2 zvols. 1 was destroyed and rebuilt, now both are inaccessable

[dc3dog]

My FreeNAS / ESXi lab in summary;

3 nics:
on board/management, 192.168.1.20/24
add on intel/MPIO, 192.168.7.9/29
add on intel/MPIO, 192.168.7.17/29

2 volumes:
zvol-mirror, (2) disks
zvol-stripe, (3) disks

2 iSCSI Portals
stripe, 192.168.7.9/29
mirror, 192.168.7.17/29

2 Targets / Extents
mirror
stripe

The above config was presented to ESXi host and all was working both fine and dandy.

Next I remove 1 small drive from the stripe volume and replace it with 2 larger disks (new total 4 disks). Destroyed stripe zvol and created raidZ zvol using ip of previous stripe Portal. FreeNAS now reports both zvols Status as HEALTHY. Now neither zvol will connect with the ESXi host. Packet captures show that FreeNAS is refusing the iSCSI connection attempts from ESXi (i.e. FreeNAS responds immediately with RST, ACK).

How can I regain access to my zvols? Again, this is a lab. I could reset FreeNAS to defaults and start over but that is not the point of this exersize. How do I regain access to my zvols? Thanks and Kind Regards.

 

[philhu]

Well, first, why did you remove a disk on the stripe without destroying the striped vol first? A stripe dies with a disk removed.

Second, was the ESXi system connected when you pulled the drive? If so, on a Centos system, the only way to regain the net disk is to reboot the Centos box. On Centos, you always unmount a net drive before you pull the network drive out from under it

Can you try to attach to vols using a windows box or something, just to verify it is what I say above and not a freeNAS problem?

 

[dc3dog]

hmmmm, I had to delete then recreate both Portals and restart the iSCSI service. Now back to fine and dandy.

 

[dc3dog]

So the question still remains, why would the equivalent of 1 failed drive take down all iSCSI?