We were showing a high level of NFS errors - lock file creation errors, lock file access errors and APD states.
StorageApdHandler: 421: Device or filesystem with identifier [xxxxxxxx-xxxxxxxx has exited the All Paths Down state.
WARNING: NFSLock: 2226:
Cannot create lock file .lck-6200000000000000 (4 retries) : Not found
As it turned out, the firmware for the VIC / enet needed to be updated. An update to the firmware for the UCS blade to version 2.1.3b fixed this issue.
do you have any documentation to back this up? we are seeing similar errors. and we are on 2.1.3a
ReplyDeleteI will look for the case and pull the notes. It was not publicly available, it took some multi-vendor cooperation to discover.
ReplyDeleteFrom the Cisco case notes:
ReplyDeleteGetting disconnects going to NFS VMs. The vmkernel logs keep showing:
WARNING: LinNet: map_pkt_to_skb:2094: This message has repeated 24 times: vmnic3: failed to find inner ethernet header for encapsulated frame
Investigating we found the ESXi 5.5 is not supported until 2.1.3b. After upgrading the blade firmware to 2.1.3b then we stopped seeing the issue.
There was no KB issued because the Cisco/VMware/NetApp opinion was that ESXi 5.5 was not supported pre 2.1.3b.
Thanks for the Update, I am working with Cisco/Flexpod Support now. I looked at the vmware compatibility matrix for 5.5 and it just lists 2.1.3 not any specific revision. I will post here when i get a resolution. I am not getting that error in the vmkernel logs though. I am getting the "all paths down" errors you described above
ReplyDelete