We have bloated (> 820GB) subversion repository that we’d like to bring over to BitBucket. We have a SVNMirror license, but want to enable our developers to start using BitBucket for current version control. The current size of new …db/revs/ directories is >=36GB
We will use the old SVN repo as an archival instance, but would not like to fill up the BitBucket drives with old and unused code.
We’d like a little clarity on whether all the …/revs/* directories will be imported into BitBucket before starting the import.
820GB is an extremely big repository, indeed, it would definitely make sense to limit the amount of data being mirrored to Bitbucket. This can be done by two features: first, it’s possible to set a minimal revision to start the import from – that is, SVN Mirror add-on will not import the whole repository starting from revision 1, but from a later revision. This minimal revision can be set on the Translation Settings tab in SVN Mirror add-on UI in a particular repository:
By default, the add-on imports the SVN data starting from the revision 1.
Another approach to limit about of imported data is to limit the set of the imported branches and tags. This can be done by tuning the mapping configuration, it allows to exclude some branches and tags and even eliminate some files and directories right in the repository. We have an article that describes this feature in detail:
In particular, there is excludeBranches setting that allows to exclude some of the branches from import; and there are includePath and excludePath setting that allow to exclude some paths in the repository. Also, the mapping configuration can be set to import only some branches ignoring all the rest.
Hope it will help, but don’t hesitate to contact us had you faced any issues.
