Re: Help with very large repositories

Michael Muller wrote:
> Many thanks to those of you who responded, this is certainly a very responsive
> mailing list!
>
> All of what I'm hearing fits - the original developers chose to organize our
> codebase into libraries with a huge number of source files. The 9200 file set
> I described is a _single library_. The RCS files are all stored in the same
> directory.
>
> I don't think the code base has any branches or tags, so at least copying of
> that directory should not be an issue. But I will try reorganizing the file
> set into more of a tree and see if that makes any difference.
>
> Thanks again!

The other thing I do when creating and importing a brand new repository is to
create the repository with the flag disabling syncing after each transaction:

svnadmin create --fs-type bdb --bdb-txn-nosync repos

This speeds up the initial import significantly and if something hoses with the
repository, you can always create a fresh one and start the import again.

After the import is complete, I make sure there are no processes attached to the
repository, and in db/DB_CONFIG, comment out the following line:

# [requires Berkeley DB 4.0]
# set_flags DB_TXN_NOSYNC

and then run 'svnadmin recover' on the repos to enable syncing after each
transaction.

Regards,
Blair

-- 
Blair Zajac, Ph.D.
<blair@orcaware.com>
Subversion and Orca training and consulting
http://www.orcaware.com/svn/
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org