Hi Robert, On Wed, 2015-05-06 at 12:21 -0400, Robert Wood wrote: > One update: it appears that Vmware ESXi 5.5 U2 is sometimes > incorrectly sensing that the LIO-ORG device is an SSD. I wonder if > that is causing issues with commands being sent? I am testing > untagging all LIO-ORG devices as SSD to see if the problem recurs. > > Changing around SCSI target feature bits reported by LIO to ESX hosts is only going to mask the real issue. >From Mike's testing earlier in the week, he's been able to verify with FC and iSCSI that it's not a target specific bug, but AFAICT it's still not clear if something specific to iscsi-target may be triggering a bug in ceph client. AFAICT, the bug is centered around ceph client reconnect logic + read-copy-update (RCU) usage in v3.19.y code. (Adding Sage CC') > > On Wed, May 6, 2015 at 11:17 AM, Robert Wood <rwood@xxxxxxxxxxxxxxxxxxx> wrote: > > Good morning, we are continuing to receive: > > > > May 6 11:08:26 roc-4r-scd212 kernel: [71898.566185] ABORT_TASK: > > ref_tag: 3847728 already complete, skipping > > May 6 11:08:26 roc-4r-scd212 kernel: [71898.566187] ABORT_TASK: > > Sending TMR_TASK_DOES_NOT_EXIST for ref_tag: 3847728 > > May 6 11:08:26 roc-4r-scd212 kernel: [71898.566191] Unable to locate > > ITT: 0x003ab632 on CID: 0 > > May 6 11:08:26 roc-4r-scd212 kernel: [71898.566191] Unable to locate > > RefTaskTag: 0x003ab632 on CID: 0. > > May 6 11:08:26 roc-4r-scd212 kernel: [71898.566254] Unexpected ret: > > -32 send data 48 > > May 6 11:08:30 roc-4r-scd212 kernel: [71902.397053] libceph: osd20 > > 10.80.3.25:6812 socket closed (con state OPEN) > -- Thanks for reporting! --nab -- To unsubscribe from this list: send the line "unsubscribe target-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
