How to speed up repository scan?

Dmitry Baryshev

I have a remote svn repository with svn+ssh:// access. Fisheye doesn't support such repositories by default, so I use a native svn client (javahl).

When fisheye start to scan my repository, it creates a separate ssh connection for each svn command, which takes 2.5 seconds or more. So it will take ~5 days to scan a complete repository with several thousands commits.

Is it possible to speed up this? Can fisheye use a shared ssh connection instead of creating separate connections?

Felipe Kraemer

Hi, Dmitry.

There are some things you could try:

1. According to the Best Practices for FishEye Configuration document, we recommend using file:/// protocol for fastest indexing performance. If you cannot install FishEye on the server where Subversion is running, we recommend using svnsync to mirror the repository onto the FishEye server.
2. If you have a large repository, you could try to configure a starting revision from which to start indexing. All commits prior to this starting point are not included in FishEye. Read more.
3. You can exclude portions of your repository that you are not interested in. Read more.
4. You can also split a single and large repository into smaller logical components such as projects or products.
5. Change the repository definition to look at a subset of your repository. Typically this involves setting the path within your repository that you wish FishEye to index. Read more.
6. These documents can also be very useful:
   http://confluence.atlassian.com/display/FISHEYE/Improve+FishEye+Scan+Performance
   http://confluence.atlassian.com/display/FISHEYE/Tuning+FishEye

Could you please give these suggestions a try and let us know how it goes?

Kind Regards,
Felipe Kraemer
Atlassian Support