Seting up a two Raid-1 disk array I've found the following problem: I have had to use persistent-superblock=0 because the disk was in a production server and couldn't be much time off so I used one existing partition with data and in the free space created the second partition of the array. (Both in the same disk, Yes, I know I lost security and performance but have to be done in that way.) Different size partitions, I supposed that it should work with the more little size of both but... Feb 28 10:40:23 atalaska kernel: nonpersistent superblock ... Feb 28 10:40:23 atalaska kernel: hda4's event counter: 00000000 Feb 28 10:40:23 atalaska kernel: hda2's event counter: 00000000 Feb 28 10:40:23 atalaska kernel: md0: max total readahead window set to 128k Feb 28 10:40:23 atalaska kernel: md0: 1 data-disks, max readahead per data-disk: 128k Feb 28 10:40:23 atalaska kernel: raid1: device hda4 operational as mirror 1 Feb 28 10:40:23 atalaska kernel: raid1: device hda2 operational as mirror 0 Feb 28 10:40:23 atalaska kernel: raid1: raid set md0 not clean; reconstructing mirrors Feb 28 10:40:23 atalaska kernel: raid1: raid set md0 active with 2 out of 2 mirrors Feb 28 10:40:23 atalaska kernel: md: syncing RAID array md0 Feb 28 10:40:23 atalaska kernel: md: minimum _guaranteed_ reconstruction speed: 100 KB/sec. Feb 28 10:40:23 atalaska kernel: md: using maximum available idle IO bandwith for reconstruction. Feb 28 10:40:23 atalaska kernel: md: using 128k window. Feb 28 10:42:23 atalaska kernel: md: md0: sync done. Sync working fine, but for example copying a file to the array and... Feb 28 10:42:53 atalaska kernel: attempt to access beyond end of device Feb 28 10:42:53 atalaska kernel: 03:04: rw=0, want=155659, limit=152617 Feb 28 10:42:53 atalaska kernel: Operation continuing on 1 devices Feb 28 10:42:53 atalaska kernel: md: recovery thread got woken up ... Feb 28 10:42:53 atalaska kernel: md: recovery thread finished ... Feb 28 10:42:53 atalaska kernel: dirty sb detected, updating. We lost one disk in the mirror so we stop the md device and have a coffee ;) Feb 28 10:46:52 atalaska kernel: interrupting MD-thread pid 319 Feb 28 10:46:52 atalaska kernel: raid1d(319) flushing signals. Feb 28 10:46:52 atalaska kernel: interrupting MD-thread pid 320 Feb 28 10:46:52 atalaska kernel: raid1syncd(320) flushing signals. Feb 28 10:46:52 atalaska kernel: marking sb clean... Feb 28 10:46:52 atalaska kernel: unbind<hda4,1> Feb 28 10:46:52 atalaska kernel: export_rdev(hda4) Feb 28 10:46:52 atalaska kernel: unbind<hda2,0> Feb 28 10:46:52 atalaska kernel: export_rdev(hda2) Feb 28 10:46:52 atalaska kernel: md0 stopped. There is also a minor problem with RAID1 and persistent-superblock=0, playing a little I set up an only partition RAID1 array, perhaps to add other partitions later, well, when you start the array and cat /proc/mdstat you could see 0% reconstruction finished for ever and time to end the sync growing up for ever too, this isn't much important but the md threads start to get CPU time and the system take a great slow down. All the tests with the lastest raidtools and raid patch for kernel 2.2.0 running 2.2.1 Well that's all for now }:-) Any suggest would be appreciatted. Raimundo Alba _______________________________ Departamento de Sistemas 957-761015 957-413781 Alcantara Internet S.L. ALCAVIA [EMAIL PROTECTED] http://personales.alcavia.net/~raimon _______________________________
