Atlassian recommends a 2GB max SVN repo size if migrating to DVCS. What 'size' is this and how do I measure it?

[Andy Worsley]

I'm migrating our FishEye SVN repository projects to Bitbucket/Git repositories. The recommendation of <2GB is in https://go-dvcs.atlassian.com/display/aod/Migration+Introduction

Is this 2GB 'size':

- The size of the SVN server deployment (e.g. a 'du' on the SVN server filesystem including the database of revision etc.)?

- The size of the SVN backup dump?

- The size on disk of project trunks when I've checked them all out?

Please advise which, and how I measure it.

[rstephens · Accepted solution]

Hi Andy,

The "size" in question is the size of the SVN backup dump or size of your SVN FS on disk (they will be very similar in size). If you file a support ticket with Atlassian support, we would be more than happy to let you know how big your repository is.

Alternatively, you could use a tool like svnsync to make a local clone of your SVN repository that includes history, and measure the size of that.

Kind Regards,

Richard Stephens

[Andy Worsley]

Thanks, Richard. The backup dump I have (from a while ago) is 1.5GB zipped, which is similar to the deployed SVN FS size of 1GB. Have raised the ticket.

[Timothy]

This would be the size of the SVN repository.

[Andy Worsley]

Thanks. And how practically do I measure that?

[Timothy]

Google it yet?

[Andy Worsley]

Google? What's that? :P

Anyway, yes all the suggestions I've found are Stack Overflow threads etc. that mostly involve having command-line access to your SVN server (which I don't have through FishEye OnDemand), or backing-up/mirroring the repository locally, which seems like overkill just to get a simple statistic.

If you are aware of an easier way to determine the SVN repository size, then please do let me know.

[Stefan Saasen]

Hi Andy, This is the size of the git/hg repository after the conversion. It's obviously difficult to estimate the size of that repository before you converted your subversion repository but looking at your current SVN repository might give you some indication. Does it contain many binary files (e.g. large media files or libraries/artifacts like .jar files)? If so did they change frequently? The best way to find out for sure is to actually convert the repository and then check the size of the repository on disk after compacting/garbage collection. E.g. for git: $> cd /path/to/git/repo $> git gc $> du -hs .git Cheers, Stefan

[Andy Worsley]

Thanks, Stefan. But is it the Git repo or the SVN repo as Richard says? Can you guys resolve between you?

In any case, I suspect they might be similar anyway, at least to order of magnitude. I'll probably do a double-pronged approach and just do a test conversion as you suggest though.

[Stefan Saasen]

Hi Andy,

Well, it really is the git/hg repository size that we mean, I'm reponsible for the svn -> git migration content on that page :)

A large SVN repo might end up being signficantly smaller in git so the final size is really what counts here.

The git/hg repository size decides whether your experience with DVCS will not be hampered by slowness. See https://go-dvcs.atlassian.com/display/aod/How+to+Reduce+the+Size+of+your+Repository for the downsides of large DVCS repositories.

Looking at https://go-dvcs.atlassian.com/display/aod/Migration+Introduction again I can see why that is not obvious. I'll look into make that clear on the page.

Hope this helps.

Cheers,

Stefan

[Andy Worsley]

Ah OK, you're the authority then :) Excellent. That's all clear now, thanks.

[Andy Worsley]

Google? What's that? :P

Anyway, yes all the suggestions I've found are Stack Overflow threads etc. that mostly involve having command-line access to your SVN server (which I don't have through FishEye OnDemand), or backing-up/mirroring the repository locally, which seems like overkill just to get a simple statistic.

If you are aware of an easier way to determine the SVN repository size, then please do let me know.