ieeadmin wrote
It looks like you're running into the same issue as myself. As I mentioned, I run about 20 websites on separate IP addresses on my server. A majority run only 1 portal, and don't cause any problems. The problem arises from the sites that run multiple portals. Those sites, even when not being heavily used, seem to balloon the w3wp.exe process to over 500 Mb (just like yourself). Since I have the luxury of using Remote Desktop, I can actually watch the memory use going up. It happens faster if I have someone working on one of the portals. What's weird is that the memory usage balloons, and ALL the sites on the server slow down, but the actual CPU usage never goes about 20 or 30 percent.
My solution is to kill the w3wp.exe process manually using the Task Manager and wait for it to restart. Once it restarts, it balloons almost immediately to about 80 Mb, then sticks there, as long as someone isn't doing any development work on one of the portals.
Definitely use host-tracker to keep your sites "alive". It's increased the response time dramatically. I'm also going to experiment with putting my lower-traffic sites into different application pools to try to pinpoint the issue.
Thanks for the further info. And please do report back if you find out anything more about it.
Since last week my sites have been nearly unusable because my host changed a setting. Where previously it would generally sit at around 500Mb all up and occasionally blow up and restart, now it's stuck in shutdown cycle.. it shuts down about 100 times an hour. It's great for the memory - never goes over about 300mb, but it also means I can hardly use the sites!! The guy who did it then promptly vanished and hasn't been back at the helpdesk since. I've been unable to get someone else to change it back either. It's just nuts. Obviously I have to get remote desktop and start doing these things myself.
But what it also shows me is that it must be able to be set up so that it recycles at some particular point... say at 600mb (I have about 730mb ram to play with).
I also find same as you that the cpu usage stayed low and the all the sites simply slowed down to near standstill when it was starting to balloon. That included all sites I have including php shopping carts and static html. Something was stuck and it wasn't using cpu in particular. What does continue to work fine at that point is ftp and Expression Web (Frontpage) access to the sites. I tried many times to disable the web config or use an app_offline file to force the main portal to let go and restart, but it never worked. Obviously it's that w3wp process messing with things. If that's a process that only services http, then perhaps ftp and such are unaffected.
I also assume that when it starts to freeze up, that I won't be able to log in with remote desktop and would have to wait for it to crash and restart regardless. hmmm mind you, given that ftp and EW works fine, perhaps remote desktop would as well?.
Thanks for the further information. It's very useful to hear.
Rob