I have created a post in the netbackup forum and opened a ticket with support about this issue with no resolution. So i thought i would give the appliance section a try.
Since upgrading to the 2.5 code i started receiving sql errors 13, 6, and 1's randomly on databases across 8 different sql 2008 servers. Some of these servers are vmware virtuals and some are physical machines. none have memory lower than 8 gb. The clients and master are netbackup 7.5.0.4 and the storage on the appliance is 100% dedup.
Support had me change the buffer size on the sql nbu clients and rearrange the jobs so that none would overlap . This did not help. The next idea from support was for me to pursue opening a ticket with Microsoft because they said this was a sql memory issue and performance tuning was needed on the sql side to correct this. Microsoft ran some performance monitors and then pointed the finger back at NBU saying it was not a sql problem considering I just started having the issue. During all of this i decided to upgrade to the 2.5.1 code to see if what that would do and it caused a slight mess of things so both the Microsoft ticket was closed and the NBU ticket.
While the appliance issue was being corrected i had to point all backups to directly run to tape. I quickly noticed that all sql errors went away and each was 100% successful and This went on for several weeks. Once the appliance was back on line all of the sql errors reappeared.
The errors will happen every night but not always on the same servers and not always the same databases. I am consistently seeing it more on the vm sql servers but it is happening to physical machines.
Event log message on sql server:
- Event ID: 18210
-
BackupVirtualDeviceFile::RequestDurableMedia: Flush failure on backup device 'VNBU0-2220-1884-1353999609'. Operating system error 995(The I/O operation has been aborted because of either a thread exit or an application request.).
-
Event ID: 1 SQLVDI
-
SQLVDI: Loc=TriggerAbort. Desc=invoked. ErrorCode=(0). Process=1348. Thread=2564. Server. Instance=MSSQLSERVER. VD=Global\VNBU0-2220-1884-1353999609_SQLVDIMemoryName_0.
Is anyone else experiencing this same issue with an appliance? Does it sound like a performance problem due to a possible bug with the appliance and sql?
Support said the event messages from above prove this is a Microsoft issue but why does changing the backup to tape and not disk solve the problem?
