After configuring a replicated data set and starting replication with Veritas Volume Replicator (VVR), an error occurs that one of the hosts closed connection while performing the operation.

Article:TECH50209  |  Created: 2006-01-21  |  Updated: 2010-01-10  |  Article URL http://www.symantec.com/docs/TECH50209
Article Type
Technical Solution

Product(s)

Environment

Issue



After configuring a replicated data set and starting replication with Veritas Volume Replicator (VVR), an error occurs that one of the hosts closed connection while performing the operation.

Error



Error V-106-58644-659 One of the host closed connection while performing the operation

Solution



This message may appear when a replication operation attempts to make configuration changes to a replicated volume group (RVG). Either replication will not start, or RVGs disappear from the Veritas Enterprise Administrator (VEA) console due to disconnection.

Cause:
Common causes of the error may include:
  • Unreliable network connectivity.
  • The operation takes so long to complete it exceeds Volume Replicator Administrative Service (VRAS) timeout.

VRAS is used by VVR for heartbeating and command execution between primary and secondary hosts. If the timeout is exceeded, a host forcibly closes the connection between primary and secondary and attempts to reattach the replication link (rlink) between them.

Solution:
If the issue is network related, verify connectivity by pinging primary and secondary host replication IP addresses. Check that ports used for communication and data traffic are not blocked (use vrport to verify the ports used). Note that other network tests may need to be performed by a network administrator to verify reliable connectivity.

Alternatively, VRAS may timeout during an initial synchronization with the SwiftSync feature enabled. Depending on the size of the volume that will be replicated, SwiftSync may take too long to determine how many blocks need to be replicated and exceed the timeout.

To determine if the SwiftSync feature is causing VRAS to timeout, disable it on each replication server by editing the Windows registry:
1. Navigate to HKLM\System\CurrentControlSet\Services\vxio\VVRParams
2. Update the SwiftSync DWORD value from 1 (default) to 0.
3. Recycle the SFW services using the command line:
  • For SFW 4.3 MP2 and below: net stop vxob & net start vxob
  • For SFW 5.x: net stop vxvm & net start vxvm

If the error still occurs after disabling SwiftSync, increase the VRAS heartbeat timeout (AE_TIMEOUT) registry value on all replication servers as shown below. The location of the key varies between different versions of SFW. Note that it may be necessary to manually create the key if it does not exist.
  • 4.3 MP2: HKLM\Software\Veritas\VxSvc\CurrentVersion\constants
  • 5.x: HKLM\SOFTWARE\VERITAS\VRTSobc\pal33\Agents\StorageAgent\constants
  • 5.x 64-bit: HKLM\Software\Wow6432Node\VERITAS\VRTSobc\pal33\Agents\StorageAgent\constants

Amend the value of AE_TIMEOUT between 90 and 180, then use the procedure below depending on your version for the change to take effect.
  • For SFW 4.3 MP2 and below, enter net stop vxob & net start vxob
  • For SFW 5.x use vxassist refresh from the command line



Supplemental Materials

SourceUMI
ValueV-106-58644-544

SourceUMI
ValueV-106-58644-659
DescriptionOne of the host closed connection while performing the operation.

Legacy ID



286632


Article URL http://www.symantec.com/docs/TECH50209


Terms of use for this information are found in Legal Notices