Reputation: 26732

SVN: Could not read status line: connection was closed by server

I started getting this error today on my subversion repo when I try to update. Any suggestions?

svn: OPTIONS of 'http://example.org/example_repo': Could not read status line: connection was closed by server (http://example.org)

It appears that when I use the svn+ssh instead of HTTP, it works.

Upvotes: 29

Answers (15)

JaydenLiu

Reputation: 1

I've found Nick Urban's comment very useful:

sudo /opt/cisco/anyconnect/bin/websecurity_uninstall.sh

Upvotes: -1

Miracle

Reputation: 49

Changing from http to https worked for me

Upvotes: 1

chris

Reputation: 531

I had the same problem. It turned out to be a conflict with the Cisco Anyconnect Client's Web Security module. I realized it because of some of the console messages I was seeing, did a search and found this: http://www.thebitguru.com/blog/view/394-Random%20Slowdown%20of%20Browsers%20in%20OS%20X%20Mountain%20Lion

Put it together in my head:

removed cisco anyconnect, success
reinstalled anyconnect without the "Web Security" module, success.

It was befuddling because it had worked fine before.

Upvotes: 53

ramrunner

Reputation: 1372

Well as anyone in this thread, same problem , different cause.

i had to change the SVNPath option to SVNParentPath . The reason is that SVNPath handles only one repository and it prefixes this to the url that the browser request. But on my setup i had a directory with all the svn repos in it and that gets handles by SVNParentPath. Below is my location directive for apache2.

<Location /svn/>
    DAV svn
    SVNParentPath /svn
    AUTHType Basic
    AUTHName "foo"
    AuthUserFile /path/to/passwd
    AuthzSVNAccessFile /path/to/authz-svn-access
    Require valid-user
</Location>

Upvotes: 0

user2045106

Reputation: 31

I usually don't post, but my team spent 12 man hours troubleshooting this.

For us this also turned out to be a conflict with the Cisco AnyConnect Client

Upvotes: 3

Steve King

Reputation: 71

This happened to me after my Cisco VPN client (AnyConnect Secure Mobility Client) updated itself. I fixed it by uninstalling and reinstalling the client with the options shown on this blog post:

http://www.thebitguru.com/blog/view/394-Random%20Slowdown%20of%20Browsers%20in%20OS%20X%20Mountain%20Lion

Upvotes: 7

sladda

Reputation: 442

Got the same problem. Even when i tried to relocate, the SAME error appeared.

My Solution: I opened the repo browser. After that, the error was gone while updating and committing. Dont ask me why :)

Upvotes: 0

Joro Mitev

Reputation: 31

I did changes to the certificates of my HTTPS server, and as it is said above, something with the caches went wrong with my repository.

I relocated the repository in the same URL and the error disappeared. (backup first)

Upvotes: 3

Bob Gustafson

Reputation: 31

I had the same problem (but I may have had a different configuration setup - seemed to be plain vanilla though)

The lines below show my last error, and my 'fix'. (Thanks to above suggestions - Joshua)

[user1@hoho6 RubymineProjects]$ svn checkout svn://localhost/home/user1/DummySVNrepo
svn: URL 'svn://localhost/home/user1/DummySVNrepo' doesn't exist

[user1@hoho6 RubymineProjects]$ svn checkout svn+ssh://localhost/home/user1/DummySVNrepo

The authenticity of host 'localhost (::1)' can't be established.RSA key fingerprint is 10:8d:10:04:00:02:b1...
Are you sure you want to continue connecting (yes/no)? yes
user1@localhost's password: 
user1@localhost's password: 
A    DummySVNrepo/test
A    DummySVNrepo/test/unit
... etc

Upvotes: 0

Steve

Reputation: 21

Had the same problem. For me, in the end, it turned out to be (probably) something to do with having messed with my apache virtual hosts, and the SSL certificate for the server (I tidied up a few of my virtual hosts and replaced my snakeoil certificate with a genuine one - the latter I suspect was the real cause of my problem).

Solution: What worked for me was to clear out any saved passwords in the clients (I was using Eclipse - subclipse or subversive, I forget which - and Tortoise). I guess that some sort of hashing based on the SSL certificate was being used somewhere along the line to encrypt the saved password, which rendered them inaccessible or invalid when I replaced the certificate.

Upvotes: 2

jamieb

Reputation: 10033

This problem was also occurring for me. The cause ended up being the transparent web proxy (Squid) that we had set up. The firewall here is configured to silently redirect all port 80 traffic through Squid. Adding an exception to the hostname of the SVN server in the firewall config fixed the issue.

Upvotes: 1

Axel

Reputation: 21

My problem was that I have my path as http://.... and I'm working with file system, then my SVN path must be file://, I try a rellocate and put it as file:// and it is working well.

Upvotes: 1