PSSDiag Deletes Trace Files

  • I am trying to use the publicly available version of PSSDiag to do some troubleshooting. I finally have everything working pretty well. The only problem is that during shutdown all of the trace files are deleted!

    According to the documentation it looks like if you run the PSSDiag service remotely, on shutdown it will try to copy the trace files to the computer from which you started the service.

    I am running this on a clustered sql 2000 sp3 server. I actually run the Diag Manager while logged into the virtual server of the cluster. I do see in the PSSDiag.Log that it says the machine is a remote machine or a cluster. Then the next line says that the target machine is not a cluster (even though it is).

    There is a file created during shutdown that looks like it is trying to copy the files to somewhere but it errors out (##ServerName__sp_trace_COPY.txt) saying the files are in use by another process. Then there is another file called ##ServerName__sp_trace_DEL.txt that is empty.

    So it looks like it tries to copy the files to somewhere (possibly itself?), the copy errors out and then the files are deleted.

    Does anyone know how I can get around this behavior? It is maddening to collect data for a half hour and then watch it disappear.

    Thanks,

    George

  • First check the pssdiag.ini file which carries your configuration.

    Search ini file for delete word.

     

    MohammedU
    Microsoft SQL Server MVP

  • Thanks for the reply. Unfortunately I don't have an .ini file. By default it uses xml that is stored in the registry. And the directions say not to fiddle with these settings manually. And I don't see anywhere in the DiagConfig GUI where I might be able to change this behavior. I am more familiar with the .ini file but the Public version of PSSDiag does not come with one.

    Thanks,

    George

  • Then copy the files to different location before shutting down pssdiag...

     

    MohammedU
    Microsoft SQL Server MVP

  • It could be you need to run it from the active node of the cluster rather than the virtual server - I don't have a test cluster to test on so it's just a best guess.

    [font="Comic Sans MS"]The GrumpyOldDBA[/font]
    www.grumpyolddba.co.uk
    http://sqlblogcasts.com/blogs/grumpyolddba/

Viewing 5 posts - 1 through 4 (of 4 total)

You must be logged in to reply to this topic. Login to reply