1

I have a SharePoint with Fast Servers. I have multiple content sources in my Fast content SSA but right now I have one crawl component. I would like to add one more crawl components. I saw this Microsoft document but I am not sure do I need to Full Crawl after this? If yes, can you please tell me why is that as all the indexes are in Fast search and crawler always keep temporary index files so If I add another crawler now in content SSA , why it affects it?

I am not sure if I want to add another database but as of now I just want to add one more crawl component on different server. http://technet.microsoft.com/en-us/library/ff599534.aspx

Improve this question

1 Answer 1

Reset to default
0

Adding more crawl components will not require re-crawling. The crawl components jobb with FAST is merely to connect to the system being crawled, pull down the item, and pass it on to the FAST farm. Therefore, adding more crawl components is not that useful for FAST as all the item handling occurs on the FAST farm.

However, adding multiple crawler databases are useful when you get multi-million items.

Why do you want another crawl component? If you want to increase crawl speed try adding crawler impact rules instead. If adding more threads to the crawling maxes out the network card (or memory/cpu which is unlikely) then you can add another crawl component.

The SharePoint crawler database keeps track of all the items being indexed, to support incremental crawls etc.

Improve this answer
5
  • I want more crawl components(atleast 2) for both performance as well as fault tolerance as, I want to increase crawling speed too, I belive with Fast crawl component job is to crawl the data and pull down the item and pass it on the Fast right ? and also how impact crawler impact rules will only help in both in crawling speed and fault tolerance?
    – EVIA
    Jun 3, 2012 at 0:16
  • No, I am not seeing memory/cpu spikes but its crawler is taking like 6 hours in crawling so I want to reduce that.Adding crawl component in another crawl database may be sounds better to me and both crawl components will crawl both databases? any thoughts? but I was not sure if that will force me to do full crawl which I wanna avoid...
    – EVIA
    Jun 3, 2012 at 0:18
  • If you want to reduce the crawl time adding crawler impact rules is the way to go first in order to pull down content in parallel. Adding crawler component for fault tolerance works out, and is the main reason to do so. If you add another database and want existing sources to be divided on both databases you may have to re-index. Not 100% sure as I have never done this. But it might be so, as data from the existing crawler database data won't move over to the new db. If you add rules to have existing sources stay in your current db, and new ones over to the other b, then you should be ok(?)
    – Mikael Svenson
    Jun 3, 2012 at 6:03
  • Thanks for replying Mikael. Yes, I was thinking the same with another crawl db( but don't wanna do re-index as of now).I will try out crawl impact rules to improve performance but do you have any link which explains that adding crawl component will only give fault tolerance and not add performance with Fast Search.
    – EVIA
    Jun 3, 2012 at 14:17
  • I don't have any links according performance. But adding multiple components in regular SP search is due to the SP server doing ifilter and parsing of the extracted data. All of this is happening on the FAST farm. So unless your crawl server is maxing a resource (CPU/ram/disk io/network), I can't really see a reason why adding another component should have any impact at all. For FAST the SP servers are "merely proxy servers" sending everything over to FAST. You should read the scaling chapter of fs4spbook.com ;)
    – Mikael Svenson
    Jun 3, 2012 at 18:08

Your Answer

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

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