On Mon, 2016-02-08 at 00:25 -0500, Dan Lane wrote: > Sorry, I've had a super busy two weeks, I've been out of town for work > the entire time :-/ You're absolutely right though, I missed that > step, I'll try again tonight. > Btw, please avoid top-posting to list emails, as it makes the thread more difficult to read. ;) Just to clarify a bit about what's going on with your setup: The system is triggering repeated ESX FC host side ABORT_TASKs at idle and initial load, and continue to generate ABORT_TASKs until triggering a TMR ABORT_TASK refcount bug. That's one specific issue. Looking further at the log however, the first few ABORT_TASKs processed would indicate I/Os have successfully reached CMD_T_COMPLETE state: Feb 7 20:30:48 dracofiler kernel: qla2xxx [0000:82:00.1]-500b:10: LOOP DOWN detected (2 5 0 0). Feb 7 20:30:51 dracofiler kernel: qla2xxx [0000:82:00.0]-500a:9: LOOP UP detected (4 Gbps). Feb 7 20:52:00 dracofiler kernel: qla2xxx [0000:82:00.1]-500a:10: LOOP UP detected (4 Gbps). Feb 7 20:52:00 dracofiler kernel: qla2xxx [0000:82:00.1]-0121:10: Failed to enable receiving of RSCN requests: 0x2. Feb 7 20:52:19 dracofiler kernel: MODE SENSE: unimplemented page/subpage: 0x1c/0x02 Feb 7 21:08:25 dracofiler kernel: ABORT_TASK: Found referenced qla2xxx task_tag: 1153732 Feb 7 21:08:25 dracofiler kernel: ABORT_TASK: ref_tag: 1153732 already complete, skipping Feb 7 21:08:25 dracofiler kernel: ABORT_TASK: Sending TMR_TASK_DOES_NOT_EXIST for ref_tag: 1153732 Feb 7 21:08:25 dracofiler kernel: ABORT_TASK: Found referenced qla2xxx task_tag: 1153776 Feb 7 21:08:25 dracofiler kernel: ABORT_TASK: ref_tag: 1153776 already complete, skipping Feb 7 21:08:25 dracofiler kernel: ABORT_TASK: Sending TMR_TASK_DOES_NOT_EXIST for ref_tag: 1153776 Feb 7 21:09:20 dracofiler kernel: ABORT_TASK: Found referenced qla2xxx task_tag: 1144580 Feb 7 21:09:20 dracofiler kernel: ABORT_TASK: ref_tag: 1144580 already complete, skipping Feb 7 21:09:20 dracofiler kernel: ABORT_TASK: Sending TMR_TASK_DOES_NOT_EXIST for ref_tag: 1144580 Feb 7 21:09:21 dracofiler kernel: ABORT_TASK: Found referenced qla2xxx task_tag: 1200548 Feb 7 21:09:29 dracofiler kernel: ------------[ cut here ]------------ Feb 7 21:09:29 dracofiler kernel: kernel BUG at drivers/scsi/qla2xxx/qla_target.c:3099! So it seems the referenced I/O tags where successfully posted into tcm_qla2xxx code, but it appears that not all FC response packets are successfully making their way back to your ESX FC host. -- 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
