I try to checkout my repository user/bar on GitHub using the cli-version of svn. This fails with error 500.
Note: Similar questions have been asked before, but are either slightly different or have (for my case) unsatisfying answers:
Error Connecting to GitHub from SVN got a different error message
git svn gives 500 internal server error got the same error, but only with git svn clone not with native svn
Subversion Unexpected server error 500 on checkout had exactly the same problem, but it seemed to have been a temporary problem on GitHub
Community Forum describes exactly the same problem, but got no response
GitHub Documentation says: make an empty checkout of the repository,
$ svn co --depth empty https://github.com/user/repo
> Checked out revision 1.
$ cd repo
So I tried to check out my repositories of foo (created months ago) and bar (created recently), and - as a cross check - a non-existing repository baz:
$ svn co --depth empty https://github.com/user/foo
Checked out revision 4949.
$ svn co --depth empty https://github.com/user/bar
svn: E170013: Unable to connect to a repository at URL 'https://github.com/user/bar'
svn: E175002: Unexpected server error 500 'Internal Server Error' on '/user/bar'
$ svn co --depth empty https://github.com/user/baz
Authentication realm: <https://github.com:443> GitHub
Password for 'user': ****************************************
svn: E170013: Unable to connect to a repository at URL 'https://github.com/user/baz'
svn: E160013: '/user/baz' path not found
Findings:
basically it works, otherwise foo would not be fine
bar exists, otherwise the error would would be like baz
A temporary problem seems unlikely (becaus foo works), but to be sure I've waited for a couple of days - no change. Configuration of foo and bar is identical (to the best of my knowledge).
Any more ideas?
More than a week after the original attempt it suddenly works as expected. So if you happen to run into the same problem:
it might affect only selected repositories
the temporary failure might last for quite some time
Related
I used to be able to fetch projects from Github using SVN externals:
svn:externals > https://github.com/J7mbo/twitter-api-php/tree/develop twitter-api-php
Github had previously announced support for SVN, and this blog posts details the use of Github projects via SVN externals.
This worked just fine previously, but I've noticed recently (the past couple of weeks?) that I can no longer fetch any project from Github: I get a Unable to connect to a repository at URL *** error. Other externals seem to be working as normal. I can't find any documentation that suggests they've dropped or changed this feature. Does anyone know why this is no longer working, or if they've dropped support?
UPDATE:
Running
svn ls https://github.com/J7mbo/twitter-api-php/branches/develop
returns the following error:
svn: E175002: OPTIONS of 'https://github.com/J7mbo/twitter-api-php/branches/develop': SSL handshake failed: SSL error: tlsv1 alert protocol version (https://github.com)
I'm guessing that's where my problem lies: my openssl version is LibreSSl 2.27, but I can't establish which version of TLS it uses, and in any instance I'm struggling to update openssl.
The svn:externals property is evaluated by your local SVN-Client, not by Github. Therefore as long as Github support the SVN-bridge, svn:externals will work.
But your URL seems to be wrong:
> svn ls https://github.com/J7mbo/twitter-api-php/tree/develop
svn: warning: W160013: URL 'https://github.com/J7mbo/twitter-api-php/tree/develop' non-existent in revision 137
svn: E200009: Could not list all targets because some targets don't exist
This URL seems to work:
> svn ls https://github.com/J7mbo/twitter-api-php/branches/develop
.gitignore
.travis.yml
LICENSE.md
README.md
TwitterAPIExchange.php
composer.json
index.php
phpunit.xml
test/
I have facing SVN problem in Mac OS X ..
When i am commit any file below are the error are generate.
svn: Commit failed (details follow):
svn: Can't get exclusive lock on file '/Volumes/SvnProject/db/txn-current-lock': Operation not supported
On the SVN server, ensure that the owner and group of the repository (and all subfolders) are properly set.
For example, I created a repo by issuing the following from the SVN server as root:
svnadmin create /path/myrepo
Initially I was getting the txn-current-lock error from my SVN client. On the SVN server, I noticed that /path/myrepo and all of its subfolders had owner:group of root:root. To fix this, I issued the following from the SVN server:
chown -R newowner:newgroup
Ensure your newowner and newgroup items match the owner/group information being supplied by the SVN clients.
Important to make it recursive with the -R option. This fixed the txn-current-lock error.
I'm having an issue with setting up jenkins for build automation, we use the source control host codebasehq, with mercurial, however it is failing to pull from the repo.
The plugin seems to connect fine with the supplied credentials (at the moment we're simply using the in http url credentials) but then gives the following error:
Building in workspace /home/ubuntu/jenkins/jobs/testJob/workspace
$ hg clone --rev tip --noupdate https://foo:bar#initech.codebasehq.com/foo/bar.hg /home/ubuntu/jenkins/jobs/testJob/workspace
abort: partial pull cannot be done because other repository doesn't support changegroupsubset.
I've tried running the command in the error message, but it simply returns the same abort message.
I have limited experience with mercurial, and google doesn't seem to return much on the error message, which makes me think it might either be an issue with the jenkins plugin, or codebasehq's mercurial setup.
Any information on the error or if anyone has experience with setting up this kind of thing on codebasehq successfully and could share any advice would be fantastic.
After a response from the codebasehq support team, it seems that 'changegroupsubset' refers to a feature that is unsupported by the https version of their mercurial repository. It is however, supported by the ssh version of their repository, so I was able to fix this by simply replacing
hg clone --rev tip --noupdate https://foo:bar#initech.codebasehq.com/foo/bar.hg /home/ubuntu/jenkins/jobs/testJob/workspace
with
hg clone --rev tip --noupdate ssh://foo:bar#initech.codebasehq.com/foo/bar.hg /home/ubuntu/jenkins/jobs/testJob/workspace
And this worked successfully.
Well I didn't do a correct backup of my settings when I updated Team City and somehow lost all of my VCS roots but the rest of my settings for my projects are still there.
When I re-add my mercurial root, I test the connection and it succeeds, however when I try to run a build, it fails on checkout. Anyone know what this error means?
[Updating sources: agent side checkout...] Failed to perform checkout on agent: 'cmd /c hg pull <MERCURIAL URL>' command failed.
stderr: transaction abort!
rollback completed
abort: connection ended unexpectedly
stdout: pulling from <MERCURIAL URL>
requesting all changes
adding changesets
adding manifests
adding file changes
This was working before I mistakenly updated without backing up :(
I am using agent side checkout with a specified checkout directory (just like I was using before the update)
Any help would be greatly appreciated.
[EDIT]: I updated tortoise hg / mercurial on the build server and got this error when trying to pull:
% hg --repository D:\Mercurial\Core pull --verbose https://<user email login>:<pw>#<kilnhg respository url>
pulling from https://<user email login>:<pw>#<kilnhg respository url>
requesting all changes
adding changesets
adding manifests
adding file changes
transaction abort!
rollback completed
abort: connection ended unexpectedly
[command returned code 255 Thu Aug 11 15:17:17 2011]
I just tried again from tortoise hg and the pull succeeded. This is very strange :-/
I've had the exact same problem. I have tried all available versions of TortoiseHG on my Windows 7 (64) machine. I can't clone or pull anything from anywhere :)
What solved it for me is the "Use uncompressed transfer" check box in TortoiseHG gui.
Equal to "hg clone --uncompressed"
I noticed a subtlety in your error message. You're using Kiln, right? I think this issue is explained on Kiln's support site. Can you manually pull from the repo to slowly pull chunks of revisions to see if it finishes with smaller sets using hg pull -r 200 as shown on Kiln's support site?
I get the following error when trying to clone from the production repo.
abort: index data/HR3/globals.php.i is corrupted!
After which mercurial promptly farts out on me and fails to clone. I've tried removing the file in question from the production repo, and that failed. I also tried deleting the .hg directory in the production repo and re-adding and committing all of the files. This produces the same error.
Just fyi, I'm trying to clone the current production copy of our app to our test server. Our test server is running in Virtual Box.
UPDATE
I fixed the issue by deleting the globals file and then deleting the .hg directory and re-running "hg init" and "hg add ." and "hg commit."
Now I get an Internal server error when I try to clone. The following is the clone command I used.
sudo hg clone http://10.1.1.25/ www
Oddly enough, mercurial tells me that the "real" url is "10.1.1.25/HR3/index.php" which is most definitely not my mercurial repository. I do not know why it is doing this. I checked in my server logs, and the following error is what Apache crapped out.
[Thu Dec 16 12:25:30 2010] [error] [client 10.1.1.102] PHP Parse error: syntax error, unexpected T_STRING in /var/www/.hg/store/data/_h_r3/includes/_b_n_a_s/_b_n_a_s-_h_r_s201.php.i on line 22
Apparently the clone command is causing Mercurial to kick off PHP and make it try to parse something in Mercurial's data store?
RESOLUTION
I suspect this was caused by the fact that I have an index.php file at the root directory of my repository that does redirects. If you try and navigate directly to http:10.1.1.25 on our network, you're automatically directed to the index page, which is not the index.php page in the root directory. Somehow I think that this was causing php to kick off and try to parse items in the mercurial store. I've resolved this issue by cloning over ssh.
I have faced such an error before. Try to run
hg verify
on your corrupted repository. Then follow the instructions from here:
http://www.softwareprojects.com/resources/programming/t-how-to-fix-mercurial-corrupt-repository-1926.html
It seems that you publish your repo with the static-http method, from the same server as your development project lives. This method only works if the web server does not mess with $URL/.hg, which your one seems to do. check if you can download the .hg folder with wget -r -np $URL/.hg in a fresh folder and run hg verify there.
Oddly enough, mercurial tells me that the "real" url is "10.1.1.25/HR3/index.php" which is most definitely not my mercurial repository.
I guess that your web server redirects some mercurial requests to your index.php url.
You may try to do that:
hg forget HR3/globals.php
hg commit
hg add HR3/globals.php
hg commit