I tried again, setting up RAID1 on 2 U320 drives, 15k, as described in
softraid(4).
Now I find the speed to be too slow. Writing to a single file is kind of
okay: [everything/pwd is /mnt, which is a softraid drive, /dev/sd3f]
# bioctl sd3
Volume Status Size Device
softraid0 0 Online 299671585280 sd3 RAID1
0 Online 299671585280 0:0.0 noencl
1 Online 299671585280 0:1.0 noencl
dump and restore is the task. It is not fast:
DUMP: Volume 1 took 0:00:07
DUMP: Volume 1 transfer rate: 2147 KB/s
DUMP: Date this dump completed: Wed May 20 16:31:08 2009
DUMP: Average transfer rate: 2147 KB/s
7 seconds for 14 MB. But data transfer itself is okay:
# dump -0ua -f testo /dev/sd0e
DUMP: Volume 1 took 0:00:01
DUMP: Volume 1 transfer rate: 15039 KB/s
DUMP: Date this dump completed: Wed May 20 16:49:53 2009
DUMP: Average transfer rate: 15039 KB/s
DUMP: level 0 dump on Wed May 20 16:49:51 2009
It is writing that takes the time:
# date && restore rf testo && date
Wed May 20 16:51:48 SGT 2009
Wed May 20 16:51:54 SGT 2009
The raw speed is good:
# dd if=/dev/zero of=nonsense.img bs=1m count=5000
5000+0 records in
5000+0 records out
524288 bytes transferred in 100.534 secs (52149868 bytes/sec)
But a dump && restore of /usr is a tad sick:
(/dev/sd0f 7.9G2.4G5.1G32%/usr)
# dump -0ua -f - /dev/sd0f | restore rf -
DUMP: Date of this level 0 dump: Wed May 20 16:53:46 2009
DUMP: Date of last level 0 dump: the epoch
DUMP: Dumping /dev/rsd0f (/usr) to standard output
DUMP: mapping (Pass I) [regular files]
DUMP: mapping (Pass II) [directories]
DUMP: estimated 2549189 tape blocks.
DUMP: Volume 1 started at: Wed May 20 16:53:48 2009
DUMP: dumping (Pass III) [directories]
DUMP: dumping (Pass IV) [regular files]
DUMP: 4.42% done, finished in 3:48
DUMP: 36.44% done, finished in 0:27
DUMP: 40.42% done, finished in 0:30
DUMP: 52.60% done, finished in 0:23
DUMP: 64.08% done, finished in 0:17
DUMP: 77.57% done, finished in 0:10
DUMP: 92.19% done, finished in 0:03
DUMP: 2717062 tape blocks
DUMP: Date of this level 0 dump: Wed May 20 16:53:46 2009
DUMP: Volume 1 completed at: Wed May 20 17:36:48 2009
DUMP: Volume 1 took 0:43:00
DUMP: Volume 1 transfer rate: 1053 KB/s
DUMP: Date this dump completed: Wed May 20 17:36:48 2009
DUMP: Average transfer rate: 1053 KB/s
DUMP: level 0 dump on Wed May 20 16:53:46 2009
DUMP: DUMP IS DONE
The LEDs of the drives were kind of continuously on.
I also tried to mount 'softdep', but that didn't make much of a
difference. When I do 'df -h' in another console, I can see at times
that the data amount transfered is huge, at other times it is moving by
steps of 0.1-0.2 MB/s. Probably it is a problem of number of files, not
of size.
Any idea what to do to improve the performance?
Uwe
