Am 21.04.23 um 16:59 schrieb Dr. Thorsten Brandau:
PS: Despooling seems unusually slow to me.
PPS: Please do not top-post.
I have noticed that both, spooling and despooling is very slow. I have
no idea why, the disc makes easily 300 MB/s and there is nothing else
going on on that raid.
I have no idea where to even look. When I start parallel other
processes, they can use the full power. iotop goes up to 300 MB/s from
time to time to drop then again to zero.
On another server where I go directly disc to tape it is faster (LTO-6).
Is your tape block size set to zero?
mt -f /dev/nst0 status | grep size
mt -f /dev/nst0 defblksize 0
mt -f /dev/nst0 setblk 0
Grab an empty tape and make some test:
Nothing will be faster than this:
# ddrescue -s 128G --force /dev/zero /dev/null
GNU ddrescue 1.21
Press Ctrl-C to interrupt
ipos: 127999 MB, non-trimmed: 0 B, current rate: 7029 MB/s
opos: 127999 MB, non-scraped: 0 B, average rate: 11636 MB/s
non-tried: 0 B, errsize: 0 B, run time: 11s
rescued: 128000 MB, errors: 0, remaining time: n/a
percent rescued: 100.00% time since last successful read: 0s
Finished
How fast can we write to the disk (ever)?
# ddrescue -D -s 128G --force /dev/zero test.bin
GNU ddrescue 1.21
Press Ctrl-C to interrupt
ipos: 127999 MB, non-trimmed: 0 B, current rate: 228 MB/s
opos: 127999 MB, non-scraped: 0 B, average rate: 727 MB/s
non-tried: 0 B, errsize: 0 B, run time: 2m 56s
rescued: 128000 MB, errors: 0, remaining time: n/a
percent rescued: 100.00% time since last successful read: 0s
Finished
How fast can we read from the disk (ever)?
# ddrescue -d --force test.bin /dev/null
GNU ddrescue 1.21
Press Ctrl-C to interrupt
ipos: 134217 MB, non-trimmed: 0 B, current rate: 941 MB/s
opos: 134217 MB, non-scraped: 0 B, average rate: 986 MB/s
non-tried: 0 B, errsize: 0 B, run time: 2m 16s
rescued: 134217 MB, errors: 0, remaining time: n/a
percent rescued: 100.00% time since last successful read: 0s
Finished
Lets generate a binary random file which is merely incompressible
# ddrescue -s 128G --force /dev/urandom test.bin
GNU ddrescue 1.21
Press Ctrl-C to interrupt
ipos: 127999 MB, non-trimmed: 0 B, current rate: 43843 kB/s
opos: 127999 MB, non-scraped: 0 B, average rate: 171 MB/s
non-tried: 0 B, errsize: 0 B, run time: 12m 27s
rescued: 128000 MB, errors: 0, remaining time: n/a
percent rescued: 100.00% time since last successful read: 0s
Finished
# ddrescue -d --force test.bin /dev/nst0
GNU ddrescue 1.21
Press Ctrl-C to interrupt
ipos: 127999 MB, non-trimmed: 0 B, current rate: 40435 kB/s
opos: 127999 MB, non-scraped: 0 B, average rate: 98497 kB/s
non-tried: 0 MB, errsize: 0 B, run time: 21m 36s
rescued: 128000 MB, errors: 0, remaining time: n/a
percent rescued: 100.00% time since last successful read: 0s
Finished
This is from our Quantum Superloader 3 LTO-7 definitions. See if it
makes any difference for you. Despooling sustained ~ 290MB/sec from an
SSD Raid 1.
bacula-sd.conf:
Device {
...
MaximumSpoolSize = 256G
# 300 Mb/sec = eof filemark every two minutes
MaximumFileSize = 36G
MaximumBlockSize = 1048576
...
}
_______________________________________________
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users
