UNSOLVED

SabarnaDeb

updated

13 years ago

S

SabarnaDeb

2 Intern

•

139 Posts

0

2807

December 10th, 2013 07:00

Too many trespasses for lun in clariion CX4-240 during VMware cloning

When we have started VMware cloning from one lun (src) to another lun (dist), we got so many number of trespasses (908) in destination LUN. We have already set failover mode=4 in all ESXi physical server then why we got so many trespasses?? In our SAN network, On each Nexus 5k SAN SW, different VSAN assigned to SPA & SP B and same for other Nexus 5k switch so in storage level, only two path is visible for each physical host.

1path.png

  • Roger_Wu

    6 Operator

    •

    4025 Posts

    535

    0

    Posted December 10th, 2013 18:00

    What's your ESX version? What's your CX4 FLARE version? What's the path selection plugin (PSP) setting now?

    You may refer to the following KBs for more information:

    emc245043 "Why is there excessive trespassing with ESX 4.x and ALUA on a CLARiiON array?" https://support.emc.com/kb/71510,

    emc99467 "What are the Initiator, Arraycommpath, and Failovermode settings for PowerPath, DMP, PVLinks, and native failover software?" https://support.emc.com/kb/31521,

    emc232355 "Using the VMware ESX native multipath plug-in (NMP) with a CLARiiON array" https://support.emc.com/kb/68544.

  • SabarnaDeb

    2 Intern

    •

    139 Posts

    535

    0

    Posted December 10th, 2013 23:00

    Hi Roger,

    ESXi version 5. CX4-240 - 04.30.000.5.524. PSP - i will check with VM team.

  • SabarnaDeb

    2 Intern

    •

    139 Posts

    535

    0

    Posted December 11th, 2013 03:00

    Yes we are using PSP. In PSP, for particular LUN we have selected RR

  • Roger_Wu

    6 Operator

    •

    4025 Posts

    535

    0

    Posted December 11th, 2013 05:00

    According to KB emc245043, can you set it to Fixed and try again?

  • SabarnaDeb

    2 Intern

    •

    139 Posts

    535

    0

    Posted December 26th, 2013 10:00

    Yes, we set it to Fixed but after that also we faced same issue, then we created Thin LUN and there is no tresspass for this LUN.

  • Storagesavvy

    474 Posts

    535

    2

    Posted December 29th, 2013 11:00

    Make sure that the Failover mode for the ESX hosts is set to 4 if you plan to use RR and ALUA.

    Some trespasses are expected in this scenario because VAAI XCOPY offload that is used for the clone of the VM runs on one SP, so if the target LUN is on the other SP, the backend IO redirector is used and the LUN will eventually be trespassed automatically.  However if the target LUN also has significant host IO, it could trespass back and forth every so often.  If the trespasses are happening very quickly and causing performance problems, there is likely configuration issue somewhere or a bug in code.  For example, there was a known issue on VNX in certain versions of 05.31 code that causes lots of trespasses.  But that particular issue did not apply to CX4.

  • SabarnaDeb

    2 Intern

    •

    139 Posts

    535

    0

    Posted December 31st, 2013 02:00

    Yes, VAAI is enabled in ESXi Host. I need to check in details.

  • o17Uu33DCF12520

    6 Operator

    •

    1058 Posts

    535

    1

    Posted January 1st, 2014 21:00

    This is designed by VAAI. when massive data blocks are moved between datastore, such as storage vmotion and vm clone, the lun will be trespassed back and forth between the clarrion/vnx storage processors even if ALUA has been enabled.

    The only workaround is make the source and destination luns owned by the same storage processor.

    Please mark my answer as "correct/helpful" answer if it helps.

  • tlindsay42

    11 Posts

    394

    0

    Posted January 4th, 2014 07:00

    Just wondering, did this cause implicit or explicit trespassing?

  • o17Uu33DCF12520

    6 Operator

    •

    1058 Posts

    79

    0

    Posted January 5th, 2014 17:00

    usually implicit.