LUNs lose connection to SQL instance
Applies to
- SAN
- iSCSI
Issue
- LUNs presented to a server fail to connect which caused a failure in SQL instances.
- iScsiPrt errors 129, 39, 27 can 9 indicate that the Repository is not reachable.
- Windows logs show the following error:
Event ID 39Source: iScsiPrtInitiator sent a task management command to reset the target. The target name is given in the dump data.Cause
iScsi traffic went sent over the same NIC as production traffic can cause timeouts and/or dropped packets as the expected response was not received in time.
Solution
- Provision a second NIC, either physical or virtual, to handle the iScsi traffic
- Enable only "TCP/IPv4" in the NIC's properties window
- Click the "Advanced" button and uncheck "Register this connection's address in DNS"
- Close this dialogue window
- With the NIC properties window open, click "Configure"
- Change Jumbo Packets to 9014 bytes
- Set the following settings to DISABLED:
- IPv4 Checksum offload
- Large Send Offload
- TCP Checksum Offload
- UDP Checksum Offload
- Close all dialogue windows
- Open an elevated PowerShell prompt
- Type the following commands:
- Netsh int tcp set global RSS=Disabled
- Netsh int tcp set global chimney=Disabled
- Netsh int tcp set global autotuninglevel=Disabled
- Netsh int tcp set global congestionprovider=None
- Netsh int tcp set global ecncapability=Disabled
- Netsh int ip set global taskoffload=disabled
- Netsh int tcp set global timestamps=Disabled
- Close PowerShell
- Open Iscsi Initiator
- Create a new connection with:
- Microsoft iSCSI Initiator Adapter
- Enter the IP address of your iScsi connected storage
- Click Save
- Open the "Favorite Targets" tab
- Remove the initiator from this list
- Click OK
Partner Notes
partnerNotes_text
Additional Information
additionalInformation_text
Internal Notes
internalNotes_text
