-
Bug
-
Resolution: Fixed
-
Medium
-
2.3.7, 2.4.1
When a git repository has sub modules (similar to svn:externals) configured it for some reason gets references to changeset ids from the submodule, which causes issues whenever FE falls back to the SCM for blame or annotated views of the file.
Either we have to add support for submodules or ensure that we don't index these changesets.
The bug can causes an error in the slurp when falling back to the SCM for blame info (RevisionCache.getBlameFallback)
The exception reported in the logs is similar to:
2011-02-11 09:34:01,929 ERROR [InitialPinger1 ] fisheye.app OneOffPingRequest-doRequest - Exception during slurp
com.cenqua.fisheye.rep.DbException: com.atlassian.fisheye.dvcs.handler.DvcsProcessException: Error while communicating with VCS: error: unable to find 216be7cb0c341e0aecf87bf47bed5f52b1fc88ea
fatal: no such path _Tools in f5c9d35c26d9186a30391faeb3768c1acb5162b3
Caused by: com.atlassian.fisheye.dvcs.handler.DvcsProcessException: Error while communicating with VCS: error: unable to find 216be7cb0c341e0aecf87bf47bed5f52b1fc88ea
fatal: no such path _Tools in f5c9d35c26d9186a30391faeb3768c1acb5162b3
Caused by: com.atlassian.utils.process.ProcessException: While executing: "git blame -l -p f5c9d35c26d9186a30391faeb3768c1acb5162b3 - _Tools"
![[Extranet] Page](/images/icons/generic_link_16.png "[Extranet] Page"){kind=icon}
[FE-3122] FIsheye is indexing changeset ids of GIT submodules when it shouldn't be causing
FishEye 2.8.2 includes a change to prevent submodule revisions from being indexed. FishEye will completely ignore submodules when it encounters them during indexing, and thus they will not be visible or browsable in the FishEye UI.
For those with reviews containing existing submodule hashes, see CRUC-6197.
We are aware that users want full indexing support for Git submodules, and this will be handled under FE-3828.
Jaron Sampson
added a comment - This appears to be a problem for our organization as well, I would like to see better git support all around in fisheye/crucible - including this feature request for submodule support.
Jaron Sampson
added a comment - This appears to be a problem for our organization as well, I would like to see better git support all around in fisheye/crucible - including this feature request for submodule support. 
Mikko Hyvärinen
added a comment - We also ran into this issue immediately when trying out FishEye on a particularly large project with dozens of submodules. It is a very frustrating situation: we would definitely want to use FishEye but we cannot collapse the git submodule hierarchy into a single repository for various reasons.
Mikko Hyvärinen
added a comment - We also ran into this issue immediately when trying out FishEye on a particularly large project with dozens of submodules. It is a very frustrating situation: we would definitely want to use FishEye but we cannot collapse the git submodule hierarchy into a single repository for various reasons. 
Herbert Balagtas
added a comment - I was also hoping using submodules would allow us to get around the 5 repository limit for the starter pack. Right now the submodules shows up as regular files and dumps error messages as soon as we click on them. We're a small company and the starter pack is all we can afford at the moment.
Herbert Balagtas
added a comment - I was also hoping using submodules would allow us to get around the 5 repository limit for the starter pack. Right now the submodules shows up as regular files and dumps error messages as soon as we click on them. We're a small company and the starter pack is all we can afford at the moment. Just wanted to add that having recently purchased Jira, GreenHopper, and Confluence licenses all of which we are very happy with, we were looking forward to giving fisheye/crucible a trial but ran into this issue on the very first day. This is frankly a deal breaker at the moment, which should be sorted out soon IMHO if fisheye can even remotely claim to 'support' git.
Submodules are very widely used throughout all our larger projects, and until this is sorted out we will not be pursuing fisheye.
Matthew Horan
added a comment - We also cannot open code reviews for repositories with submodules. Matthew Horan
added a comment - This is a major blocker for us, as a re-index is required to pull in any commits to the repository, and auto indexing doesn't work because of the error. This basically makes Fisheye useless for repositories containing commits with submodule references.
Matthew Horan
added a comment - This is a major blocker for us, as a re-index is required to pull in any commits to the repository, and auto indexing doesn't work because of the error. This basically makes Fisheye useless for repositories containing commits with submodule references.
Our log scanner is picking this issue up as a "known problem". However, we do not use submodules, but subtrees