05-19-2015 05:56 AM
Reviewed prior threads on Status 46 but did no see solution to this particular issue.
SQL backups running finie for months until this past weekend. Of course "nothing has changed".
Noted in dbclient log for the failed jobs the image names preceded by slash '/' as in /NODENAME.MSSQL7.CLUSTERNAME.db.DBNAME...
Can't figure out where that slash is coming from.
05-19-2015 12:31 PM
Recent Windows update?
> wmic qfe list brief
...and filter / sort by dates? What was installed between the dates of last good working, and first now broken, backup?
05-19-2015 01:21 PM
Thanks for that command. So there were some KB updates on 5/17. But the backups failed 5/16.
05-19-2015 02:27 PM
Did the backups finish before the date-time of the installs?
(can you tell whether I'm clutching at straws... :p)
...and I googled 'netbackup sql slash' and it would seem that extra '/' or '\' is bit of a theme? Red herring though?
05-20-2015 06:08 AM
Sounds like a glitch or a change in DNS....
Where exactly is the status 46 coming from? SQL client?
All I can suggest is to troubleshoot as normal status 46:
C:\>bperror -S 46 -r
server not allowed access
The server tries to access a client, but access is blocked.
Possible causes are as follows:
* The server is not listed on the client as a valid server.
Do you have bpcd on the client?
05-20-2015 07:32 AM
Name resolution looks fine, forward and back, Master <> Client <> Media
---------------
bpcd log snippet from yesterday (the one from 5/16 looks fine):
CLIENT is not a master server
CLIENT is not a media server either
...
exit status 46
---------------
05-20-2015 07:56 AM
Is there a connection request from itself in bpcd?
Is CLIENT also a (SAN) media server?
05-20-2015 10:33 AM
Noticed in the client bpcd log for the successful job (prior week), it was the media server and client connecting at same point in the log, which was successful.
That snippet above the client was trying to connect to itself. It is not a SAN media server.
05-20-2015 11:09 AM
05-22-2015 12:44 AM
Will - could you be out of ports on client side:
> netstat -an | find /i /c "tcp"
...how many ports open?
.
Some example steps to query, and change, and reset:
reg query "HKLM\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters" | find /i "MaxUserPort" (FYI - the default is that it does not exist) netsh int ipv4 show dynamicportrange udp netsh int ipv4 show dynamicportrange tcp netsh int ipv6 show dynamicportrange udp netsh int ipv6 show dynamicportrange tcp (FYI - the default for all four above is start: 49152 number: 16384) (FYI - The following extended the range from 16384 ports to 20000 ports) netsh int ipv4 set dynamicportrange udp start=20000 number=45500 netsh int ipv4 set dynamicportrange tcp start=20000 number=45500 netsh int ipv6 set dynamicportrange udp start=20000 number=45500 netsh int ipv6 set dynamicportrange tcp start=20000 number=45500 netsh int ipv4 show dynamicportrange udp netsh int ipv4 show dynamicportrange tcp netsh int ipv6 show dynamicportrange udp netsh int ipv6 show dynamicportrange tcp reg query "HKLM\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters" /v "MaxUserPort" FYI - THESE ARE NOTES ON HOW TO RESET TO DEFAULT: netsh int ipv4 set dynamicportrange udp start=49152 number=16384 netsh int ipv4 set dynamicportrange tcp start=49152 number=16384 netsh int ipv6 set dynamicportrange udp start=49152 number=16384 netsh int ipv6 set dynamicportrange tcp start=49152 number=16384 reg delete "HKLM\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters" /v "MaxUserPort"
05-22-2015 10:50 AM
dbclient log now shows Status 18 CONNECT FAILED and Status 25 cannot connect on socket.
"The license for the requested feature is not available" :(
Don't know what the server admins did but I'm going to try to remove/reinstall NetBackup Client.
05-22-2015 11:07 AM
05-22-2015 12:54 PM
It had been working for months without issue per OP. Works fine again after client reinstall.
Very odd. I've never seen so many weird errors from one client.
05-22-2015 01:08 PM
Solved - removed & reinstalled NetBackup Client software (7.6.0.4).