Re: Some problems with the raid-stuff in 2.4.0-test12pre3
On Sat, Dec 02, 2000 at 06:56:43AM +1100, Neil Brown wrote: > On Friday December 1, [EMAIL PROTECTED] wrote: > > It's so slow that it's unusable. Especially writing. open() and > > close()-calls often hang for 20 seconds or more. > > write-calls hang for 3-4 seconds. This has to be a bug. > > But yes, after a long time, it finishes ;) > > Well, that does sound slower than I would expect > > 1/ Could you try: > >http://cgi.cse.unsw.edu.au/~neilb/patches/linux/2.4.0-test12-pre3/patch-E-raid5 > >and tell me how much that helps. That helped a _lot_! It's still slower than 2.2.x, but I'm happy with it! Thank you. I'm at your service if you need more testing ;) Regards Florian Heinz - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] Please read the FAQ at http://www.tux.org/lkml/
Re: Some problems with the raid-stuff in 2.4.0-test12pre3
On Friday December 1, [EMAIL PROTECTED] wrote: > On Fri, Dec 01, 2000 at 01:11:45PM +1100, Neil Brown wrote: > > On Thursday November 30, [EMAIL PROTECTED] wrote: > > > Hello people, > > > > > > I have some trouble with the raid-stuff. ... > > > > Is it just "very slow", but it eventually finishes, it is it so slow, > > that it actually stops and doesn't make any progress at all? > > > > raid5 in 2.4 is definately slower than in 2.2. Could that be all that > > you are seeing? > > It's so slow that it's unusable. Especially writing. open() and > close()-calls often hang for 20 seconds or more. > write-calls hang for 3-4 seconds. This has to be a bug. > But yes, after a long time, it finishes ;) Well, that does sound slower than I would expect 1/ Could you try: http://cgi.cse.unsw.edu.au/~neilb/patches/linux/2.4.0-test12-pre3/patch-E-raid5 and tell me how much that helps. 2/ Try a larger chunk size. My testing suggests 64K is a good starting point. NeilBrown - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] Please read the FAQ at http://www.tux.org/lkml/
Re: Some problems with the raid-stuff in 2.4.0-test12pre3
On Fri, Dec 01, 2000 at 01:11:45PM +1100, Neil Brown wrote: > On Thursday November 30, [EMAIL PROTECTED] wrote: > > Hello people, > > > > I have some trouble with the raid-stuff. > > My machine is a Pentium-III, 256 MB ram and 7 scsi-disks (IBM DNES-318350W > > 17B). I'm using raid5 for 6 of these disks (chunk-size 8). > > Machine boots, I do mkraid /dev/md0 and then mke2fs /dev/md0 and that's > > where the problems start. mkfs tries to write 684 inode-tables and after the > > first 30 it gets very slow. ps ax (with wchan) tells me it hangs in > > wakeup_bdflush. > > I'm rather sure it's related to the raidcode, because without raid the disks > > work as expected. > > I'm using an Adaptec 7892A with the aic7xxx-driver, I have disabled the TCQ > > and the extra checks for the new queueing code, but I have tried with both > > activated, too. > > No related messages from the kernel in the syslog. > > It worked fine with 2.2.x. > > Is it just "very slow", but it eventually finishes, it is it so slow, > that it actually stops and doesn't make any progress at all? > > raid5 in 2.4 is definately slower than in 2.2. Could that be all that > you are seeing? It's so slow that it's unusable. Especially writing. open() and close()-calls often hang for 20 seconds or more. write-calls hang for 3-4 seconds. This has to be a bug. But yes, after a long time, it finishes ;) - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] Please read the FAQ at http://www.tux.org/lkml/
Re: Some problems with the raid-stuff in 2.4.0-test12pre3
On Fri, Dec 01, 2000 at 01:11:45PM +1100, Neil Brown wrote: On Thursday November 30, [EMAIL PROTECTED] wrote: Hello people, I have some trouble with the raid-stuff. My machine is a Pentium-III, 256 MB ram and 7 scsi-disks (IBM DNES-318350W 17B). I'm using raid5 for 6 of these disks (chunk-size 8). Machine boots, I do mkraid /dev/md0 and then mke2fs /dev/md0 and that's where the problems start. mkfs tries to write 684 inode-tables and after the first 30 it gets very slow. ps ax (with wchan) tells me it hangs in wakeup_bdflush. I'm rather sure it's related to the raidcode, because without raid the disks work as expected. I'm using an Adaptec 7892A with the aic7xxx-driver, I have disabled the TCQ and the extra checks for the new queueing code, but I have tried with both activated, too. No related messages from the kernel in the syslog. It worked fine with 2.2.x. Is it just "very slow", but it eventually finishes, it is it so slow, that it actually stops and doesn't make any progress at all? raid5 in 2.4 is definately slower than in 2.2. Could that be all that you are seeing? It's so slow that it's unusable. Especially writing. open() and close()-calls often hang for 20 seconds or more. write-calls hang for 3-4 seconds. This has to be a bug. But yes, after a long time, it finishes ;) - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] Please read the FAQ at http://www.tux.org/lkml/
Re: Some problems with the raid-stuff in 2.4.0-test12pre3
On Friday December 1, [EMAIL PROTECTED] wrote: On Fri, Dec 01, 2000 at 01:11:45PM +1100, Neil Brown wrote: On Thursday November 30, [EMAIL PROTECTED] wrote: Hello people, I have some trouble with the raid-stuff. ... Is it just "very slow", but it eventually finishes, it is it so slow, that it actually stops and doesn't make any progress at all? raid5 in 2.4 is definately slower than in 2.2. Could that be all that you are seeing? It's so slow that it's unusable. Especially writing. open() and close()-calls often hang for 20 seconds or more. write-calls hang for 3-4 seconds. This has to be a bug. But yes, after a long time, it finishes ;) Well, that does sound slower than I would expect 1/ Could you try: http://cgi.cse.unsw.edu.au/~neilb/patches/linux/2.4.0-test12-pre3/patch-E-raid5 and tell me how much that helps. 2/ Try a larger chunk size. My testing suggests 64K is a good starting point. NeilBrown - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] Please read the FAQ at http://www.tux.org/lkml/
Re: Some problems with the raid-stuff in 2.4.0-test12pre3
On Sat, Dec 02, 2000 at 06:56:43AM +1100, Neil Brown wrote: On Friday December 1, [EMAIL PROTECTED] wrote: It's so slow that it's unusable. Especially writing. open() and close()-calls often hang for 20 seconds or more. write-calls hang for 3-4 seconds. This has to be a bug. But yes, after a long time, it finishes ;) Well, that does sound slower than I would expect 1/ Could you try: http://cgi.cse.unsw.edu.au/~neilb/patches/linux/2.4.0-test12-pre3/patch-E-raid5 and tell me how much that helps. That helped a _lot_! It's still slower than 2.2.x, but I'm happy with it! Thank you. I'm at your service if you need more testing ;) Regards Florian Heinz - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] Please read the FAQ at http://www.tux.org/lkml/
Re: Some problems with the raid-stuff in 2.4.0-test12pre3
On Thursday November 30, [EMAIL PROTECTED] wrote: > Hello people, > > I have some trouble with the raid-stuff. > My machine is a Pentium-III, 256 MB ram and 7 scsi-disks (IBM DNES-318350W > 17B). I'm using raid5 for 6 of these disks (chunk-size 8). > Machine boots, I do mkraid /dev/md0 and then mke2fs /dev/md0 and that's > where the problems start. mkfs tries to write 684 inode-tables and after the > first 30 it gets very slow. ps ax (with wchan) tells me it hangs in > wakeup_bdflush. > I'm rather sure it's related to the raidcode, because without raid the disks > work as expected. > I'm using an Adaptec 7892A with the aic7xxx-driver, I have disabled the TCQ > and the extra checks for the new queueing code, but I have tried with both > activated, too. > No related messages from the kernel in the syslog. > It worked fine with 2.2.x. Is it just "very slow", but it eventually finishes, it is it so slow, that it actually stops and doesn't make any progress at all? raid5 in 2.4 is definately slower than in 2.2. Could that be all that you are seeing? NeilBrown - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] Please read the FAQ at http://www.tux.org/lkml/
Re: Some problems with the raid-stuff in 2.4.0-test12pre3
On Thursday November 30, [EMAIL PROTECTED] wrote: Hello people, I have some trouble with the raid-stuff. My machine is a Pentium-III, 256 MB ram and 7 scsi-disks (IBM DNES-318350W 17B). I'm using raid5 for 6 of these disks (chunk-size 8). Machine boots, I do mkraid /dev/md0 and then mke2fs /dev/md0 and that's where the problems start. mkfs tries to write 684 inode-tables and after the first 30 it gets very slow. ps ax (with wchan) tells me it hangs in wakeup_bdflush. I'm rather sure it's related to the raidcode, because without raid the disks work as expected. I'm using an Adaptec 7892A with the aic7xxx-driver, I have disabled the TCQ and the extra checks for the new queueing code, but I have tried with both activated, too. No related messages from the kernel in the syslog. It worked fine with 2.2.x. Is it just "very slow", but it eventually finishes, it is it so slow, that it actually stops and doesn't make any progress at all? raid5 in 2.4 is definately slower than in 2.2. Could that be all that you are seeing? NeilBrown - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] Please read the FAQ at http://www.tux.org/lkml/