We also use 2.4.19-64GB-SMP (SLES-8) and have the same problems. I looked into the kernel sources, the patch from Andy Cress is already applied. Worse, we got a broken disk last week. A cp-process became hanging in kernel (status 'D' with ps command) on a mirrored device. So we want to reboot the server. It did not work, we had to powercycle the machine. Our equipment: Compaq DL380, Adaptec AHA-3960D / AIC-7899A U160/m (rev 01), 10 34GB/70GB data disks on two busses each mirrored across plus one hotspare per bus. Andreas.Kahnt@coware.de ---------- Forwarded message ---------- Date: Tue, 8 Apr 2003 07:03:29 -0700 From: "Cress, Andrew R" <andrew.r.cress@intel.com> To: 'Alfred Isele' <Alfred.Isele@fujitsu-siemens.com> Cc: linux-raid@vger.kernel.org Subject: RE: raid1: resync hanging until reboot I've seen something similar in 2.4.18, and here's how we patched it. http://scsirastools.sourceforge.net/kern/2.4.18/raid_resync.patch It's short, so I've also included it below, but the formatting of the email may wrap some lines. Andy Cress --- linux-2.4.18/drivers/md/md.c.orig Thu Jan 30 10:35:01 2003 +++ linux-2.4.18/drivers/md/md.c Thu Jan 30 10:37:13 2003 @@ -3424,6 +3424,10 @@ wake_up(&mddev->recovery_wait); if (!ok) { // stop recovery, signal do_sync .... + if (mddev->pers->stop_resync) + mddev->pers->stop_resync(mddev); + if (mddev->recovery_running) + md_interrupt_thread(md_recovery_thread); } } @@ -3578,7 +3582,7 @@ * this also signals 'finished resyncing' to md_stop */ out: - wait_event(mddev->recovery_wait, atomic_read(&mddev->recovery_active)==0); + wait_disk_event(mddev->recovery_wait, atomic_read(&mddev->recovery_active)==0); up(&mddev->resync_sem); out_nolock: mddev->curr_resync = 0; -----Original Message----- From: Alfred Isele [mailto:Alfred.Isele@fujitsu-siemens.com] Sent: Monday, April 07, 2003 9:25 AM To: linux-raid@vger.kernel.org Subject: raid1: resync hanging until reboot Hello! When we disable a partition from a raid meta device and try to reattach it again without a reboot in between the resynch often starts hanging. After a reboot the resynch does complete. In detail: We use Linux version 2.4.19-64GB-SMP We create a raid1 meta devivce: mdadm -Cv /dev/md0 -l1 -n2 /dev/sda1 /dev/sdb1 We disable one piece from the meta device: mdadm /dev/md0 -f /dev/sda1 -r /dev/sda1 We reattach it to the meta device: mdadm /dev/md0 -a /dev/sda1 And very often but not always the resych hangs with cat /proc/mdstat showing: md0 : active raid1 sda1[2] sdb1[1] 80256 blocks [2/1] [_U] [>....................] recovery = 0.0% (0/80256) finish=26.7min speed=0K/sec Does anybody know a solution to this problem? Thank you very much Alfred - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
