subversion 1.0.5, mod_dav_svn, gentoo, db4.1.25, and database corruption

["paul." <po...@umich.edu>]

Hello again list,

After taking everyone's screaming recommendation to use the https:// 
method (because svn+ssh doesn't work), I ran into the following 
problem:

[glass:~/business/haquery] crescent% svn update
svn: PROPFIND request failed on '/svn/haquery'
svn: PROPFIND of '/svn/haquery': 500 Internal Server Error 
(https://phonics.no-ip.com)

which led me to hop onto the server, and administer a can of svnadmin 
recover:

vcr svn # pwd
/var/svn
vcr svn # ls
conf  fnord  fnord2  haquery
vcr svn # svnadmin recover haquery
Please wait; recovering the repository may take some time...

Recovery completed.
svn: Berkeley DB error while opening 'uuids' table for filesystem 
haquery/db:
Invalid argument
vcr svn #

Interesting! A quick google leaves many impressions --
"it's db4.1 -- you need db4.2"
"it's the --mutex flag in bdb"
"it's the phase of the moon"

I really like svn. I have a backup of the broken repository. However, 
I've spent more time fighting this version control system in the past 
few months than I have using it. I even learned how to admin apache 
just so I could use it. I wish I had the time and ability to hop on the 
bugtracker and find the error points in the code. In the end, i just 
want to use a good versioning system.

1) how do i fix this problem?
2) when is SVN expected to hit 2.0 (1.0)? I might try it again then.

- paul


---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org

Re: subversion 1.0.5, mod_dav_svn, gentoo, db4.1.25, and database corruption

["paul." <po...@umich.edu>]

On Jul 19, 2004, at 8:28 PM, Shawn Erickson wrote:
> On Jul 19, 2004, at 3:14 PM, paul. wrote:
>> After taking everyone's screaming recommendation to use the https:// 
>> method (because svn+ssh doesn't work)
> svn+ssh doesn't work? How so?

Nope. Permissions and BDB logfiles. A well discussed topic.

I'm running an svn install on a server where I can't just 'move the 
binary and replace it with a umask script' and as such, svn+ssh is 
broken for me. I asked the list about it a month ago, because I spent a 
good chunk of the day su-ing and chmod-ing and chown-ing files. The 
answer was "huh, that sure sucks." Not exactly *working* in my opinion, 
but that's what the new filesystem support in svn is about. Too bad 
that's not ready for prime-time either.

Anyhow, now I'm having an unexplained BDB problem, this time taking the 
solution to "just have a single point of access" with apache2. BDB 
screws up again, this time in such a manner as to fail svnadmin 
recover. Awesome. So, how do I un-nuke data that shows the symptoms 
given in the original post? When can I expect that not to happen 
anymore?

- paul


---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org

Re: subversion 1.0.5, mod_dav_svn, gentoo, db4.1.25, and database corruption

[Shawn Erickson <sh...@freetimesw.com>]

On Jul 19, 2004, at 3:14 PM, paul. wrote:

> After taking everyone's screaming recommendation to use the https:// 
> method (because svn+ssh doesn't work)

svn+ssh doesn't work? How so?

-Shawn


---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org