Problem
VMware SAN based restore fails with "socket write failed" status 2820 error
- vCenter sees virtual machine created successfully but during write to datastore fails.
- Attempts to restore to alternate datastore results in same issue
- All permissions on vCenter and backup host has been checked.
Error Message
...
22/07/2013 11:29:22 AM - begin reading
22/07/2013 11:42:40 AM - Critical bpbrm(pid=7060) from client nbexxxxxxxxxxxxx: FTL - Virtual machine restore: file write failed
22/07/2013 11:42:40 AM - Critical bpbrm(pid=7060) from client nbexxxxxxxxxxxxx: FTL - Virtual machine restore: file write failed
22/07/2013 11:42:42 AM - Info tar32(pid=3068) done. status 5
22/07/2013 11:42:42 AM - Info bptm(pid=1924) EXITING with status 24 <----------
22/07/2013 11:42:42 AM - Info tar32(pid=3068) done. status: 185
22/07/2013 11:42:42 AM - Info tar32(pid=3068) done. status: 24: socket write failed
22/07/2013 11:42:42 AM - Error bpbrm(pid=7060) client restore EXIT STATUS 24: socket write failed
22/07/2013 11:44:12 AM - restored image WP200A006_1373666722 - (the restore failed to recover the requested files(5)); restore time 00:18:24
22/07/2013 11:44:12 AM - end Restore; elapsed time: 00:18:26
VMware policy restore error(2820)
Cause
Virtual machine was originally created with VMware thin provisioning disk option.
As the behavior of thin provisioning disk will dynamically initialize or commit blocks as the data is written to the disk there may have been a problem with dynamic provision in VMware causing allocating block error in vCenter event logs and a series of write errors in the Netbackup logs.
Solution
Change VMware disk format from "Original Provisioning" to "Thick Provisioning with Lazy Zeroed" and triggered same restore using same recovery host/vCenter server.
Ensure there is enough space available in the target datastore to perform thick provisioning disk format.
Applies To
ESX: 4.1
Netbackup 7.5.0.4
Media server: Windows 2003 64bit(64GB RAM)
Restore: vCenter server
Transport Mode: SAN