You are not logged in.
I would like to check if my XFS volume is fragmented...(cause pacman is really slow with -Ss and -Sy these days..not sure why)
$ xfs_db -c frag -r /dev/sdb2
this returns..
$ Segmentation fault
I'm pretty sure that the command is correct..any ideas?
Thank in advance
Offline
Happened to me when I tried XFS... And not just that. Other applications, for example links, would segfault every time I ran them; while very similar software would often work (e.g. elinks). I don't know what that was about... Maybe something to do with the ACL support?
Offline
I'm sorry I googled for it but I don't quite understand what it means..Did you have pacman slowness problems too?
Offline
I'm sorry I googled for it but I don't quite understand what it means..Did you have pacman slowness problems too?
have you tried running pacman-optimize? That'll re-align the inodes and data of the pacman database directories into one contingous chunk.
Offline
Doesn't help...I've tried adjutsing /etc/pacman.d/* to better mirrors..doesn't seem to help
Offline
1) Is your / volume actually sdb2? sda2?
2) Try to run frag interactively..to report fragmentation
root@vw-linux /data # xfs_db -r /dev/sda3
xfs_db> frag
actual 62504, ideal 440, fragmentation factor 99.30%
xfs_db>
3) try the defrag...
http://oss.sgi.com/archives/linux-xfs/2 … 00090.html
"Be conservative in what you send; be liberal in what you accept." -- Postel's Law
"tacos" -- Cactus' Law
"t̥͍͎̪̪͗a̴̻̩͈͚ͨc̠o̩̙͈ͫͅs͙͎̙͊ ͔͇̫̜t͎̳̀a̜̞̗ͩc̗͍͚o̲̯̿s̖̣̤̙͌ ̖̜̈ț̰̫͓ạ̪͖̳c̲͎͕̰̯̃̈o͉ͅs̪ͪ ̜̻̖̜͕" -- -̖͚̫̙̓-̺̠͇ͤ̃ ̜̪̜ͯZ͔̗̭̞ͪA̝͈̙͖̩L͉̠̺͓G̙̞̦͖O̳̗͍
Offline
ACL = access control lists, something XFS supports. My line of thinking was that a segfault occurs when a program tries to access memory that isn't allocated for its use, so it might have something to do with the ACL support. I'm probably dead wrong of course.
Offline
cactus: doesn't work interactively either..
For some odd reason, xfs_fsr doesnt exist in the xfsprogs package..In fact I dont think its in any xfs package anymore cause its not there in Gentoo 2005.1 either.
Anyway I ran the command thru the gentoo live CD and it reported a fragmentation of 1%..so no worries there..I think it maybe the arch package which may be borked? (xfsprogs 2.3.36-1)
Another strange thing is bonnie showed the best results for XFS(mkfs.xfs -f -d su=128k,sw=2) (2 HDD raid0 with 64k stripe)compared to all other File systems(including Reiser4), but realworld tests show that reiser4 is 10 seconds faster in moving the kernel source from one directory to another. The tests were carried out on a non root drive and i restarted after each test so that stuff loaded in the RAM doesn't mess with the results...why?
and pacman is still slow with pacman -Q or pacman -Sy....
Offline
I'm getting Segmentation fault aswell, maybe one of the packages are busted? Of course I only use it for my anime harddrive so it doesn't matter much to me.
Offline
Its worth fixing if it was on the Hentai drive right? :twisted:
Offline
Its worth fixing if it was on the Hentai drive right? :twisted:
hehe definitely. I'm planning to use xfs on my next harddrive (whenever i get the money ) so it would be nice to see exactly what is causing the Segmentation fault.
Offline
Maybe you should try ext3 with directory indexing - according to Namsys benchmarks, it seems to be quite good for reading and writing large files.
Offline
2) Try to run frag interactively..to report fragmentation
root@vw-linux /data # xfs_db -r /dev/sda3 xfs_db> frag actual 62504, ideal 440, fragmentation factor 99.30% xfs_db>
ryan@dogma ~ $ xfs_db -r /dev/hdb1
xfs_db> frag
xfs_db: out of memory
Offline
Hi,
Yeah, I'm getting the seg fault as well, haven't had a chance to look into yet. Was planning in looking at in tonight.
Jon
Offline
Right if this applies to everyone then we should file a bug report. Looks most likely like a bad package. PKGBUILD shows a 'compile.patch' being used too.
Offline
I'm having the segfault as well. But I didn't have it a few weeks ago... strange.
For some odd reason, xfs_fsr doesnt exist in the xfsprogs package..In fact I dont think its in any xfs package anymore cause its not there in Gentoo 2005.1 either.
xfs_fsr is part of the xfsdump package. I started a thread on that a while back:
http://bbs.archlinux.org/viewtopic.php? … ht=xfsdump
And here's a feature request:
http://bugs.archlinux.org/index.php?do=details&id=3178
but yeah, let's file a bugreport. I mean, we're quite a few people who have the segfault prob.
Offline
I got all xfs related packages by using zeldan's PKGBUILDS.
http://bbs.archlinux.org/viewtopic.php? … ht=xfsdump
Note that you have to change the version numbersin the PKGBUILDS to match the ones listed here
ftp://oss.sgi.com/projects/xfs/cmd_tars/
Also note that xfsprogs wont build unless you use the compile patch in /var/abs/system/xfsprogs. So modify these PKGBUILDS to include the compile patch.
Now I'm able to defragment my drive, but still not able to check fragmentation status (using xfs_db -c frag -r /dev/hdx), Segmentation Fault again..
Is this because of the compile patch?
Can someone please file a bug report cause I haven't set up an account yet.
Offline
I've filed a bugreport. http://bugs.archlinux.org/index.php?do=details&id=3199.
How about uploading the new PKGBUILDs to AUR?
Offline
archliux.org was down for a while yesterday.. ill do it today
Offline
Hi,
Ok downloaded xfs_progs from sgi and complied it myself and xfs_db now works fine, so it gotta be a problem with the package for sure now.
Its a no brainer to compile btw, so if you need xfs_db now just grab the tar ball from ftp://xfs.org/mirror/SGI/Release-1.3.1/cmd_tars/ and do the usual configure, make, make install
Works for me anyway. Be interesting to see if this works for others.
Cheers,
Jon
Offline
Hi,
Slight correction, the version I tested above was 2.5.6 not the current version which 2.6.36. However, I decided to test 2.6.13 out of interest.
Got that copy and complied it, and it works fine as well. Lets try 2.6.36 next
Jon
Offline
OK now the results of 2.6.36 testing, available from ftp://oss.sgi.com/projects/xfs/cmd_tars/
This complied OK, but seg faults on xfs_db. Hmmm, interesting.
It would appear that 2.6.36 has a little problem. It would seem safest to drop back to 2.6.13 for the moment.
Comments?
Jon
Offline
Last post for today honest, looks like there _is_ a problem with xfsprog, found this bug reports on SGI Bugzilla for xfs_check
http://oss.sgi.com/bugzilla/show_bug.cgi?id=248
Well, at least it looks like its not just limited to Arch. Thinking about putting one in for xfs_db, but would be helpful if one of you guys could do some testing as well before hand.
Can we down grade the current production version for now pls
Jon
Offline
and can the production version include a 'make install-dev' for xfs progs cause xfsdump will not work without that. I was about to upload to the five xfs packages to the AUR but rules say that you should not upload a package already in extra/current/whatever, so i cannot upload an xfsprogs with make install-dev
Offline
Still a problem. Just tried with the new arch package..
xfsprogs 2.7.3-1
still a no go..
# xfs_db -rc frag /dev/sda6
Segmentation fault
"Be conservative in what you send; be liberal in what you accept." -- Postel's Law
"tacos" -- Cactus' Law
"t̥͍͎̪̪͗a̴̻̩͈͚ͨc̠o̩̙͈ͫͅs͙͎̙͊ ͔͇̫̜t͎̳̀a̜̞̗ͩc̗͍͚o̲̯̿s̖̣̤̙͌ ̖̜̈ț̰̫͓ạ̪͖̳c̲͎͕̰̯̃̈o͉ͅs̪ͪ ̜̻̖̜͕" -- -̖͚̫̙̓-̺̠͇ͤ̃ ̜̪̜ͯZ͔̗̭̞ͪA̝͈̙͖̩L͉̠̺͓G̙̞̦͖O̳̗͍
Offline