Hi ,all big guys,I find many lock messages in vmkernel.log, In almost all esxi5.1 host.
I don't know if this impact our production . happend that one or two vm'vmdk changed to read-only ..I'm not sure if this "lock" or not.
Please look insight of this message ,and give me some advice,Thank you very much !
log lik this :
2013-12-10T22:23:47.374Z cpu22:13263202)DLX: 3661: vol 'xxxxxxx', lock at 75927552: [Req mode 1] Checking liveness:
gen 33, mode 1, owner 516444ef-b7b9240f-5805-xxxxxxxxmtime 6405918
2013-12-10T22:23:51.399Z cpu26:13263202)DLX: 4185: vol 'xxxxxx', lock at 75927552: [Req mode: 1] Not free:
gen 33, mode 1, owner 516444ef-b7b9240f-5805-xxxxxxxxmtime 6405918
2013-12-10T22:23:51.399Z cpu26:13263202)DLX: 3661: vol 'xxxxxxx', lock at 75927552: [Req mode 1] Checking liveness:
gen 33, mode 1, owner 516444ef-b7b9240f-5805-xxxxxxxxmtime 6405918
2013-12-10T22:23:55.403Z cpu26:13263202)DLX: 4185: vol 'xxxxxxx', lock at 75927552: [Req mode: 1] Not free:
gen 33, mode 1, owner 516444ef-b7b9240f-5805-xxxxxxxxmtime 6405918
2013-12-10T22:24:00.744Z cpu2:13263202)DLX: 3661: vol 'xxxxxxxx', lock at 75927552: [Req mode 1] Checking liveness:
gen 33, mode 1, owner 516444ef-b7b9240f-5805-xxxxxxxx mtime 6405918
2013-12-11T02:20:34.574Z cpu16:13263202)DLX: 3661: vol 'xxxxxxx', lock at 230592512: [Req mode 1] Checking liveness:
gen 9, mode 1, owner 517186c1-1b4dfbe4-0e13-xyyyyyyymtime 5279956
Also, does your storage support VAAI? VAAI allows the ESXi host to offload some tasks to the storage array. This also removes some of the locking processes, since ATS locking and hardware acceleration (both things need to be supported and done by the storage array) can be offloaded to your array.
A few locks is not perse a bad thing, but don't place too many virtual machines on a datastore as this can lead to over excessive locking.
@Avinash: Better would be to look at disk device (u) in esxtop. Then activate the H field: RESVSTATS = Reserve Stats. CON/s shows SCSI conflicts, when a host tries to do I/O on a LUN that is currently being locked. RESV/s shows when something is happening on the VMFS volumes that use SCSI reservations. In example metadata updates.
Hello,
The Host is trying to access a data store that is being locked by a MAC address. We call this as scsi reservation locks.
If we see excessive of these messages then it could cause slowness or performance problems. Check if your getting any reservation conflict messages or compare the esxtop davg - if those two are good then we need not worry about the above.
About Roy: Deep Dive Into SCSI Reservation
Thanks,
Avinash
Also, does your storage support VAAI? VAAI allows the ESXi host to offload some tasks to the storage array. This also removes some of the locking processes, since ATS locking and hardware acceleration (both things need to be supported and done by the storage array) can be offloaded to your array.
A few locks is not perse a bad thing, but don't place too many virtual machines on a datastore as this can lead to over excessive locking.
@Avinash: Better would be to look at disk device (u) in esxtop. Then activate the H field: RESVSTATS = Reserve Stats. CON/s shows SCSI conflicts, when a host tries to do I/O on a LUN that is currently being locked. RESV/s shows when something is happening on the VMFS volumes that use SCSI reservations. In example metadata updates.
Thanks,Avinash,I will see the SCSI Reservation .
Thanks tomtom901,as far as I know ,it suport VAAI,storage are in many backup job.
Thanks.