Post back problem (CPU at 99%)

Post back problem (CPU at 99%)

  
Hi,

We have an application with several e-spaces published. However, we are experiencing a few problems with one of them. More accurately, with ONE web page of that e-space.

The page lists records from a table, stored in a DB in another machine, and gets about 5000 records at once. Some days ago, the page started to provoke an abnormal growth in resource consumption and now the worker process (w3wp.exe) easily reaches 99% of CPU. The first time the page is loaded it returns what it's supposed to. It presents paging, search fields and the first page of results. However, if we try to move to the next page, or search, or order the results...the page takes forever to process. Sometimes it results in a "Page cannot be displayed" message. In IIS logs we can see 301 and 302 http errors. In Service Center, in the Screen Log, we sometimes get huge loading times (when the pages doesn't end up in an error) of about 375602 ms, 535459 ms, 445011 ms, ...

We suspected that the total number of records to show could be too much, but we have another page that list a similar kind of entity, with almost the same record count and the page loads perfectly, and we can order the records, navigate through pages, without problem. Of course that, once the worker process reaches 99% and starts to grow, other pages start to fail as well.

We also noticed that the w3wp, once it starts to "eat" all resources, it also starts to generate millions of page faults!

We recently had hard drive space problems in both the hub server and the DB machine. We thought that it could be the source of our issues, but we cleared out some Gb of space and the situation still occurs, even after some restarts.

We also suspected of some javascript code on the page's 'onload'. So we eliminated the code and republished the e-space, but still nothing new.

We even tried to isolate this e-space in an Application Pool of its own, so we could manage its recycling without affecting other e-spaces. But the problem resists.

Any clues to what could be going on?
Thanks,
R.Matos
Hello Ricardo

If you are having this problem in a project for a customer, you should definitely contact OutSystems Technical Support to get help.

Typically high CPU consumption from w3wp.exe is a symptom of infinite loops. You mention that this happens after accessing a particular page. Does the page have any loop of any kind in the preparation (like an if statement that returns to itself), in the preparation for any of its web blocks, or in the the action that fills a table records (4.2)?

Without looking at the OML it is difficult to know what is happening, but you can try to put some audits in the preparation for that screen to see if they are being written too many times, or even if they are written at all.