Forum Discussion

Ptolemee's avatar
Ptolemee
Level 4
7 years ago

Exchange 2010 GRT restore failing with status code 23 "socket read failed"

- Configured NFS on media servers (windows 2008 r2) and Exchnage 2010 Clients

- GRT enabled in Exchange Policy,  Client list: dag01.domain.local

- DAG name: dag01.domain.local; DAG nodes: exc01.domain.local, exc02.domain.local, exc03.domain.local

- In folder Netbackup\db\altnames created 3 files: exc01.domain.local, exc02.domain.local, exc03.domain.local and in each of those added dag01.domain.local as the single line

- Full backup to MSDP disk Storage Unit is successfull

- Running  Backup Archive and Restore from master server. Can browse full backup image of dag01.domain.local and select a single mailbox.

- When I start the restore it fails with Status Code 23 "Socket Read Failed". Detaied Status in job monitor shows "BPRESOLVER has executed on server EXC01" socket read failed (23)

- Added EXC01 (short name) file in folder Netbackup\db\altnames with same contents dag01.domain.local

- Tried restore again but it failed with the same status cde 23 error.

Any idea what might be wrong?

Thnaks.

  • Each DAG node is mailbox server and CAS server. Exchage environment has a load balancer for the CAS servers.

    Found solution at the bottom of Article ID:100038019 "Everything you need to know to get Exchange Granular backups to work".

    Sectin 7e:

    If using a load balancer for your Exchange CAS server, you will need to put a host file entry on your master and media server to point the load balancer names to 1 of your CAS servers.
           Otherwise, the restore will fail with a communication error.
           For example:
           CAS01 and CAS02 are the names of my CAS servers.
           1.2.3.4 and 1.2.3.5 are the IP addresses of my CAS servers.
           CASARRAY.COM is the name of my load balancer.
           1.2.3.6 is the IP of the load balancer.

           For the host file, choose to point the loadbalancername to CAS01 (only choose 1)
           1.2.3.4       CAS01        CAS01.domain         CASARRAY.COM

  • Much outdated tribal knowledge here.

    Don't use altnames to undermine your security. Use the Distributed Application Restore Mapping (DARM) to give individual Exchange servers access to a backup image that is catalogued under the DAG name. With current Exchange (2013 or later) and NetBackup (8.1 or later) versions you shouldn't even need the DARM, because NetBackup discovers the relationship between the DAG and its nodes.

    If you really have NetBackup 7.5, then yes, you need the hosts file trick to get past your CAS load balancer. You don't need it from 7.6.0.3 forward, because NetBackup doesn't need to find the CAS for GRT restore. Rather than using a local URL to EWS on a CAS, NetBackup uses EWS autodiscovery to obtain a remote URL that it can use on any mailbox server.

    Note that if you upgrade your Exchange you will have to upgrade NetBackup. NetBackup 7.5 did not support Exchange 2013.

  • Each DAG node is mailbox server and CAS server. Exchage environment has a load balancer for the CAS servers.

    Found solution at the bottom of Article ID:100038019 "Everything you need to know to get Exchange Granular backups to work".

    Sectin 7e:

    If using a load balancer for your Exchange CAS server, you will need to put a host file entry on your master and media server to point the load balancer names to 1 of your CAS servers.
           Otherwise, the restore will fail with a communication error.
           For example:
           CAS01 and CAS02 are the names of my CAS servers.
           1.2.3.4 and 1.2.3.5 are the IP addresses of my CAS servers.
           CASARRAY.COM is the name of my load balancer.
           1.2.3.6 is the IP of the load balancer.

           For the host file, choose to point the loadbalancername to CAS01 (only choose 1)
           1.2.3.4       CAS01        CAS01.domain         CASARRAY.COM