CPU Maxed after (maybe?) double click of submit?

[OKUser]

First, thanks for providing this script. I'm especially impressed with the notification if someone tries to abuse the script. I missed a setup step and was notified when I submitted my own form. Great feature.

I just implemented the script for the first time for one of my clients. It was a breeze to set up, and testing went very well. Since we actually turned it on, we've had to reboot the server three times because the CPU maxed out. Prior to this the only time we've had to reboot the box was either a kernel update or an update to HSPhere. It's a RH 7.3 box running HSphere 2.4 if that matters. All patches are applied. Sendmail seems to be the culprit, judging from the log files. We've niced the task so it can't kill everything on the box again, but I just wondered if anyone had seen this before?

I suspect the double click because after the last reboot I got two identical messages. As far as I know this is the only new activity on this server in the last few days, although I could be wrong.

Any ideas/suggestions?

TIA,
LP

[russellr]

Hi,

Clicking the submit button is one submission. Double-clicking it is still one submission. So, I don't think that would be the issue.

However, sendmail could be the culprit.

If you can Telnet or SSH into your server, run a command called top.

It shows you the current processes and CPU load. The processes that are using the most CPU are shown at the top.

A lightly loaded server will have a load average of less than 1. A load average greater than, say, 3 or 4 is OK but starting to get busy. A load average of 20 or more means the server is going to be very unresponsive.

To test your theory, run top and do some quick form submissions. If Sendmail is using a lot of your CPU you'll see it straightaway.

If you're using a shared hosting service, your hosting provider should be able to track down the problem for you.

If not, you might consider hiring an expert on RH to diagnose the problem.

[OKUser]

Thanks Russell.

I'm familiar with top - problem is the server won't respond to get it running. My Cacti monitoring shows me the CPU maxes out, but not what's doing it. The maillog shows sendmail as rejecting connections because the load average is too high, which is why I've zeroed in on this. There is one more customer who just brought up a PostNuke site though, so it could be them. The fun part will be figuring out which one is doing it.

I will post what we find (when we find it...) in case others may run into something similar.

Thanks,
LP

<edit spelling>

[russellr]

Hi,

I'm suggesting you login and run top before the problem occurs, not after it.

Of course, this only works if you can trigger the problem or if it happens frequently enough.

[OKUser]

Kept it running all day yesterday and of course it never went off line. Didn't have it pulled up yet this morning when it maxed again - of course.

I'm really more suspicious of the PostNuke site after some further testing yesterday. Need to put my head together with the two guys who worked on it and see what we can come up with.

Thanks again Russell.

LP

[OKUser]

We finally found the problem. It was in the PostNuke site, where a cut-and-paste went terribly wrong. They pasted a link, which actually contained back side instructions to PostNuke instead of posting the path to an image. Pure coincidence on the timing.

Thanks for the reads, suggestions, and the script. I plan on recommending it to any of my users who want to collect online info.

LP

[russellr]

Hi,

Thanks for the closure. Let us know if you need any further assistance.