[dm-devel] multipath, Engenio 6998 and Volume Snapshot
aaron
aaron at sysdev.oucs.ox.ac.uk
Tue Jan 17 15:53:30 UTC 2006
(or depending on your choice of branding an IBM DS4800 (flashcopy),
SGI TP9700 (SnapCopy) and no doubt lots of other brandings)
I'm also having a problem using multipath, the DS4800 and the DS4800's
inbuilt flashcopy feature. I'd like to use the DS4800's inbuilt
flashcopy feature becuase (i) it is there and (ii) it will push the
load of backing up from the indiviual SAN attached machines to the
DS4800 and backup servers.
The problem is as follows. There are 8 volumes mounted on server A,
which is Debain/Sarge system running a 2.6.15 kernel with multipath
tools built from the git tree on 2005-12-14. If I try to create a
snapshot on the DS4800, not involving any of the mounted volumes, and
there is some disk IOi (*) on one of the mounted volumes then the
following occurs:
1) There are messages like:
multipathd: 65:160: tur checker reports path is down
multipathd: checker failed path 65:160 in map imap225-x4
kernel: device-mapper: dm-multipath: Failing path 65:160.
multipathd: imap225-x4: remaining active paths: 1
multipathd: 8:240: tur checker reports path is down
multipathd: checker failed path 8:240 in map imap224-x4
kernel: device-mapper: dm-multipath: Failing path 8:240.
multipathd: imap224-x4: remaining active paths: 1
multipathd: 65:160: tur checker reports path is up
multipathd: 65:160: reinstated 3370:Jan 16 18:41:58 imap224 multipathd: imap225-x4: remaining active paths: 2
multipathd: 8:240: tur checker reports path is up
multipathd: 8:240: reinstated
multipathd: imap224-x4: remaining active paths: 2
in syslog for each of the mounted volumes.
2) I've seen errors like there are SCSI errors for the volume with disk IO:
kernel: sd 2:0:0:6: rejecting I/O to offline device
kernel: sd 2:0:0:6: SCSI error: return code = 0x10000
kernel: end_request: I/O error, dev sdw, sector 60402552
(with numberous repititions) and I've also seen errors like:
kernel: qla2300 0000:04:02.0: scsi(1:0:6): Abort command issued -- 188fa3 2002. Abort command issued -- 188fbe 2002.
kernel: qla2300 0000:04:02.0: scsi(1:0:6): Abort command issued -- 188fbf 2002.
kernel: qla2300 0000:04:02.0: scsi(1:0:6): Abort command issued -- 188fc0 2002.
which result in the IP for this disk stopping
The program generating IO is also unhappy, and gives errors like:
(3735) open clients/client42/~dmtmp/PARADOX/__414F2.DB succeeded for handle -1
(49517) ERROR: handle 13114 was not found
(62475) rmdir clients/client71/~dmtmp/ACCESS failed (Directory not empty)
(62476) rmdir clients/client71/~dmtmp failed (Directory not empty)
(62477) rmdir clients/client71 failed (Directory not empty)
which seems to be a bad thing.
The multipath configutation is attached.
Has anyone come accross this before and/or does any one have any hints
as to what (if anything) is going wrong ?
aaron
(*) My favourite form of disk IO at the moment is dbench 100 (where
dbench is part of the Samba benchmarking suite).
-------------- next part --------------
multipaths {
multipath {
wwid 3600a0b8000114d9a0000953d4396577c
alias imap224-x0
}
multipath {
wwid 3600a0b8000114e5a0000945c439feea3
alias imap224-x1
}
multipath {
wwid 3600a0b8000114d9a000095ef43a14230
alias imap224-x2
}
multipath {
wwid 3600a0b8000114e5a0000946d43a142f5
alias imap224-x3
}
multipath {
wwid 3600a0b8000114d9a000095f343a1440e
alias imap224-x4
}
multipath {
wwid 3600a0b8000114e5a0000947043a14483
alias imap224-x5
}
multipath {
wwid 3600a0b8000114d9a000095f543a1449a
alias imap224-x6
}
multipath {
wwid 3600a0b8000114e5a0000947143a144ff
alias imap224-x7
}
multipath {
wwid 3600a0b8000114d9a00009417438d7eaa
alias imap225-x0
}
multipath {
wwid 3600a0b8000114e5a00009332438d8135
alias imap225-x1
}
multipath {
wwid 3600a0b8000114d9a00009422438d8142
alias imap225-x2
}
multipath {
wwid 3600a0b8000114e5a00009333438d8185
alias imap225-x3
}
multipath {
wwid 3600a0b8000114d9a00009424438d8188
alias imap225-x4
}
multipath {
wwid 3600a0b8000114e5a00009334438d81c3
alias imap225-x5
}
multipath {
wwid 3600a0b8000114d9a00009426438d81c4
alias imap225-x6
}
multipath {
wwid 3600a0b8000114e5a00009335438d81fd
alias imap225-x7
}
}
devices {
device {
vendor "IBM"
product "1815 FAStT"
path_grouping_policy group_by_prio
prio_callout "/sbin/mpath_prio_tpc /dev/%n"
path_checker tur
}
}
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://listman.redhat.com/archives/dm-devel/attachments/20060117/4100ebb4/attachment.sig>
More information about the dm-devel
mailing list