You are not logged in.
After the last BIG update, i've noticed that my pacman is a lot slower when sync and/or installing packages.
I've done a pacman-optimize, a XFS_fsr and nothing changed this...
Any ideas?
Edit: New info.. moderators please change this thread to the right forum index :oops:
Offline
I wouldn't blame it on the update, XFS is just extremely slow with pacman by nature - it's designed to deal with huge files, but has abominable performance with very small ones.
Offline
the same for me... running pacman on XFS for 1 month, everything was fine, after the BIG UPDATE it is VERY slow!
Maybe because of the new Kernel?
Or even because of mkinitcpio??
digiKam developer - www.digikam.org
Offline
Weird, that doesn't really make sense, not if pacman-optimize didn't help... :?
Offline
Im running pacman-optimize for the fourth time now, its getting even slower!
Before the BIG UPDATE I did a pacman-optimize and it took 30 seconds, and after another run it took only 15 seconds.
Now it takes 3 minutes, and the second and third time takes even a little longer... that is VERY strange.
I did a rsync backup and will format my / with reiserfs now, play back the backup and see if it is faster.
If not you really can't blame it on XFS I think...
digiKam developer - www.digikam.org
Offline
OK my root partition is running ReiserFS now.
Pacman is VERY fast now, BUT:
Maybe this is just because the files on that partition are not fragmented or something like that.
I will test it for 1 week or 2 and see what will happen... but I think that ReiserFS is slower than XFS, Firefox is booting slower and KDE is also starting slower...
but pacman is fast now!
digiKam developer - www.digikam.org
Offline
OK my root partition is running ReiserFS now.
Pacman is VERY fast now, BUT:
Maybe this is just because the files on that partition are not fragmented or something like that.
I will test it for 1 week or 2 and see what will happen... but I think that ReiserFS is slower than XFS, Firefox is booting slower and KDE is also starting slower...
but pacman is fast now!
maybe you should think about using more stable file systems. Ext2 and Ext3 should work eprfectly fine with good speeds. I've used Ext3 for a long time now and havn't had much of any speed issues. But since you think it fragmenting that's slowing the system down, switching to ReiserFS probablt wasn't the best move, it fragments very easily.
Offline
The strange thing:
starting xfs_db, it said: 0.33% Fragmentation, so this is nearly nothing...
I don't know why it was so slow, but now I switched my home-partition to ReiserFS, too.
EXT3 was the FS I used before, but it was definitly slower than XFS.
digiKam developer - www.digikam.org
Offline
OK my root partition is running ReiserFS now.
Pacman is VERY fast now, BUT:
Maybe this is just because the files on that partition are not fragmented or something like that.
I will test it for 1 week or 2 and see what will happen... but I think that ReiserFS is slower than XFS, Firefox is booting slower and KDE is also starting slower...
but pacman is fast now!
ReiserFS indeed does have slower read speeds than XFS.
Offline
Maybe I should format my partitions every week, so pacman is fast... :cry:
digiKam developer - www.digikam.org
Offline
so now I switched to EXT3, the FS I had used the last time 3 years ago.
I created the partition with
mke2fs -j -O dir_index
like Gullible Jones told in another thread.
Maybe EXT3 is faster with these settings, I will see.
digiKam developer - www.digikam.org
Offline
It's somewhere between ReiserFS and XFS with pacman speed, but has good read speeds in my experience. JFS is also in that middle ground.
BTW... If you come back to XFS, you could try this.
Offline
It's somewhere between ReiserFS and XFS with pacman speed, but has good read speeds in my experience. JFS is also in that middle ground.
BTW... If you come back to XFS, you could try this.
Hmmm I don't know, all this switching is getting boring !!
If EXT3 is nearly the same speed as my XFS installation, and pacman speed will be consistent, I will stay with EXT3 for now...
digiKam developer - www.digikam.org
Offline
I think the problem may not be of pacman... i'm getting a lot of loadAVG and system locks when copying data from/to disk/usb.
It may be some IDE chipset driver change or sheduler change...but things are a LOT slower after last big update, so i think it's not just XFS because i'me talking big file transfers (about 600~4000MB) so it's not XFS fault, I defrag regulary.
Offline
That's right, I noticed that too.
Copying data from HDD to my USB drive is slower than before...
digiKam developer - www.digikam.org
Offline
Try hdparm. Carefully.
Offline
hdparm -tT /dev/hda
Timing cached reads: 2008 MB in 2.00 seconds = 1003.35 MB/sec
Timing buffered disk reads: 90 MB in 3.01 seconds = 29.90 MB/sec
dmesg | grep hda
hda: 117304992 sectors (60060 MB) w/8192KiB Cache, CHS=16383/255/63, UDMA(100)
hda: cache flushes supported
dmesg | grep IDE
PCI: Ignoring BAR0-3 of IDE controller 0000:00:02.5
Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2
SIS5513: IDE controller at PCI slot 0000:00:02.5
SIS5513: SiS 962/963 MuTIOL IDE UDMA133 controller
Offline
you could try the "pacman-drive" script. it's somewhere in the forums... it has reduced the time pacman takes to initialize from 50+sec to 5sec!
Offline
I think the problem here goes quite a bit beyond that
hdparm -tT /dev/hda
Timing cached reads: 2008 MB in 2.00 seconds = 1003.35 MB/sec Timing buffered disk reads: 90 MB in 3.01 seconds = 29.90 MB/sec
What kind of hard drive are you using? The buffered read speed is quite a bit lower than what I get, even though you've got UDMA5 enabled and all... :?
(Any HDD experts around here?)
Offline
This is a laptop so the HD isn't that fast like the "big ones" :twisted:
The problem was fixed in some way using pacman-drive script, i recommend that to pacman dev's, it's a very very fast way of getting those anoying files get all around the discs making pacman slow.
The loadAVG of the HD transfers has lowered a bit since some updates (kernel 2.6.17.3 and so on) so i'm not getting the freezes when copying to/from HD/USB to HD/USB and that was the point of this thread, i've read the kernels changelog and found nothing in there that could make that happen, but it may be a miss configuration in the kernel compilation.
So if you got XFS (and also improvments for others) use pacman-drive script it made wonders for me
Offline