Sharepoint Forum

Ask Question   UnAnswered
Home » Forum » Sharepoint       RSS Feeds

CPU usage

  Asked By: Phenil    Date: Aug 24    Category: Sharepoint    Views: 3918

We found that CPU usage has been reached 80%-90% [almost 5time in a quick time] and after our initial investigation come to know that 'w3wp.exe ([running on webapplication's system account])' causes the issue but we would like to know the reason behind this. How we can move ahead with this, Any clues?



5 Answers Found

Answer #1    Answered By: Mike Lamb     Answered On: Aug 24

What is running  in that w3wp process (what I am asking is basically what app-pool is running and where and which web-app is using that app-pool). That will tell you and isolate the web-app, then try to find are there any custom components running on that web-app.

It will be a little difficult to quickly identify this problem, you will have to keep on isolating the regions one by one, i.e. web-app then site collection then custom components, etc.

Possible some custom component that is causing more CPU cycles.

Answer #2    Answered By: Jose Scott     Answered On: Aug 24

Using the IIS Logs we found  that crawling happened during that time  and it may caused the issue.

few things:
1. Crawler is trying to access the pages using the account  "-" (may be an localmachine account) and failed to login, so it uses the correponding service account to crawl which means accessing the same URL/Pages two times, I don't know that why it uses the localmachine account here?

Lets say, If you have 100pages
try accessing the pages using "-" account 100hits
try accessing the pages using "service" account 100hits
so, the total hits becomes 200hits instead of 100hits.

2. I posted the below questioin some time back and would like to confirm the same again.
Crawling happening in anyone (only in one server) of the server at a given point of time and rest of the servers are idel during that time, if it is by design what is the use of having more than a server in the farm?

Answer #3    Answered By: Taylor Clark     Answered On: Aug 24
Answer #4    Answered By: Anthony Rutledge     Answered On: Aug 24

Finally, would like to know, We have a few WFEs and an Index server, I want the crawling to be processed in the Index server itself instead of sending the request to the WFEs to get it done.

Is there a way to achieve this, if so what is the best possible way to do it?

Answer #5    Answered By: Heena Nagori     Answered On: Aug 24

If i understand it correct you would like the Indexer box to crawl itself instead of going to another server (that has the WFE service running), if that is correct then all you will have to do is enable the WFE Role on the Indexer box.

Dedicated WFE for indexing (not part of NLB) could be a better solution than using all WFEs, especially when indexing large amount of data. If it's indexer + WFE is on one box, high network traffic is eliminated.

Didn't find what you were looking for? Find more on CPU usage Or get search suggestion and latest updates.