Unsolved
This post is more than 5 years old
1 Rookie
•
18 Posts
0
3826
May 8th, 2013 05:00
SQL Server VM (vadp) - FULL backup fine, Incremental backup a disaster!
Networker server 7.6.3 backing up a Windows 2008 R2 VM running SQL Server, NOT using the SQL Module for Networker, just regular VADP backup.
The weekly full backups always work fine, the nightly incremental backups always fail and the Networker logs suggest to look within the VM for clues.
The event log on the VM shows hundreds of errors at this time each night (11:00pm to 11:05pm) most of them referring to SQL. Also when I view VSS writers after the fail, several of them (IIS, SQL, WMI) are in failed state.
I've attached a cut of the event log, here are some examples of entries:
| MSSQLSERVER | 18210 | Server | BackupVirtualDeviceFile::SendFileInfoBegin: failure on backup device '{597E9B68-D363-4ECF-AC31-85021DA05FD4}1'. |
Operating system error 995(failed to retrieve text for this error. Reason: 15100).
| SQLWRITER | 24583 | None | "Sqllib error: OLEDB Error encountered calling ICommandText::Execute. hr = 0x80040e14. SQLSTATE: 42000, Native |
Error: 3013
Error state: 1, Severity: 16
Source: Microsoft SQL Server Native Client 10.0
Error message: BACKUP DATABASE is terminating abnormally.
SQLSTATE: 42000, Native Error:
3271
Error state: 1, Severity: 16
Source: Microsoft SQL Server Native Client 10.0
Error message: A nonrecoverable I/O error occurred on file ""{597E9B68-D363-4ECF-AC31-
85021DA05FD4}19:"" 995(failed to retrieve text for this error. Reason: 15100).
SQLSTATE: 01000, Native Error: 4035
Error state: 1, Severity: 0
Source: Microsoft SQL Server Native
Client 10.0
Error message: Processed 0 pages for database 'master', file 'master' on file 1.
"
| SQLVDI | 1 | None | SQLVDI: Loc=TriggerAbort. Desc=invoked. ErrorCode=(0). Process=1680. Thread=7512. Server. Instance=MSSQLSERVER. |
VD=Global\{597E9B68-D363-4ECF-AC31-85021DA05FD4}17_SQLVDIMemoryName_0.
MSSQLSERVER 3041 Backup BACKUP failed to complete the command BACKUP DATABASE msdb. Check the backup application log for detailed
I've had multiple calls open with EMC Support for months and they have no idea what the problem is.



1goyalp
18 Posts
0
May 8th, 2013 14:00
Hi GBY,
STEP 1: Had you enabled CBT using command line at proxy host? If not please use the following format:
directory>nsrvadp_modify_vm.exe -H vCenter server -P protocol -u user -p password -l lookup method -k lookup key -c command.
greenblueyellow
1 Rookie
•
18 Posts
0
May 9th, 2013 09:00
Hi 1goyalp,
I've just tried that command with -c info and it says CBT=Enabled so I don't think the problem is there. I'm going to try stopping the SQL VSS service on the client machine tonight and see if that makes any difference but would be keen to hear any other suggestions.
Thanks
1goyalp
18 Posts
0
May 9th, 2013 14:00
Hi GBY,
The CBT is enabled and it is taking incremental backup. As per the working architecture of CBT it should make the snapshot for each virtual disk. It looks like:
vmname.vmdk
vmname-flat.vmdk
vmname-ctk.vmdk
vmname-000001.vmdk
vmname-000001-delta.vmdk
vmname-000001-ctk.vmdk
In some cases, CBT might reset and lose track of incremental changes and snaphot disk do not get deleted automatically so
STEP 1: Delete all the old snapshots. (Don't try the backup now).
STEP 2: For each virtual disk, this entry must be there scsix:x.ctkEnabled = "TRUE".
Now you have to enable CBT in SQL. The command is something like this:
STEP 3: SQL> alter database enable block change tracking;
You enable block change tracking by making an entry in the controlfile:
(Please take help from SQL DBA to enable CBT in SQL DB, I have bit less knowledge in SQL).
CBT in SQL checks 64KB of every block and it will track all the changed block from DB before backing up and will not pause the running DB to compare and cache the changed block. This will avoid SQL VSS to invoke interrupt handler.
STEP 4: Now try the incremental backup.
Could you please give me the application log event in .evt format and version of vSphere, VDDK, Networker build version of server and client?
greenblueyellow
1 Rookie
•
18 Posts
0
May 13th, 2013 02:00
1goyalp I think you may be onto something as far as as the VMDK files go.
This VM lives in 2 different datastores, the first one has:
vmname-ctk.vmdk
vmname_2-ctk.vmdk
vmname_2-000001-delta.vmdk
vmname.vmdk
vmname_2.vmdk
the second has:
vmname.vmdk
vmname-ctk.vmdk
vmname-000004.vmdk
vmname-000012.vmdk
vmname-000009.vmdk
vmname-000006.vmdk
I'm hesitant to delete the CTK, DELTA and 00000x files. Is there a command I can run from the ESX host to see which files are actually in use right now?
Also you say this:
STEP 2: For each virtual disk, this entry must be there scsix:x.ctkEnabled = "TRUE"
Whereabouts should I see/set this?
Thanks.
Umraan
96 Posts
0
May 30th, 2013 00:00
Hi
how often do you run the incremental backups ?
have you tried running a level 1 (differential) backup
sechm
1 Rookie
•
79 Posts
0
May 30th, 2013 07:00
At a guess, the backup is initiating a VSS backup of SQL using the SQL VSS writer. Why this fails for incremental backups but works for fulls I'm not sure. Possibly an incremental backup is interpreted as a transaction log backup. Your error messages indicate problems with the "master" and "msdb" databases, the system portion of SQL. These databases cannot be quiesed.
Try looking into the database recovery model for your databases, full or simple. Also, the VMware KB 1031200 might be some help, on how to disable specific VSS writers in VMware tools.
Since you're not using the SQL module, perhaps a file system level consistent image backup is enough for your needs.
tonyalbers
75 Posts
1
September 9th, 2013 06:00
Maybe it's just me, but shouldn't you use NMM for the SQL backups?
AFAIK VADP doesn't do applications, so I don't think you'll be able to do a recovery.
Just my 2 cents,
/tony