1

This is my setup:

2xSQL, 2xAPP, 2xWFE where;

  • SQLs is 16GB ram

  • APPs are 12GB ram

  • WFEs are 12GB ram

  • C: 40GB

  • Log: 50GB

  • Index: 50GB

Search is running on one of the APP servers, and Requests are on the other. I have a combined DB size of 700GB.

So, first time, Full Crawl was running for 30h, indexed around 1.6mill documents. After the Full Crawl the Incremental Crawl started and it ran for 29h, a second Incremental Crawl startede and i stopped it after 10h, because those times did not seem right to me.

Even the MySites that is in a own WebApp and a own Content Source ran for 2h Full Crawl.

Im not that strong in Search, but I discovered that my initial configuration was downgraded to:

  • APPs 5GB ram
  • WFEs 4GB ram

So could this have been the issue for the slow times? The small amount of RAM, and waht about the drive space? When it crawls the C drive only has around 10GB free, could the drive place also bottleneck something?

I have asked them to roll the configuration back to the inital, and size the drives so the C is minimum 80GB and the Index is 20 procent of the combined DB size, that is 140GB.

What else should I look into? Could this be suficient enough to solve the Crawl issue?

I am still waiting on the server-team to upgrade the config, before I can test again.

3
  • Can you clarify your topology too, where are each of the 6 components running? Commented Mar 17, 2015 at 13:06
  • I will look into this when the server is up and running. The SQL server was to full so the connection was terminated to SQL, maybe this and the low hardware specs was a part of the slow crawl.
    – emo
    Commented Mar 20, 2015 at 9:34
  • Okay, I can see know that the QueryProcessingComponent, the IndexComponent, the Admincomponent, the CrawlComponent, the contentProcessingComponent and the AnalyticsProcessingComponent all are on one server.
    – emo
    Commented Mar 23, 2015 at 9:51

2 Answers 2

0

5GB ram is killer for the sharepoint farm, you expect more performance issue.If you read "Choose hardware resources for the host servers" technet article, so minimum Ram for the crawl server is 8GB for small farm and 16GB for the Query server.

In order to get the optimum performance then you need to increase your Hardware.

8
  • I really hope that it will solve the issue, boosting the hardware. I would hate to tweak on some services to cope with the slow crawl issue.
    – emo
    Commented Mar 16, 2015 at 13:14
  • that is the 1st thing to do, with 4 /5 gb of ram...you cant event troubleshoot the issue....I bet even MSFT not help u until you increase the hardware.
    – Waqas Sarwar MVP
    Commented Mar 16, 2015 at 13:17
  • The performance is alot faster now with the hardware upgrade. But, after my Full Crawl when I start the incremental it runs and indexes just as many items as the full crawl. I get alot of "processing this item failed because of a xml parser error" errors.
    – emo
    Commented Mar 23, 2015 at 9:53
  • I would open a new question on this error so that get more people attention. ...meanwhile check this blogs.msdn.com/b/spblog/archive/2013/10/08/…
    – Waqas Sarwar MVP
    Commented Mar 23, 2015 at 13:29
  • Thanks, I will. I did check that, and did all that, but still same amount of errors. Will try with a new question.
    – emo
    Commented Mar 23, 2015 at 13:30
0

Considering your topology and overall lack of resources you NEED to make adjustments. I too have everything running on a demo server, but I have 16GB RAM and it works fine for demos. In a production environment I would suggest separating the Index and Content processing roles. I cover this in my Pluralsight class on search. (http://bit.ly/abSearch) You can also review the recommended architectures at: http://www.microsoft.com/en-us/download/details.aspx?id=30383

2
  • Would it function if I had all the components on each of my two servers? So the index, content, crawl, etc, is both on Server1 and Server2. Thanks.
    – emo
    Commented Mar 23, 2015 at 13:03
  • Usually you would provision the services separately. It all depends on how your performance and resources are used. Commented Mar 24, 2015 at 12:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.