locked
[Microsoft][ODBC SQL Server Driver]Communication link failure - During snapshots removal in VMware ESX RRS feed

  • Question

  • Hi, I got a problem that occurs during backups (snapshot removal) of our virtual SQL 2014 server in VMware. 

    01:00:51     Operating System   : Windows Server 2012(6.2 build 9200) 
    01:00:51     RDBMS              : Microsoft SQL Server 12.00.4416
    01:00:51     ODBC Driver        : SQLSRV32.DLL ver 06.03.9600

    01:06:15 ERROR : Error (0) Error in AGRFetchCur: SQL Error (0): 08S01 [Microsoft][ODBC SQL Server Driver]Communication link failure
    01:06:25 ERROR : Error (0) 0(08S01):[Microsoft][ODBC SQL Server Driver]Communication link failure
    01:06:25 ERROR : Error (1) ERROR 1: Fetching with cursor
    01:06:25 ERROR : Error (0) 0(08S01):[Microsoft][ODBC SQL Server Driver]Communication link failure
    01:06:25 ERROR : Error (1) ERROR 1: ODBC: Performing ROLLBACK
    01:06:25 Transaction aborted.

    Probably the SQL server are freezing a couple of seconds during this time because of high IO on storage. Is it possible to tune the timeout values to get this more stable?


    Tuesday, December 15, 2015 12:31 PM

Answers

All replies

  • refer thread also share output of following.

    cmd ->Netsh int tcp show global

    cmd ->Netsh int tcp show chimneystats

    https://support.microsoft.com/en-us/kb/942861


    Please click Mark As Answer if my post helped.

    Tuesday, December 15, 2015 2:30 PM
  • Microsoft Windows [Version 6.3.9600]
    (c) 2013 Microsoft Corporation. All rights reserved.

    C:\Users\xxxxx>Netsh int tcp show global
    Querying active state...

    TCP Global Parameters
    ----------------------------------------------
    Receive-Side Scaling State          : enabled
    Chimney Offload State               : disabled
    NetDMA State                        : disabled
    Direct Cache Access (DCA)           : disabled
    Receive Window Auto-Tuning Level    : normal
    Add-On Congestion Control Provider  : none
    ECN Capability                      : enabled
    RFC 1323 Timestamps                 : disabled
    Initial RTO                         : 3000
    Receive Segment Coalescing State    : enabled
    Non Sack Rtt Resiliency             : disabled
    Max SYN Retransmissions             : 2


    C:\Users\xxxxx>Netsh int tcp show chimneystats

    Your System Administrator has disabled TCP Chimney.

    This is from my client, Win2012R2.

    What settings could I change to see if gets better? Max SYN retransmissions?

    Tuesday, December 15, 2015 2:53 PM
  • This is normal and expected and annoying behavior of VMware.

    Please see:

    http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1002836

    • Marked as answer by Eric__Zhang Thursday, December 24, 2015 8:46 AM
    Tuesday, December 15, 2015 3:07 PM
  • Does it exists a good workaround? Faster Storage ? 
    Wednesday, December 16, 2015 9:57 AM
  • More info from vmware.log:

    Checkpoint_Unstun: vm stopped for 27016967 us

    = 27 seconds halt!

    Probably not a setting to tune, only more hardware or stop using snapshot of SQL VMs. 

    Wednesday, December 16, 2015 12:23 PM
  • That would be a better question for VMware.

    However, No, there is no workaround.  The larger the number of snapshots removing, the longer the "pause". 

    Wednesday, December 16, 2015 12:44 PM