Adding a tag to git is very resource intensive in fisheye

[Kingsquare]

We've started using git / bamboo / git-flow (sort of) to start tagging our releases... Every new tag though creates massive resource consumption for the server runining Fisheye. (status in administration for the given repo Adding ta [tagname] ([commit hash])

Is there any way to avoid this? It's a fairly large project, but why would this have to take so long (as to a quick git command everywhere else?)

[joe · Accepted solution]

The KB article Git Repository Indexing is Too Slow when Creating a New Branch/Tag suggests and documents a --Xenable-git-content-hash-resolving-heuristic flag that may help in this case.

[Sergey Svishchev]

So, indexing of b/t creations can be slow for Subversion and can be slow for git... is there a supported SCM backend which is NOT resource-intensive for Fisheye?

[Kingsquare]

I'll try the provided solution when i have the time... It seems weird though that a 'special' flag would be needed for something that other systems (bitbucket / github etc etc) don't seem to have a problem with?

Since we've moved away from FeCru as our main introspection tool of our repo and into Bitbucket / Github we just dont see the added value of FeCru over something as trivial as i.e. Bitbucket. It's just 'better' (imho)

[Kingsquare]

It was converted from SVN to GIT with full history (>10K commits)

It still remains a mystery to me for now if that has anything to do with it though?

[Kelly Schoenhofen]

I'm not sure. It's a private fork of the Linux 2.6 kernel. So I don't know what Linus et al did in the early days before they switched to Git - did they migrate it from bitkeeper? Or start fresh? or?

[Sergey Svishchev]

Was it converted from other SCM?

[Kelly Schoenhofen]

I'm not sure. It's a private fork of the Linux 2.6 kernel. So I don't know what Linus et al did in the early days before they switched to Git - did they migrate it from bitkeeper? Or start fresh? or?

[Kelly Schoenhofen]

We're having the same problem. We have a very large git repo with a very large amount of history, and tags take greater than 12 hours to process in FishEye. I've gone though the FeCru performance best practices guide already; is this normal for FeCru?