01-22-2013 05:33 AM
Good Morning,
I'm trying to backup an Exchange server with GRT "option".
I follwed the istruction that i found here :http://www.symantec.com/connect/videos/exchange-2010-sp2-grt-netbackup-75
On Media server i stopped and disable the Server and Client NFS services
I execute the command " sc config portmap start= auto "
and i created the empty file No.Restricions under C:\Program Files\Veritas\Netbackup\db\altnames
On the Exchange server the client disable the Server NFS services.
I created the backup policy ( MS-Exchage Server ) and I "flag" Perform snapshot backup and Enable granula recovery.
I submitted the backup job and a first result is
Info bptm(pid=5716) using 32 data buffers
22/01/2013 13:00:07 - Error bpbrm(pid=992) from client "clientname": ERR - Client for NFS service is either not installed or not running.
22/01/2013 13:00:09 - Info bptm(pid=5716) start backup
22/01/2013 13:00:09 - Info bptm(pid=5716) backup child process is pid 5420.4608
22/01/2013 13:00:09 - Info bptm(pid=5420) start
22/01/2013 13:00:09 - begin writing
22/01/2013 13:00:24 - Error bpbrm(pid=992) from client "clientname"- Error encountered while attempting to get additional files for Microsoft Information Store:\clientname_MB_01_Do_Not_Use\Logs_1358855898\
22/01/2013 13:00:24 - Error bpbrm(pid=992) from client "clientname": ERR - Exchange granular restore from this image may not work.
So at this point where I wrong ?
I'm trying to backup on DataDomain
Many thanks
R.
Solved! Go to Solution.
01-29-2013 02:28 AM
Sometimes worth setting a proxy host for restores (under client host properties for the exchange servers) - this would be you media server
As it looks like that media server is an appliance then i guess the fragment size it writes to disk is the default at 25600 MB.
To reduce the database system errors it is worth increasing the client read timeouts (to at least 1800) on the Media servers and also for GRT backups i like to use a smaller fragement size as it mounts smaller chunks of the backup images to do the restore - i like to use 5000MB which is a lot smaller than an appliances default
Hope this helps
01-23-2013 01:47 AM
EDIT:
I solved the proble regarding NFS service..
Now i have another problem
Error bpbrm(pid=2536) from client CLIENTNAME: ERR - Unable to initialize nbfsd.
22/01/2013 15:16:25 - Error bpbrm(pid=2536) from client CLIENTNAME: ERR - Error encountered while attempting to get additional files for Microsoft Information Store:\CLIENTNAME_MB_01_Do_Not_Use\Logs_1358863991\
22/01/2013 15:16:25 - Error bpbrm(pid=2536) from client CLIENTNAME: ERR - Exchange granular restore from this image may not work.
01-23-2013 11:28 AM
Moved from "NetBackup Appliances" forum to "NetBackup" forum
01-28-2013 04:58 PM
01-29-2013 01:35 AM
01-29-2013 02:28 AM
Sometimes worth setting a proxy host for restores (under client host properties for the exchange servers) - this would be you media server
As it looks like that media server is an appliance then i guess the fragment size it writes to disk is the default at 25600 MB.
To reduce the database system errors it is worth increasing the client read timeouts (to at least 1800) on the Media servers and also for GRT backups i like to use a smaller fragement size as it mounts smaller chunks of the backup images to do the restore - i like to use 5000MB which is a lot smaller than an appliances default
Hope this helps
01-29-2013 03:08 AM
01-29-2013 03:15 AM
Ok - try the proxy host and the fragment size then - if you change the fragment size you will need another full backup first in order to test restores from that one
01-29-2013 03:44 AM
01-29-2013 03:20 PM
01-30-2013 12:25 AM
Hello aleco many thanks for your suggestion but my media server is W2008 R2.
The strange thing in this moment is that client side when we try to expand a single public forlder i saw , master side,always a restore process.
01-30-2013 01:27 AM
The restore is perfectly normal - the backup image gets mounted and read to expand the trees as you browse the image - it always happens with GRT backups
The issue is if that restore job finishes before the tree has expanded after which you get nothing or the database system error
This is why client read timeouts (clients and especially the media server) are so important as it needs to be more time than it takes to read the image and this is also where the smaller storage unit fragment size really comes into play as it can just pick the fragment involved and work on that rather than the default huge fragment which again could casue the timeout
The database system error usually is a combination of bpcd and bpdbm timeouts - much of which were hard coded with much higer values in the later releases of NetBackup for the very reason of GRT restores
Still worth a try with a smaller fragment size
01-31-2013 12:47 AM
01-31-2013 03:33 AM
OK - so would you be able to try a 5000MB fragment size to see if it resolves your issue?
#edit#
One other thing I just remembered .....
When opening the BAR GUI it by default selects a lot of backup images ... this can also cause an issue.
Make sure that you only select one backup image (one little green orb) before trying to expand anything
Hope this helps
01-31-2013 11:43 PM
02-01-2013 01:41 AM
Ok - So maybe getting there
Put you client read timeout right up on your media server first, i know it is already 7200 but increase it anyway.
Next look to see if you use a proxy for restores - thsi si on the clients host properties - Windows Client - Exchange tab (the exchange granualr restore proxy host)
If you are not then put your media server in there, if you are then remove it - it is all trial and error for what is best in your environment - do this for all mailbox servers and see if it helps
Also keep an eye the restore job to see if that finishes long before you get the error - thsi indicates where the timeout is ocurring
It is worth setting the client read timeout on the mailbox servers host proerties to a high value too
See if this helps
02-01-2013 04:00 AM
mmmm I inserted as proxy the media server that performed the backup. Client Read timeout is now setting to 14400 but i have always the same result ERROR: Cannot connect on socket
EDIT: moreover.. client side for every "browsing" to search the item I saw a restore of 20 minutes... is it normal ?