.NET virtualization frozen cache

We have TS 6.1 installed on a unix solaris box. One of our sites uses .NET and virtualizes on a separate windows server. Here's the behavior we are seeing:

1. Create a new .aspx page and it virtualizes correctly on first generation

2. Edit the DCR, regenerate and the old version of the file is displayed. We can download the file from TeamSite and see that the changes were made, so it is a virtualization problem. We can go to a different user computer and try to preview the page and will get the old version, so it does not appear to be a cache issue on the user's machine.

3. Refreshing the page a few times will finally get the changes to show up. The time we have to wait after generation to virtualize increases from minutes to hours as user demand increases. Rebooting the windows server resets the dealy to 15-30 seconds.

4. The site makes use of server side includes. We get occaisional "failed to start monitoring changes in file" errors, always related to one of the include files. Refreshing the page makes the error go away.

5. We have edited the nocache registry key on the windows server -- no luck.

Any ideas on what might be going on?

Find more posts tagged with

Comments

sixtrojans

Someone suggested increasing the samba configuration change notification updates from 1 second to 5 seconds. Does anyone know what the optimum setting should be on this?

Migrateduser

I may not be correct... but it seems an error caused by the .net enviornment. Please check the IIS settings. Check whether you are able to update the dll's ....

sixtrojans

I may not be correct... but it seems an error caused by the .net enviornment. Please check the IIS settings. Check whether you are able to update the dll's ....

Do you think the whole problem is related to the .net envrionment or just the "failed to start monitoring changes" error? We can't figure out if there is one problem causing both issues or if it is 2 separate problems.

sixtrojans

It looks like solved our problem. For what it's worth, here's what we did:

The 'change notify' problem appears to have been caused by hitting a limit in the number of network BIOS commands (see http://support.microsoft.com/kb/810886).

Per the MS KB, we added/updated the MaxCmds value in the windows registry for the lanmanworkstation to 10000 and rebooted.

We also added a global paramameter to the SAMBA configuration file named max mux and set its value to 10000 as well and restarted Samba on the Solaris server. The max mux is the equvalent of the MaxMpxCt for lanmanserver on windows.