Timeouts, Connectivity Issues, and Remembering the Repositor

andrew00 · Post by **andrew00** » Wed Mar 31, 2004 3:12 pm

I'm having a few problems with 2.01. First, the Vault connection times out like every 3-4 minutes. I know there's a way to correct this - what is it?

When it times out, it hangs for about 10 seconds (very annoying) and then says "The Vault Server could not be contacted." This is, of course, untrue, because the next keystroke I hit, the system gets the Vault server and proceeds with checkout.

This happens every 3-4 minutes, so it's getting pretty old.

Also - the system refuses to remember my repository. I mean, I only have one, but it insists on prompting me, even when a connection is lost. Is there a setting for this?

Thanks in advance,
Andrew

dan · Post by **dan** » Wed Mar 31, 2004 9:03 pm

The Admin Tool -> Server Options pane contains the timeout setting for the server timeout, but it looks like this problem isn't necessarily the vault timeout - it could be an IIS timeout.

Has anything changed in your IIS settings recently, or in your network config? It doesn't make sense that this would suddenly start timing out.

For Vault to remember your repository, create a login profile from the login dialog, and set it to automatically select the default repository.

sterwill · Post by **sterwill** » Wed Mar 31, 2004 9:10 pm

IIS has an option to recycle its worker process (the thing that does all the server work executing Vault) based on certain thresholds (elapsed time, number of requests served, idle time, etc.). Windows Server 2003 has easy-to-use control panels to set these variables, and we've heard from several Vault customers that some hosting companies (who host Vault servers) will have their servers set to recycle this process often (I guess to avoid IIS leaking resources). If this is done too often. it will make Vault unusable. The default time span for recycling is around 1 day, which isn't so bad (but still hurts a lot for big trees which we can't cache for more than a day). If these options got changed somehow, this could be causing your server to restart.

Are there startup messages in the Vault server logs that correspond to the timeouts/reconnects you see (startup messages are multiple lines, and include the version of the server and such)? This would confirm that the worker process is being recycled, and your problem isn't strictly network related. The logfile can be found at %WINDIR%\temp\sgvault\sgvault.log.