[Bug 231494] New: abysmal ext3 performance
https://bugzilla.novell.com/show_bug.cgi?id=231494 Summary: abysmal ext3 performance Product: openSUSE 10.2 Version: Final Platform: Other OS/Version: Other Status: NEW Severity: Normal Priority: P5 - None Component: Kernel AssignedTo: kernel-maintainers@forge.provo.novell.com ReportedBy: seife@novell.com QAContact: qa@suse.de OtherBugsDependingO 215208 nThis: seife@strolchi:/local/video> mount /dev/sda5 on / type ext3 (rw,acl,user_xattr) proc on /proc type proc (rw) sysfs on /sys type sysfs (rw) debugfs on /sys/kernel/debug type debugfs (rw) udev on /dev type tmpfs (rw) devpts on /dev/pts type devpts (rw,mode=0620,gid=5) /dev/sda2 on /boot type ext2 (rw,acl,user_xattr) /dev/sda7 on /local type ext3 (rw) securityfs on /sys/kernel/security type securityfs (rw) /dev/sda6 on /home type ext3 (rw,noatime,loop=/dev/loop0,encryption=twofish256) /home/seife/local/news on /var/spool/news type none (rw,bind) automount(pid16746) on /mounts type autofs (rw,fd=4,pgrp=16746,minproto=2,maxproto=4) automount(pid16748) on /suse type autofs (rw,fd=4,pgrp=16748,minproto=2,maxproto=4) automount(pid16750) on /secret type autofs (rw,fd=4,pgrp=16750,minproto=2,maxproto=4) seife@strolchi:/local/video> ls -l bigdir/ total 1304027 -rw-r--r-- 1 seife suse 1329380501 3. Jan 03:25 001.vdr -rw-r--r-- 1 seife suse 719576 3. Jan 03:25 index.vdr -rw-r--r-- 1 seife suse 2508 3. Jan 02:25 info.vdr seife@strolchi:/local/video> du -sh bigdir/ 1,3G bigdir/ seife@strolchi:/local/video> time rm -rf bigdir/ real 0m53.093s user 0m0.004s sys 0m2.100s So it takes 1 second per 24MB to delete files. This is bad. With reiserfs and with xfs, the rm takes practically mo time at all. Setup is a 10.2, fresh installed with standard ext3 options on a Pentium M 1,6GHz, with 2MB RAM, intel 855 chipset, using the libata drivers (ata_piix) -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=231494 ------- Comment #1 from seife@novell.com 2007-01-05 10:04 MST ------- Note that the harddrive of this machine is not the fastest: root@strolchi:/local# hdparm -tT /dev/sda /dev/sda: Timing cached reads: 1456 MB in 2.00 seconds = 727.54 MB/sec Timing buffered disk reads: 78 MB in 3.02 seconds = 25.82 MB/sec root@strolchi:/local# hdparm -tT --direct /dev/sda /dev/sda: Timing O_DIRECT cached reads: 32 MB in 2.06 seconds = 15.55 MB/sec Timing O_DIRECT disk reads: 46 MB in 3.08 seconds = 14.95 MB/sec (the values are exactly the same if i use the old piix driver instead of ata_piix) The drive is a hitachi 40G Notebook harddrive, Model Number: HTS424040M9AT00 I will probably replace it next week with an 80G seagate drive. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=231494 jack@novell.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |ASSIGNED ------- Comment #2 from jack@novell.com 2007-01-05 10:26 MST ------- At one point there was a bug in Yast and it configured ext3 with 1KB blocksize instead of 4KB. What is the output of dumpe2fs for /local? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=231494 ------- Comment #3 from seife@novell.com 2007-01-05 10:50 MST ------- Created an attachment (id=111670) --> (https://bugzilla.novell.com/attachment.cgi?id=111670&action=view) dumpe2fs First block: 1 Block size: 1024 Fragment size: 1024 Blocks per group: 8192 Fragments per group: 8192 Inodes per group: 1024 Inode blocks per group: 128 so it looks like it did that. All the FS on that machine are ext3. I guess i will have to reinstall the machine to fix that? (and all machines at home :-() That was, btw, an installation of oS10.2 final, so the bug is still in the final version. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=231494 jack@novell.com changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |aj@novell.com ------- Comment #4 from jack@novell.com 2007-01-05 12:01 MST ------- Andreas Jaeger seemed to confirm in the bug 220200, that this has been fixed. Are you sure that you are installing final oS10.2 (it has been fixed really in the last moments...)? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=231494 ------- Comment #5 from seife@novell.com 2007-01-05 13:14 MST ------- ARGH! i installed RC2 and later updated to RC4/final, now that you mention it. Anyway, the root fs (/dev/sda5) has root@strolchi:~# dumpe2fs /dev/sda5 dumpe2fs 1.39 (29-May-2006) Filesystem volume name: root Last mounted on: <not available> Filesystem UUID: 4b2191ee-47e7-4283-a98a-8aceaa59abf1 Filesystem magic number: 0xEF53 Filesystem revision #: 1 (dynamic) Filesystem features: has_journal resize_inode dir_index filetype needs_recovery sparse_super large_file Default mount options: (none) Filesystem state: clean Errors behavior: Continue Filesystem OS type: Linux Inode count: 853088 Block count: 1704890 Reserved block count: 85244 Free blocks: 228034 Free inodes: 590674 First block: 0 Block size: 4096 Fragment size: 4096 Reserved GDT blocks: 416 Blocks per group: 32768 so i guess i do not need to reinstall, just backup the /local partition sda7, which i need to do anyway when putting in a larger harddrive (the other partitions will just be dd'ed over from the old drive. So does the 1k blocksize explain the performance problem? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=231494 jack@novell.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |DUPLICATE ------- Comment #6 from jack@novell.com 2007-01-05 13:41 MST ------- In bug 220200 it did. We'll see if it's true also in your case :). So I'll mark this bug as a duplicate of 220200 and if your problems won't be solved by changing the blocksize, then please reopen... *** This bug has been marked as a duplicate of bug 220200 *** -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=231494 ------- Comment #7 from seife@novell.com 2007-01-05 15:06 MST ------- Ok. I already did it on the cryptohome (there it also was 1k, i backed it up, mkfs.ext3 -O dir_index /dev/loop0; mounted it and restored the backup) and the machine already "feels better" :-) I'll do the same stunt with the /local partition on monday (i don't have enough free space available right now) and will complain if it does not help. Unless there is a method to do this without mkfs, then i can do it immediately, but i don't think so... :-) -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
participants (1)
-
bugzilla_noreply＠novell.com