Start a Conversation

Unsolved

J

1 Message

1046

February 22nd, 2022 04:00

DDBoost Backups Failing on SQL Availability Group Secondary Nodes

Hi,

I've been trying to setup the DDBoost backups on a SQL HA environment we have setup but am struggling to get it working on anything other than the currently active node.

Current config is a set of SQL 2017 Enterprise instances sitting on Server 2019 hosts on separate hardware split across 2 DCs and connected to local DDBoost hardware (DD6800s I'm told) that is local to each DC. Current SQL native backups (Ola Hallengren based) run on the secondary nodes without issue and per design to shift the IO off the primary node - this means copy only fulls and Tlogs due to the AG usage but works fine for DR. DB backup sizes range from a few GB through to around 8.5TB per backup cycle for the larger main DB we host here.

When trying to setup a federated backup via the DDBoost XSPs or via the SSMS Tool we get an error that the backup cannot occur on the secondary as it can't run a query on the DB to get the filegroup or filelist.

Errors/Logs;

raw_WillExecute: 'SELECT groupname, status FROM [DBName]..sysfilegroups'

appagent_19_8/nsr/db_apps/bsmsql/sqlxbsacompforsave.cpp(931): Unable to detect the Filegroup or Filelist of database [DBName]. Backup cannot proceed for this database.

DDBoost Command via SQL SSMS Query;

EXEC dbo.emc_run_backup
'-c [WSFC FQDN Address] -D 4 -S 16 -l full -y +30d -a "NSR_DFA_SI_DD_HOST=[IpAddr]" -a "NSR_DFA_SI_DD_USER=[ddboostuser]" -a "NSR_DFA_SI_DEVICE_PATH=/[Path]" -a "NSR_DFA_SI_DD_LOCKBOX_PATH=C:\Program Files\DPSAPPS\common\lockbox" -a "NSR_COPY_ONLY=TRUE" "MSSQL#[SQL AG Listener Name]:[DBName]"'

DDBoost Command (from DDBoost logs);

ddbmsqlsv.exe -m [SecondarySQLServerName] -D 4 -S 16 -l full -y +30d -a NSR_DFA_SI_DD_HOST=[IPAddr] -a NSR_DFA_SI_DD_USER=[ddboostuser] -a NSR_DFA_SI_DEVICE_PATH=/[Path] -a NSR_DFA_SI_DD_LOCKBOX_PATH=C:\Program Files\DPSAPPS\common\lockbox -a NSR_COPY_ONLY=TRUE -z FEDERATED_SLAVE=true -z FEDINDEX_NAME=MSSQL -z FEDCLIENT_NAME=[WSFC Cluster Name] -z NSR_JOB_PARENT_JOBID=0 -z FEDAAG_NAME=[SQL Listener Name] -N MSSQL#[ListenerName]: MSSQL:[DNName]

Is there a way to command DDBoost to skip this check for federated backups where secondaries are, by default, unreadable and in a recovering state? We can't allow these DBs to be read-only due to various factors so skipping this check or finding a working way to take a backup here is the best shot. Standard SQL backups work so seems strange that DD Boost can't stream the backup out either as I'd expect it to use the same essential process and stream the output into the DDBoost system rather than a backup file similar to other 3rd party backup tools (NetBackup, Litespeed etc). It does print details to the logs that it checks the SQL HA status etc so I'm surprised it doesn't 

Thanks!

2.4K Posts

March 3rd, 2022 03:00

You forgot one important detail - the backup application. 'NSR' obviously points to NetWorker using the appropriate modules.

Unfortunately, I cannot help you directly as I do not have the necessary DB knowledge. During my last employment, the DB admins backed up without a NW module:

  -  they exported the DB to a disk file

  -  from where these files were backed up as ordinary files

Yes - you must provide additional disk space to keep this data for 1-2 days (depending on your policies) but the solution is so much easier for the backup admins as they do not have to learn about the DB details. And you have some buffer time which helps when you have to maintain the backup environment.

With respect to your original question - may I suggest that you start a thread in the NetWorker User group - those guys seem to have the better knowledge about your problem.

 

No Events found!

Top