All Versions
Fisheye 4.2 DocumentationFisheye 4.1 Documentation
Fisheye 4.0 Documentation
More...
Understanding How FishEye Indexes Repositories
FishEye's administration interface now allows you to configure the number of threads used for repository indexing. FishEye uses two types of indexing operation:
- Initial indexing - the complete scan of the repository that FishEye does when FishEye first accesses the repository, and
- Incremental indexing - the quick, background process that regularly scans repositories for changes.
The number of threads available for these indexing operations is now configurable.
Prior to FishEye 1.5, a transient error during processing the initial indexing operation would cause a repository to move to the incremental indexing thread, even when a large amount of indexing remained. This would impact the incremental indexing of other repositories in the FishEye instance. In FishEye 1.5 this has been corrected. A repository will remain on the initial indexing thread until it has successfully completed the initial indexing operation.
Setting FishEye's Total Threads
To configure the number of threads FishEye uses for indexing, open the FishEye 'Admin Screen', then click 'Server Settings'. Now, click 'Edit Details' under 'Resource Limits'. You can increase the number of threads available for each indexing phase independently. Enter the desired number of threads for each process and Click 'Update'.
Screenshot: FishEye Multi-threading Configuration
!
Choosing a Thread Setting
As changing the number of threads always involves a trade-off in performance, you should generally change these settings in increments of one, then observe the performance impact. A safe range is from 1-3 threads on both settings. Using more than three threads on either setting is not recommended; the more threads, the busier the server will become when indexing. There is certainly no benefit in having more threads than you have repositories. Adding a lot of threads may impact overall system performance.
By default, FishEye uses one thread each for the two indexing modes. This is equivalent to the functionality in FishEye prior to version 1.5.
For more information on managing FishEye's performance, see the FishEye Tuning page.
This feature was introduced in FishEye 1.5.
Overview
Content Tools
Apps
Log a request with our support team.
Raise an issue for our developers.
See answers from the community.
Blog and update our documentation.
Except where otherwise noted, content in this space is licensed under a Creative Commons Attribution 2.5 Australia License.
5 Comments
Bob Swift
Sept 24, 2009Is the recommendation of 1-3 still relevant in the age of multiple quad-core processors?
George Barnett
Sept 25, 2009It's going to depend on the number of CPUs you have an the data you are using. 3 is the sweet spot for most people, although you may get benefits from raising the number. As always, it's best to test the difference in your environment.
Anonymous
Jun 03, 2011Is the default, "Unlimited" threads or should setting these >1 generally yield faster indexing?
Daniel R
Mar 06, 2012The default for FishEye is one thread each for the two indexing modes. Depending on the number of repositories and the performance of your hardware more threads will lead to faster indexing. Given the fact that a thread can only index one repository at a time, more than one thread for a single repository will not lead to faster indexing.
Anonymous
May 15, 2012I have 80 repos 6 quad core processor, 16GB RAM and SAN storage. 6 of my repos are in initial scan (and could be for days or weeks.) Are you sure number between 1 and 3 is as high as this should go? If so, you need to explain a bit more about what is under the covers here. My performance is bad.