Symptoms
After the unexpected outage of a node, the ESXi node cannot provide iSCSI target service. The vSAN iSCSI target service fails to start.
Run the following command to check the status of the service:
/etc/init.d/vitd status
Restarting the vitd service does not resolve the issue, nor does rebooting the ESXi host.
The UI shows all the vmk's as normal, similar to a working node:
vitd[2493716]: VITD: Thread-0x1ca7f5de40 invalid listen address vmk3:3260
vitd[2493732]: VITD: Thread-0x7d979e5e40 getaddrinfo for vmk3:3260 failed: Name or service not known
vitd[2493732]: VITD: Thread-0x7d979e5e40 invalid listen address vmk3:3260
vitd[2493748]: VITD: Thread-0xaf045e9e40 getaddrinfo for vmk3:3260 failed: Name or service not known
vitd[2493748]: VITD: Thread-0xaf045e9e40 invalid listen address vmk3:3260
vitd[2493765]: VITD: Thread-0x7d27ef1e40 getaddrinfo for vmk3:3260 failed: Name or service not known
Cause
This issue is caused due to a faulty vmk0, though it is not shown as missing on the UI. Even the command
esxcfg-vmknic -l shows the vmk0.
By design, if the customer needs the vSAN iSCSI target service, vmk0 must be functioning. On the VxRail server, vmk0 is used for node discovery. It is also used to start the vSAN iSCSI target service.
Resolution
To resolve this issue: Review the troubleshooting steps in VMware.
VMware KB
vSAN Health Service
Restart the vSAN iSCSI services on the host.
To restart the vSAN iSCSI service without interrupting current I/O operations, run this command:
etc/init.d/vitd restart
To restart vSAN iSCSI and rest I/O, run the following command.
This may interrupt and cancel current I/O operations.
/etc/init.d/vitd io_reset
If the above does not resolve the issue, delete the failing vmk0 and re-create it. Restart the vSAN iSCSI target service again.
etc/init.d/vitd restart