[Iscsitarget-devel] some more on "Abort Task (01) issued on ... (Function Complete)"
Marko Kobal
So, I'm using ESXi 5 on the initiator side and the latest stable version of
IET od the target side (on top of CentoOS 5.5 x64), software iSCSI on both
sides.
I've double checked that:
- there are no hw issues
- networking works as expected, no packet loss (I'm not using jumbo frames
neither Flow control)
On normal load, everything is OK, however, under heavy load I'm
experiencing; on target:
...
iscsi_trgt: Abort Task (01) issued on tid:1 lun:0 by sid:8731221840167424
(Function Complete)
last message repeated 11 times
iscsi_trgt: Abort Task (01) issued on tid:1 lun:0 by sid:9015995351761408
(Function Complete)
iscsi_trgt: Abort Task (01) issued on tid:1 lun:0 by sid:9860420281893376
(Function Complete)
last message repeated 32 times
iscsi_trgt: Abort Task (01) issued on tid:1 lun:0 by sid:9575646770299392
(Function Complete)
last message repeated 32 times
...
and on ESXi 5:
...
nmp_PathDetermineFailure:2084: SCSI cmd RESERVE failed on path
vmhba34:C1:T8:L0, reservation state on device
t10.94544500000000009637363796D20727F646D20353D20313 is unknown
...
Device t10.94544500000000009637363796D20727F646D20353D20313 performance has
deteriorated. I/O latency increased from average value of 11230 microseconds
to 236583 microseconds.
...
Device t10.94544500000000009637363796D20727F646D20353D20313 performance has
improved. I/O latency reduced from 45071 microseconds to 22559 microseconds.
...
Can anybody help me debug those issues? Thanks very much!
Kind regards, Marko Kobal