[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Svn server core dumps when user connects with svn+ssh for the fir st time

From: Lee Jason <Jason.Lee_at_HCAhealthcare.com>
Date: 2004-12-28 19:49:05 CET

Hello,

I have run into what I believe is a bug, and one that I have consistently
been able to reproduce. I have built SVN on several servers, and whenever a
client connects to the repository server (using svn+ssh) from a freshly
built svn client on a separate server, the repository server core dumps. I
then end up having to run a "svnadmin restore." This only occurs when a
client connects for the -first time- from a freshly built client. In other
words, subsequent connections from the client or other clients on the same
remote system do not result in my repository crashing.

I believe I have a sane build of SVN in all cases, as running a "make check"
shows all tests completing successfully.

I performed my builds on Solaris 9 (Sparc). The versions of the relevant
components are as follows:
Subversion-1.1.2
OpenSSH_3.7.1p2
OpenSSL 0.9.7d
autoconf-2.59
db-4.2.52
libiconv-1.8
libtool-1.5

My build was under the following environment:
LDFLAGS='-R/usr/local/lib -R/usr/local/ssl/lib
-R/usr/local/BerkeleyDB.4.2/lib -L/usr/local/lib -L/usr/local/ssl/lib
-L/usr/local/BerkeleyDB.4.2/lib'

And my 'configure' was simply the following:
./configure --with-berkeley-db=/usr/local/BerkeleyDB.4.2 --with-ssl
--mandir=/usr/local/man

Although I am new to using Subversion, I don't believe this problem is a
user error. After recovering the repository, the connecting clients which
crashed my svn server have no issues after their first episode.

When connecting to the repository with the "svn+ssh://" directive, the
following is displayed in my syslog (note, two separate events shown here):

Dec 28 07:23:39 netopweb.mgmt.medcity.net sshd[14204]: [ID 800047 auth.info]
Accepted password for leeja from 10.26.31.4
1 port 45515 ssh2
Dec 28 07:23:40 netopweb.mgmt.medcity.net genunix: [ID 603404 kern.notice]
NOTICE: core_log: svnserve[14207] core dumped
: /var/core/core.svnserve.14207.netopweb.mgmt.medcity.net.101.1.1104240220

Dec 28 11:48:18 netopweb.mgmt.medcity.net sshd[14546]: [ID 800047 auth.info]
Accepted password for leeja from 10.26.31.5
4 port 1023 ssh2
Dec 28 11:48:18 netopweb.mgmt.medcity.net genunix: [ID 603404 kern.notice]
NOTICE: core_log: svnserve[14549] core dumped
: /var/core/core.svnserve.14549.netopweb.mgmt.medcity.net.101.1.1104256098

Honestly, I have not explored other connection methods as the svn+ssh seems
like the best fit for my environment. I would add that I was able to use
the repository for hours when connected locally using the (file:///
<file:///> ) directive without any problems. But the very instant I
connected from a remote server using svn+ssh the svn server went south.

Thanks,
Jason

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Tue Dec 28 19:52:31 2004

This is an archived mail posted to the Subversion Users mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.