* resync on every raidstart
@ 2003-09-21 18:05 Dick Streefland
2003-09-23 6:47 ` Neil Brown
0 siblings, 1 reply; 5+ messages in thread
From: Dick Streefland @ 2003-09-21 18:05 UTC (permalink / raw)
To: linux-raid
Every time I stop a RAID1 device with raidstop, and start it again
with raidstart, it resyncs. Is this normal, or am I doing something
wrong? Here is an example with a RAID1 device on two loopback devices:
# cat /etc/raidtab
raiddev /dev/md0
raid-level 1
nr-raid-disks 2
nr-spare-disks 0
chunk-size 4
device /dev/loop1
raid-disk 0
device /dev/loop2
raid-disk 1
# mkraid /dev/md0
handling MD device /dev/md0
analyzing super-block
disk 0: /dev/loop1, 10240kB, raid superblock at 10176kB
disk 1: /dev/loop2, 10240kB, raid superblock at 10176kB
Sep 21 19:30:16 zaphod kernel: md: bind<loop1>
Sep 21 19:30:16 zaphod kernel: md: bind<loop2>
Sep 21 19:30:16 zaphod kernel: raid1: raid set md0 active with 2 out of 2 mirrors
Sep 21 19:30:16 zaphod kernel: md: syncing RAID array md0
Sep 21 19:30:16 zaphod kernel: md: minimum _guaranteed_ reconstruction speed: 1000 KB/sec/disc.
Sep 21 19:30:16 zaphod kernel: md: using maximum available idle IO bandwith (but not more than 200000 KB/sec) for reconstruction.
Sep 21 19:30:16 zaphod kernel: md: using 128k window, over a total of 10176 blocks.
Sep 21 19:30:17 zaphod kernel: md: md0: sync done.
# raidstop /dev/md0
Sep 21 19:30:31 zaphod kernel: md: md0 stopped.
Sep 21 19:30:31 zaphod kernel: md: unbind<loop2>
Sep 21 19:30:31 zaphod kernel: md: export_rdev(loop2)
Sep 21 19:30:31 zaphod kernel: md: unbind<loop1>
Sep 21 19:30:31 zaphod kernel: md: export_rdev(loop1)
# raidstart /dev/md0
Sep 21 19:30:46 zaphod kernel: md: autorun ...
Sep 21 19:30:46 zaphod kernel: md: considering loop2 ...
Sep 21 19:30:46 zaphod kernel: md: adding loop2 ...
Sep 21 19:30:46 zaphod kernel: md: adding loop1 ...
Sep 21 19:30:46 zaphod kernel: md: created md0
Sep 21 19:30:46 zaphod kernel: md: bind<loop1>
Sep 21 19:30:46 zaphod kernel: md: bind<loop2>
Sep 21 19:30:46 zaphod kernel: md: running: <loop2><loop1>
Sep 21 19:30:46 zaphod kernel: raid1: raid set md0 active with 2 out of 2 mirrors
Sep 21 19:30:46 zaphod kernel: md: ... autorun DONE.
Sep 21 19:30:46 zaphod kernel: md: syncing RAID array md0
Sep 21 19:30:46 zaphod kernel: md: minimum _guaranteed_ reconstruction speed: 1000 KB/sec/disc.
Sep 21 19:30:46 zaphod kernel: md: using maximum available idle IO bandwith (but not more than 200000 KB/sec) for reconstruction.
Sep 21 19:30:46 zaphod kernel: md: using 128k window, over a total of 10176 blocks.
Sep 21 19:30:47 zaphod kernel: md: md0: sync done.
I'm using raidtools2 version 1.00.3, and kernel 2.6.0-test5 with the
following configuration:
CONFIG_MD=y
CONFIG_BLK_DEV_MD=y
# CONFIG_MD_LINEAR is not set
# CONFIG_MD_RAID0 is not set
CONFIG_MD_RAID1=y
# CONFIG_MD_RAID5 is not set
# CONFIG_MD_MULTIPATH is not set
CONFIG_BLK_DEV_DM=y
# CONFIG_DM_IOCTL_V4 is not set
--
Dick Streefland //// De Bilt
dick.streefland@xs4all.nl (@ @) The Netherlands
------------------------------oOO--(_)--OOo------------------
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: resync on every raidstart
2003-09-21 18:05 resync on every raidstart Dick Streefland
@ 2003-09-23 6:47 ` Neil Brown
2003-09-23 20:43 ` Dick Streefland
0 siblings, 1 reply; 5+ messages in thread
From: Neil Brown @ 2003-09-23 6:47 UTC (permalink / raw)
To: Dick Streefland; +Cc: linux-raid
On Sunday September 21, dick.streefland@xs4all.nl wrote:
> Every time I stop a RAID1 device with raidstop, and start it again
> with raidstart, it resyncs. Is this normal, or am I doing something
> wrong? Here is an example with a RAID1 device on two loopback
devices:
I cannot reproduce this.
Does it happen with real devices, or only loopback devices?
Can you try with mdadm and see if that makes a difference?
After raidstop, and before raidstart, can you
mdadm -E /dev/loop1
mdadm -E /dev/loop2
and show me the results?
Thanks,
NeilBrown
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: resync on every raidstart
2003-09-23 6:47 ` Neil Brown
@ 2003-09-23 20:43 ` Dick Streefland
2003-09-27 14:52 ` Dick Streefland
0 siblings, 1 reply; 5+ messages in thread
From: Dick Streefland @ 2003-09-23 20:43 UTC (permalink / raw)
To: linux-raid
Neil Brown <neilb@cse.unsw.edu.au> wrote:
| I cannot reproduce this.
| Does it happen with real devices, or only loopback devices?
|
| Can you try with mdadm and see if that makes a difference?
|
| After raidstop, and before raidstart, can you
|
| mdadm -E /dev/loop1
| mdadm -E /dev/loop2
|
| and show me the results?
This is the output of mdadm-1.3.0:
# mdadm -E /dev/loop1
/dev/loop1:
Magic : a92b4efc
Version : 00.90.00
UUID : 44f2804a:f7890a34:bd644666:becddad4
Creation Time : Tue Sep 23 21:08:17 2003
Raid Level : raid1
Device Size : 10176 (9.94 MiB 10.42 MB)
Raid Devices : 2
Total Devices : 2
Preferred Minor : 0
Update Time : Tue Sep 23 21:08:33 2003
State : dirty, no-errors
Active Devices : 2
Working Devices : 2
Failed Devices : 0
Spare Devices : 0
Checksum : e0ba675d - correct
Events : 0.2
Number Major Minor RaidDevice State
this 0 7 1 0 active sync /dev/loop1
0 0 7 1 0 active sync /dev/loop1
1 1 7 2 1 active sync /dev/loop2
# mdadm -E /dev/loop2
/dev/loop2:
Magic : a92b4efc
Version : 00.90.00
UUID : 44f2804a:f7890a34:bd644666:becddad4
Creation Time : Tue Sep 23 21:08:17 2003
Raid Level : raid1
Device Size : 10176 (9.94 MiB 10.42 MB)
Raid Devices : 2
Total Devices : 2
Preferred Minor : 0
Update Time : Tue Sep 23 21:08:33 2003
State : dirty, no-errors
Active Devices : 2
Working Devices : 2
Failed Devices : 0
Spare Devices : 0
Checksum : e0ba6760 - correct
Events : 0.2
Number Major Minor RaidDevice State
this 1 7 2 1 active sync /dev/loop2
0 0 7 1 0 active sync /dev/loop1
1 1 7 2 1 active sync /dev/loop2
I guess the "dirty" state is the problem?
I also created a RAID1 array on another machine (2.6.0-test5 kernel,
raidtools-1.00.3, mdadm-0.7.2) with real disks:
# mkraid /dev/md3
handling MD device /dev/md3
analyzing super-block
disk 0: /dev/hda3, 535752kB, raid superblock at 535680kB
disk 1: /dev/hdb3, 535752kB, raid superblock at 535680kB
... waiting for sync completion ...
# mdadm -E /dev/hda3
/dev/hda3:
Magic : a92b4efc
Version : 00.90.00
UUID : 8e85542d:e073310b:b5478b9a:67ea873f
Creation Time : Tue Sep 23 21:38:39 2003
Raid Level : raid1
Device Size : 535680 (523.12 MiB 548.53 MB)
Raid Disks : 2
Total Disks : 2
Preferred Minor : 3
Update Time : Tue Sep 23 21:49:36 2003
State : dirty, no-errors
Active Drives : 2
Working Drives : 2
Failed Drives : 0
Spare Drives : 0
Checksum : b43f696d - correct
Events : 0.1
Number Major Minor RaidDisk State
this 0 3 3 0 active sync /dev/hda3
0 0 3 3 0 active sync /dev/hda3
1 1 3 67 1 active sync /dev/hdb3
# raidstop /dev/md3
# mdadm -E /dev/hda3
/dev/hda3:
Magic : a92b4efc
Version : 00.90.00
UUID : 8e85542d:e073310b:b5478b9a:67ea873f
Creation Time : Tue Sep 23 21:38:39 2003
Raid Level : raid1
Device Size : 535680 (523.12 MiB 548.53 MB)
Raid Disks : 2
Total Disks : 2
Preferred Minor : 3
Update Time : Tue Sep 23 21:50:45 2003
State : dirty, no-errors
Active Drives : 2
Working Drives : 2
Failed Drives : 0
Spare Drives : 0
Checksum : b43f69b4 - correct
Events : 0.2
Number Major Minor RaidDisk State
this 0 3 3 0 active sync /dev/hda3
0 0 3 3 0 active sync /dev/hda3
1 1 3 67 1 active sync /dev/hdb3
A raidstart resulted in a second resync, but after that, it is clean:
# mdadm -E /dev/hda3
/dev/hda3:
Magic : a92b4efc
Version : 00.90.00
UUID : 8e85542d:e073310b:b5478b9a:67ea873f
Creation Time : Tue Sep 23 21:38:39 2003
Raid Level : raid1
Device Size : 535680 (523.12 MiB 548.53 MB)
Raid Disks : 2
Total Disks : 2
Preferred Minor : 3
Update Time : Tue Sep 23 22:20:08 2003
State : clean, no-errors
Active Drives : 2
Working Drives : 2
Failed Drives : 0
Spare Drives : 0
Checksum : b43f709c - correct
Events : 0.4
Number Major Minor RaidDisk State
this 0 3 3 0 active sync /dev/hda3
0 0 3 3 0 active sync /dev/hda3
1 1 3 67 1 active sync /dev/hdb3
However, the RAID1 array on the loopback devices stays "dirty", even
after multiple raidstop/raidstart commands.
--
Dick Streefland //// De Bilt
dick.streefland@xs4all.nl (@ @) The Netherlands
------------------------------oOO--(_)--OOo------------------
^ permalink raw reply [flat|nested] 5+ messages in thread* Re: resync on every raidstart
2003-09-23 20:43 ` Dick Streefland
@ 2003-09-27 14:52 ` Dick Streefland
2003-09-27 21:25 ` Dick Streefland
0 siblings, 1 reply; 5+ messages in thread
From: Dick Streefland @ 2003-09-27 14:52 UTC (permalink / raw)
To: linux-raid
I did some more tests, and it looks like my problem may be related to
SMP. When I setup a RAID1 array on my SMP machine, the component
devices stay "dirty", but when I do the same on an UP machine, the
state is "clean". However, when I boot the SMP machine with "nosmp",
or boot a non-SMP kernel, the problem persists. Both machines run a
2.6.0-test5 kernel (with your oops fix from 2003-09-12), and have
mdadm-1.3.0.
This is the test script:
#!/bin/sh
dd bs=1024k count=10 if=/dev/zero of=/tmp/img1
dd bs=1024k count=10 if=/dev/zero of=/tmp/img2
losetup /dev/loop1 /tmp/img1
losetup /dev/loop2 /tmp/img2
mdadm -C -n 2 -l 1 /dev/md1 /dev/loop1 /dev/loop2
while grep resync /proc/mdstat
do sleep 1
done
echo done
sleep 3
mdadm -E /dev/loop1
mdadm -S /dev/md1
losetup -d /dev/loop1
losetup -d /dev/loop2
rm /tmp/img1
rm /tmp/img2
This is the output on the UP machine:
[======>..............] resync = 30.0% (3584/10176) finish=0.1min speed=716K/sec
done
/dev/loop1:
Magic : a92b4efc
Version : 00.90.00
UUID : 6553bca6:9c4152b1:960d65fc:75de3d79
Creation Time : Sat Sep 27 16:23:58 2003
Raid Level : raid1
Device Size : 10176 (9.94 MiB 10.42 MB)
Raid Devices : 2
Total Devices : 2
Preferred Minor : 1
Update Time : Sat Sep 27 16:24:11 2003
State : clean, no-errors
Active Devices : 2
Working Devices : 2
Failed Devices : 0
Spare Devices : 0
Checksum : 35976528 - correct
Events : 0.1
Number Major Minor RaidDevice State
this 0 7 1 0 active sync /dev/loop1
0 0 7 1 0 active sync /dev/loop1
1 1 7 2 1 active sync /dev/loop2
This is the output on the SMP machine:
[>....................] resync = 0.0% (0/10176) finish=1.6min speed=0K/sec
done
/dev/loop1:
Magic : a92b4efc
Version : 00.90.00
UUID : dc1e7844:0571f448:75ec8ac0:df235bed
Creation Time : Sat Sep 27 16:23:18 2003
Raid Level : raid1
Device Size : 10176 (9.94 MiB 10.42 MB)
Raid Devices : 2
Total Devices : 2
Preferred Minor : 1
Update Time : Sat Sep 27 16:23:19 2003
State : dirty, no-errors
Active Devices : 2
Working Devices : 2
Failed Devices : 0
Spare Devices : 0
Checksum : 5eb70538 - correct
Events : 0.1
Number Major Minor RaidDevice State
this 0 7 1 0 active sync /dev/loop1
0 0 7 1 0 active sync /dev/loop1
1 1 7 2 1 active sync /dev/loop2
--
Dick Streefland //// De Bilt
dick.streefland@xs4all.nl (@ @) The Netherlands
------------------------------oOO--(_)--OOo------------------
^ permalink raw reply [flat|nested] 5+ messages in thread* Re: resync on every raidstart
2003-09-27 14:52 ` Dick Streefland
@ 2003-09-27 21:25 ` Dick Streefland
0 siblings, 0 replies; 5+ messages in thread
From: Dick Streefland @ 2003-09-27 21:25 UTC (permalink / raw)
To: linux-raid
spam@streefland.xs4all.nl (Dick Streefland) wrote:
| I did some more tests, and it looks like my problem may be related to
| SMP. When I setup a RAID1 array on my SMP machine, the component
| devices stay "dirty", but when I do the same on an UP machine, the
| state is "clean". However, when I boot the SMP machine with "nosmp",
| or boot a non-SMP kernel, the problem persists. Both machines run a
| 2.6.0-test5 kernel (with your oops fix from 2003-09-12), and have
| mdadm-1.3.0.
It turns out that the problem is not related to SMP, but to
CONFIG_PREEMPT. I built two identical kernels for a non-SMP machine,
except for CONFIG_PREEMPT. With the CONFIG_PREEMPT kernel, the state
is dirty, with the other kernel, the state is clean.
--
Dick Streefland //// De Bilt
dick.streefland@xs4all.nl (@ @) The Netherlands
------------------------------oOO--(_)--OOo------------------
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2003-09-27 21:25 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2003-09-21 18:05 resync on every raidstart Dick Streefland
2003-09-23 6:47 ` Neil Brown
2003-09-23 20:43 ` Dick Streefland
2003-09-27 14:52 ` Dick Streefland
2003-09-27 21:25 ` Dick Streefland
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).