anyone getting short timeouts on windows servers

[crocman]

Customers shopping cart admin area on nt44 appears to be dropping the session and forcing them to log back in at exactly 10 minute intervals. Maybe its the app pool being recycled anyone else experiencing this behavior. It happened when they first upgraded nt44 a few weeks back, I had them move the site to a new app pool and the problem disappeared, but now its back again

[Brangwyn]

Are they active during those 10 minutes? sounds like a normal timeout to me, how are you handling sessions?

[crocman]

Ive tested that twice and kept clicking around so no idle time, it times out right on 10 minutes exactly, doesnt actually give a timeout page just boots you out of the admin area and forces you to log back in, cart session value is set to 20 minutes, its all been fine up until recently. Im thinking its possibly something to do with the app pool recycling but then again I aint no network guru, if memory serves me right session ids are stored in the cart database

[Brangwyn]

App Pool recylcing only happens once per day on default install settings, would be very unusual for it to be set to recycle every 10 mins, the load that would put on the server would be kinda crazy forcing every thing to reload into memory every 10 minutes.

Is it an ASP Cart, or PHP?

[crocman]

I couldnt agree more, can cpu usage settings in iis sometimes affect things like this? its an ASP cart(uniwin), again its been fine up until recently, no settings have been changed, i realize it could be another application or maybe one of those intermittent things. Have had sites on other servers(different hosting company) have similar problems and when they've been moved to different pools it always appears to have solved the problem. Like I said Im no network or IIS guru and appreciate the hazards of shared environments

[jmbeach]

was having a similar issue on NT11, they moved me to another app pool. Still the problem was intermitent... was also told that "It recycles every 480 minutes or when it reaches 400 MB virtual or 192 MB physical memory usage."

You may search around as I've read that IIS6 has more of a tendency to give up session state that IIS5. Maybe Brangwyn can confirm/deny that...

Best of luck!

[crocman]

Thanks for the replies
IMHO their just running everything to capacity and trying to squeeze the most out of the hardware/software. Hopefully things will improve sooner rather than later. I have the same cart software running on a vps with iis set at its out of the box settings and so far never a hitch with that cart installation

[Brangwyn]

192MB for an application pool servicing multiple domains sounds a little too low to me, they would be better off IMO not recycling when memory reaches peak and letting IIS manage the memory instead of the fairly resource expensive recycle when IIS has to recache the domain once someone hits it again. It's quite possible this is the root cause of your problems and the reason why I'm getting a lot of complaints lately about server unavailable messages (normally a sure sign the app pool is recycling which normally only take a few seconds but seems to take several minutes on some of the servers here).

IIS6 seems to manage memory quite well, i've one server with about 70 application pools on it (2GB RAM) and most of the time it's very smooth running, app pool recycles are every 29 hours only nothing else.

[crocman]

Thanks Brangwyn
I asked support what the session timeout setting in IIS was to which they replied 20 mins which I think is the setting out of the box. Where did you get the ram allocation for app pool figure from(Nagios)?

Very informative. Eventually, as resellers (not network admins) we end up needing to know this kind of thing. This is why I took on a vps just to get that extra real life experience in supporting and running a public web server. Are you running each app on your server in its own pool ?

[stanj]

Originally Posted by Brangwyn

192MB for an application pool servicing multiple domains sounds a little too low to me, ......... It's quite possible this is the root cause of your problems and the reason why I'm getting a lot of complaints lately about server unavailable messages (normally a sure sign the app pool is recycling which normally only take a few seconds but seems to take several minutes on some of the servers here).

This sounds like it might be related to my problem of, among other things, Service Unavailable several times a day for up to 20 minutes at a time. These problems with slow servers, timeouts, Service Unavailable, and web sites being down daily seem to have started in the spring. What happened 3-4 months ago either technically or policy wise that could account for the change performance? Is is just overloaded servers or are reasonably loaded servers working with some handicap that was introduced about that time?