FWIW, I had to convert over to using svn://-style URLs (vs. file://-
style) because my mirrored SVN roots are on an NFS filesystem;
executing a bunch of checkouts simultaneously from the same root is
Bad (and causes some checkouts to fail).
Instead, I setup an svnserve on my cluster head node to serve the SVN
repos in a read-only manner out to my cluster. It was trivial to
setup the svnserve, and it seems to work quite well in my MTT setup.
On Dec 20, 2007, at 2:28 PM, Ethan Mallove wrote:
> On Thu, Dec/20/2007 11:51:30AM, Jeff Squyres wrote:
>> On Dec 20, 2007, at 11:38 AM, Ethan Mallove wrote:
>>>> FYI: per my post to the OMPI devel list, I discovered the
>>>> "svnsync" command yesterday. svnsync allows you to
>>>> mirror an entire SVN repository in a read-only fashion; I
>>>> mirrored the ompi-tests repository locally at Cisco. A
>>>> one-line cron job keeps my ompi-tests SVN repo mirror up
>>>> to date every night.
>>> Can you post that crontab?
>> It's just
>> 0 19 * * * . $HOME/.bashrc ; svnsync sync file:///home/mpiteam/mirrors-svnroots/ompi-test
>> (repeated for the other SVN repo's I'm mirroring)
>> See the SVN book docs for svnsync -- they discuss how to setup a
>> repo for mirroring with svnsync (there's a few steps you have to do
>> set it up).
>>>> Hence, my SVN checkouts are no longer serialized -- which
>>>> had been causing significant delays when the "test get"
>>>> phases of many MTT jobs were running at more-or-less the
>>>> same time.
>>> If I see this:
>>> $ ls /foo
>>> README.txt conf dav db format hooks locks
>>> How do I "svn checkout" from /foo using a "file:" URL?
>> svn co file:///foo
>> A file:// URL is just like an http[s]:// URL -- just put the absolute
>> directory name after it (to include the leading "/" -- so a file URL
>> will have 3 /'s).
> Ah. The *3* /'s are important. 2 slashes didn't do it :-)
>> Jeff Squyres
>> Cisco Systems
>> mtt-users mailing list
> mtt-users mailing list