As James Smart suggested, you can try to decrease the HBA driver nodev timeout value to 1 sec. I have done the cable unplug / plug test regularly and it work reliably for me. I'll try the controler restart test tomorow. regards, cvaroqui Le lundi 13 d?embre 2004 �1:24 +0100, Nicola Ranaldo a �it : > > Indeed, > > can you audit your fixes in > > http://christophe.varoqui.free.fr/multipath-tools/multipath-tools-0.4.0.tar.bz2 > > before I release it ? > > Ok, now the tools does not segs, but the last check i have to do is about > the clone syscall, on my system (slackware 10.0) i have to use fork in > order to have multipathd daemons > run. > While using clone strace multipathd gives: > > brk(0) = 0x8051000 > brk(0x8052000) = 0x8052000 > brk(0) = 0x8052000 > brk(0) = 0x8052000 > brk(0x8056000) = 0x8056000 > clone(child_stack=0x8055040, flags=CLONE_NEWNS) = 2443 > exit_group(0) = ? > > and the process dies... > it's the clone call necessary? does the process run properly even if i use > fork? > > > ... and report on general behaviour. > > Ok, some progress is done :))) > > Failover initiated by an "sg_start /dev/sgx 1" works properly! and i can do > a lot of switches between active and ghost path, with a 1/2 second delay > between each other, with no process disruption! great :) > > howewer a failover initiated by a "restart other" on the hsg80 console > gives: > > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 655327 > Dec 13 11:13:55 m3 kernel: SCSI error : <0 0 1 2> return code = 0x20000 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 656343 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 656351 > Dec 13 11:13:55 m3 kernel: SCSI error : <0 0 1 2> return code = 0x20000 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 657367 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 657375 > Dec 13 11:13:55 m3 kernel: SCSI error : <0 0 1 2> return code = 0x20000 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 658391 > Dec 13 11:13:55 m3 kernel: SCSI error : <0 0 1 2> return code = 0x20000 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 658399 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 658407 > Dec 13 11:13:55 m3 kernel: SCSI error : <0 0 1 2> return code = 0x20000 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 658903 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 658911 > Dec 13 11:13:55 m3 kernel: SCSI error : <0 0 1 2> return code = 0x20000 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 659927 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 659935 > Dec 13 11:13:55 m3 kernel: SCSI error : <0 0 1 2> return code = 0x20000 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 660951 > Dec 13 11:13:55 m3 kernel: SCSI error : <0 0 1 2> return code = 0x20000 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 660959 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 660967 > Dec 13 11:13:55 m3 kernel: SCSI error : <0 0 1 2> return code = 0x20000 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 661983 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 661991 > Dec 13 11:13:55 m3 kernel: SCSI error : <0 0 1 2> return code = 0x20000 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 663007 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 663015 > Dec 13 11:13:55 m3 kernel: SCSI error : <0 0 1 2> return code = 0x20000 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 664031 > Dec 13 11:13:55 m3 kernel: SCSI error : <0 0 1 2> return code = 0x20000 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 664039 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 664047 > Dec 13 11:13:55 m3 kernel: SCSI error : <0 0 1 2> return code = 0x20000 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 664791 > Dec 13 11:13:55 m3 kernel: end_request: I/O error, dev sdb, sector 664799 > Dec 13 11:13:55 m3 kernel: Incorrect number of segments after building list > Dec 13 11:13:55 m3 kernel: counted 8, received 1 > Dec 13 11:13:55 m3 kernel: req nr_sec 1024, cur_nr_sec 8 > Dec 13 11:13:55 m3 kernel: Buffer I/O error on device dm-1, logical block > 81523 > Dec 13 11:13:55 m3 kernel: lost page write due to I/O error on dm-1 > Dec 13 11:13:55 m3 kernel: Buffer I/O error on device dm-1, logical block > 81524 > Dec 13 11:13:55 m3 kernel: lost page write due to I/O error on dm-1 > Dec 13 11:13:55 m3 kernel: Buffer I/O error on device dm-1, logical block > 81525 > Dec 13 11:13:55 m3 kernel: lost page write due to I/O error on dm-1 > Dec 13 11:13:55 m3 kernel: Buffer I/O error on device dm-1, logical block > 81526 > Dec 13 11:13:55 m3 kernel: lost page write due to I/O error on dm-1 > Dec 13 11:13:55 m3 kernel: Buffer I/O error on device dm-1, logical block > 81527 > Dec 13 11:13:55 m3 kernel: lost page write due to I/O error on dm-1 > Dec 13 11:13:55 m3 kernel: Buffer I/O error on device dm-1, logical block > 81528 > Dec 13 11:13:55 m3 kernel: lost page write due to I/O error on dm-1 > Dec 13 11:13:55 m3 kernel: Buffer I/O error on device dm-1, logical block > 81529 > Dec 13 11:13:55 m3 kernel: lost page write due to I/O error on dm-1 > Dec 13 11:13:55 m3 kernel: Buffer I/O error on device dm-1, logical block > 81530 > Dec 13 11:13:55 m3 kernel: lost page write due to I/O error on dm-1 > Dec 13 11:13:55 m3 kernel: Buffer I/O error on device dm-1, logical block > 81531 > Dec 13 11:13:55 m3 kernel: lost page write due to I/O error on dm-1 > Dec 13 11:13:55 m3 kernel: Buffer I/O error on device dm-1, logical block > 81532 > Dec 13 11:13:55 m3 kernel: lost page write due to I/O error on dm-1 > Dec 13 11:13:55 m3 kernel: Incorrect number of segments after building list > Dec 13 11:13:55 m3 kernel: counted 8, received 1 > Dec 13 11:13:55 m3 kernel: req nr_sec 1024, cur_nr_sec 8 > Dec 13 11:14:06 m3 kernel: SCSI error : <0 0 1 2> return code = 0x10000 > Dec 13 11:14:06 m3 multipathd: 8:16 : tur checker reports path is down > Dec 13 11:14:06 m3 kernel: SCSI error : <0 0 1 2> return code = 0x10000 > Dec 13 11:14:06 m3 last message repeated 4 times > Dec 13 11:14:06 m3 multipathd: event checker startup : disk1 > Dec 13 11:14:16 m3 multipathd: 8:0 : tur checker reports path is up > Dec 13 11:14:18 m3 kernel: SCSI error : <0 0 1 2> return code = 0x10000 > Dec 13 11:14:18 m3 last message repeated 2 times > Dec 13 11:14:42 m3 kernel: Incorrect number of segments after building list > Dec 13 11:14:42 m3 kernel: counted 8, received 1 > Dec 13 11:14:42 m3 kernel: req nr_sec 1024, cur_nr_sec 8 > Dec 13 11:14:42 m3 multipathd: devmap event on disk1 > Dec 13 11:14:42 m3 kernel: Incorrect number of segments after building list > Dec 13 11:14:42 m3 kernel: counted 8, received 1 > Dec 13 11:14:42 m3 kernel: req nr_sec 1024, cur_nr_sec 8 > Dec 13 11:14:42 m3 kernel: Incorrect number of segments after building list > Dec 13 11:14:42 m3 kernel: counted 8, received 1 > Dec 13 11:14:42 m3 kernel: req nr_sec 1024, cur_nr_sec 8 > Dec 13 11:14:44 m3 kernel: SCSI error : <0 0 1 2> return code = 0x10000 > Dec 13 11:14:44 m3 last message repeated 2 times > Dec 13 11:14:44 m3 multipathd: event checker startup : disk1 > > after a long delay the random write operation (blocked due to the fail) > restarts! > > but in the log i have: > > Dec 13 11:15:43 m3 kernel: SCSI error : <0 0 1 2> return code = 0x10000 > Dec 13 11:16:16 m3 last message repeated 12 times > Dec 13 11:16:44 m3 last message repeated 17 times > > and multipath -l -v3 gives > 0:0:1:2: sg_io failed status 0x0 0x1 0x0 0x0 > 0:0:1:2: Unable to get INQUIRY vpd 1 page 0x0. > disk1 (360001fe1001613800009205005470164) > [size=33 GB][features="0"][hwhandler="0"] > \_ round-robin 0 [active][first] > \_ 0:0:0:2 sda 8:0 [ready ][active] > > the second path is lose! > > to double check giving an sg_start on the lose path i get: > start_stop: Host_status=0x01 [DID_NO_CONNECT] > > all this without oops > > thanks > > Nicola Ranaldo > > -- > > dm-devel@xxxxxxxxxx > https://www.redhat.com/mailman/listinfo/dm-devel -- christophe varoqui <christophe.varoqui@xxxxxxx>