Messages in this thread | ![/](/images/icornerl.gif) | | From | Michael Buesch <> | Subject | Re: [2.6] system is very slow during disk access | Date | Thu, 7 Aug 2003 11:53:02 +0200 |
| |
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On Thursday 07 August 2003 06:13, Nuno Silva wrote: > Hi!
Hi,
> There are some references to: hdparm -a 512 /dev/hda. > For some folks this improves fs performance. > > Does it work for you?
Changing read-ahead doesn't work. Neither to 8 nor to 512. I think it doesn't have something to do with read-ahead, because the system-performance-drop occurs while writing to the disk. My dd-example wrote to disk:
dd if=/dev/zero of=./t.test
I just ran a few other tests:
dd if=/dev/hda of=/dev/null doesn't drop system performance and the system is very usable while dd is running. So reading from a plain partition is OK.
dd if=/dev/zero of=/dev/hdc7 This slows down the system _very_ much, but it's still usable. The mouse curser doesn't jump randomly over the screen, like it did while dd if=/dev/zero of=./t.test
So my guess, it may have something to do with reiserFS. So I ran another test:
I did mke2fs /dev/hdc7 mount /dev/hdc7 /mnt/data_1 cd /mnt/data_1 dd if=/dev/zero of=./t.test
And now the surprise. :) Here the system-behaviour is exactly the same, as while writing to the plain partition. The cursor is not smooth, but it's usable and it doesn't jump ramdonly over the screen.
So I think it has something to do with reiserFS. Where can I start to track it down, why reiserFS is doing this? Hmm, I'll try to enable CONFIG_REISERFS_CHECK and CONFIG_REISERFS_PROC_INFO
Can somebody tell me how to debug reiserFS with these options enabled?
> Regards, > Nuno Silva
Short note: hda and hdc are both exactly the same devices.
- -- Regards Michael Buesch [ http://www.8ung.at/tuxsoft ] Penguin on this machine: Linux 2.6.0-test2 - i386
-----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.1 (GNU/Linux)
iD8DBQE/MiGKoxoigfggmSgRAt+PAJ9gI8Z5Osm4/DcnFZsPtR7x4UQpZACfZw7R GxC2PwCVOAWFFE39Fw9ttKM= =v0nD -----END PGP SIGNATURE-----
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| ![\](/images/icornerr.gif) |