We are in socket error hell here.
We have a small to midsize environment 1 Master Server and 3
Media servers all running NetBackup 6.5.1a all servers are Windows 2003, the Master
is 32bit and the Media servers are 64 bit. All Dell hardware the media servers
are quad core dell 2850’s and the Master a Dual core 1850, Broadcom NIC(s).
My backend storage is a Data Domain 580 appliance, we are using the VTL option.
We are primarily backing up windows servers, but, there is Linux in the
environment too.
We have been converting from a mixed NetBackup 5.4mp6 and
Backup Exec 11D environment all over to a our NetBackup 6.5.1a environment. As
boxes are being added we noticed these socket errors (25) that have been occurring
more and more. We have reached out for support from both a 3rd
party and Symantec and though both appear to be making efforts to assist we are
not making headway fast enough at all. The majority of the errors occur during the
weekend, Friday night fulls (we run incremental differentials during the week);
however, we only run about 10 more policies during the weekend than during the
week.
At this point we have added the registry key HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters\TcpTimedWaitDelay
setting the timeout to 30 seconds and it has appeared to help a little. We also
spaced out the jobs more; but, our servers never got taxed and still got the
errors.
Has anyone ever been in the spot where they have gotten a
ton of 25 errors over different clients at different times with no real rhyme
or reason. The documentation for these errors are poor at best. Picking a place
to begin trouble shooting or focus on has been difficult at best.
Thanks in advance.
Feel free to call me out on anything I may have left out.
Steve
