blocks you've probably got more problems than you realize. Quite often bad blocks are a symptom of HUGE problems in the future and the disks should (usually) be removed as soon as possible in my experiance...ESPECIALLY in a server environment. You really don't want the server going up due to a hard drive failure.

As for 5 hours? Don't know what to tell you. I use ReiserFS partitions in the 40-60 gig range. I've had to do a rebuild-tree maybe 3 times, two of which were using an early stage ReiserFS (def: early stage - just out of beta) after power failures. It didn't take nearly that long and that was a slow machine. On my old AthlonXP 1600+ a rebuild tree on a 60 gig partition only took like 5 minutes or something like that. Don't know what to tell you beyond that, and the only times I've EVER had to do that are on power failures...maybe one of the times was a nasty hardlock but that was my fault anyway and it was a test-machine to boot, lotsa experimental stuff. So if its a production machine, you might consider a UPS that can warn the kernel of the power failure so the machine can go down safely.

As for fragmentation, I've never had the speed problems with fragmentation personally and I store files of a fair variety of sizes ranging from cookies up to 3.2 gig drive images formatted with ReiserFS themselves for cross-compiling for my K6-2 downstairs. I've never had a performance problem, even with the Reiser-in-Reiser situation.

This thread has just become a stupid flamewar (the thread being what is stupid not you amax, just so we're clear). I think I'm gonna unwatch it just so the temptation to post is removed.

[Ag] · n00b Joined: 24 Mar 2004 Posts: 1

ectospasm · Posted: Thu Mar 25, 2004 3:55 am Post subject:

Evangelion · Posted: Thu Mar 25, 2004 10:49 am Post subject:

Corona688 · Veteran Joined: 10 Jan 2004 Posts: 1204

Another vote for ReiserFS. Been running perfectly, and a lot more efficiently than ext2/ext3. It's also survived a dual power surge(first surge probably didn't hurt much; second was right in the middle of the bootup process! ) I needed to restore one file from backup. Everything else was A-OK.

(side note: I got my system running back in Jan, when the liveCD's didn't support SATA, so I needed to bootstrap from an ancient 1GB drive. ext2 couldn't fit the kernel source/build on a 1GB drive. reiserfs could.

)