Nicola, Detaching and re-attaching up to 3 out of 4 paths with in-flight IO works just fine for me using 2 Emulex HBAs to a CLARiiON AX150, a version 2.6.16-rc5 linux kernel, and a very recent multipath-tools.0.4.7 package. Is it possible for you to try with Emulex HBAs instead of Qlogic ones? Ed > -----Original Message----- > From: dm-devel-bounces@xxxxxxxxxx > [mailto:dm-devel-bounces@xxxxxxxxxx] On Behalf Of Nicola Murino > Sent: Tuesday, July 11, 2006 8:26 AM > To: device-mapper development > Subject: RE: dm-emc doesn't works with vanilla kernel > > Thanks for your reply, > > I'm currently using multipath-tools 0.4.7, the problem happen when I > manually detach a path with a kernel>2.6.13, > > > The system works fine with only an active path, here a re the logs: > > Jul 7 12:29:00 nodob qla2300 0000:0a:03.0: LIP reset occured (f8ef). > Jul 7 12:29:35 nodob rport-2:0-0: blocked FC remote port time out: > removing target > Jul 7 12:29:45 nodob scsi2 (0:1): rejecting I/O to dead device > Jul 7 12:29:45 nodob device-mapper: dm-multipath: Failing path 8:112. > Jul 7 12:29:45 nodob device-mapper: dm-emc: emc_pg_init: sending > switch-over command > > > but when I reattach the failed path the system is unable to react to > this event, I have this in my logs (kernel-2.6.14 and 2.6.16): > > Jul 7 12:30:05 nodob qla2300 0000:0a:03.0: LIP occured (f8ef). > Jul 7 12:30:06 nodob Vendor: DGC Model: RAID 5 Rev: > 0207 > Jul 7 12:30:06 nodob Type: Direct-Access ANSI > SCSI revision: 04 > Jul 7 12:30:06 nodob SCSI device sdl: 629145600 512-byte hdwr sectors > (322123 MB) > Jul 7 12:30:06 nodob sdl: asking for cache data failed > Jul 7 12:30:06 nodob sdl: assuming drive cache: write through > Jul 7 12:30:06 nodob SCSI device sdl: 629145600 512-byte hdwr sectors > (322123 MB) > Jul 7 12:30:06 nodob sdl: asking for cache data failed > Jul 7 12:30:06 nodob sdl: assuming drive cache: write through > Jul 7 12:30:06 nodob sdl:<6>sd 2:0:0:0: Device not ready. > Jul 7 12:30:06 nodob end_request: I/O error, dev sdl, sector 0 > Jul 7 12:30:06 nodob Buffer I/O error on device sdl, logical block 0 > Jul 7 12:30:06 nodob sd 2:0:0:0: Device not ready. > Jul 7 12:30:06 nodob end_request: I/O error, dev sdl, sector 0 > Jul 7 12:30:06 nodob Buffer I/O error on device sdl, logical block 0 > Jul 7 12:30:06 nodob unable to read partition table > Jul 7 12:30:06 nodob Attached scsi disk sdl at scsi2, > channel 0, id 0, > lun 0 > Jul 7 12:30:06 nodob Attached scsi generic sg7 at scsi2, > channel 0, id > 0, lun 0, type 0 > Jul 7 12:30:06 nodob Vendor: DGC Model: RAID 5 Rev: > 0207 > Jul 7 12:30:06 nodob Type: Direct-Access ANSI > SCSI revision: 04 > Jul 7 12:30:06 nodob error 1 > Jul 7 12:30:06 nodob scsi: Unexpected response from host 2 > channel 0 id > 0 lun 0 while scanning, scan aborted > > and I have to reboot the server to have a working system, > > with kernel-2.6.13 on the same system with the same udev and > multipath-tools all is fine, > > thanks > Nicola > > > Il giorno lun, 10/07/2006 alle 11.07 -0400, egoggin@xxxxxxx > ha scritto: > > I suspect that the problem is with multipath-tools not the > > dm-emc code. You may need to upgrade your multipath-tools > > package to a version 0.4.7 base. Could you try this and > > let me know the status? > > > > I have no problem of this sort with my CX series CLARiiONs > > with version 2.6.(13,14,15,16) linux kernels. > > > > > -----Original Message----- > > > From: dm-devel-bounces@xxxxxxxxxx > > > [mailto:dm-devel-bounces@xxxxxxxxxx] On Behalf Of Nicola Murino > > > Sent: Saturday, July 08, 2006 5:09 AM > > > To: dm-devel@xxxxxxxxxx > > > Subject: dm-emc doesn't works with vanilla kernel > > > > > > Hi all, > > > > > > I'm successfully running multipath-tools on a gentoo server > > > since almost > > > a year, only a problem: > > > > > > if I use a kernel different from 2.6.13 after detaching a > fibre path > > > dm-emc doesn't send trespass and the cx300 luns are not > accessible, I > > > got this problem with kernel-2.6.14 and kernel-2.6.16.19, > > > kernel-2.6.13 > > > works ok. > > > > > > On another installation with centos, last kernel > > > (2.6.9-34.01EL) and the > > > same storage all is fine. > > > > > > Would be great if dm-emc works properly with the upstream kernels, > > > > > > thanks > > > Nicola > > > > > > -- > > > > > > dm-devel@xxxxxxxxxx > > > https://www.redhat.com/mailman/listinfo/dm-devel > > > > > > > -- > > > > dm-devel@xxxxxxxxxx > > https://www.redhat.com/mailman/listinfo/dm-devel > > -- > > dm-devel@xxxxxxxxxx > https://www.redhat.com/mailman/listinfo/dm-devel > -- dm-devel@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/dm-devel
