fisheye cannot connect to mercurial repos over https using self-signed cert on server

[Ianto McGuckin]

I am trying to connect mercurial repos to fisheye.  The server that mercurial is running on is using a self-signed certificate for serving the mercurial repositories, and when I try to add the repo in fisheye (using password for https authentication and supplying the fisheye admin password for this), and testing the connection to the repo I get the error message: 

Error testing connection.
Non-zero exit code: 255
at com.atlassian.utils.process.PluggableProcessHandler.complete(PluggableProcessHandler.java:69)
at com.atlassian.utils.process.ExternalProcessImpl.wrapUpProcess(ExternalProcessImpl.java:509)
at com.atlassian.utils.process.ExternalProcessImpl.finish(ExternalProcessImpl.java:423)
at com.atlassian.utils.process.ExternalProcessImpl.execute(ExternalProcessImpl.java:378)
at com.atlassian.fisheye.dvcs.DvcsScmConfig.executeCommand(DvcsScmConfig.java:139)

Connection failed.

As far as I can see, everything is configured correctly.

Mercurial appears to be configured correctly

Mercurial executable (hg) = /Program Files/TortoiseHg/hg.exe

Does the stack trace above point to anything useful I could try next?

[Themis]

Hi Ianto,

a quick search using the stack trace lead to this issue:

Test Connection fails for some mercurial repositories

It suggests that while the test connection reports a failure, actually adding/saving the repo results in correct indexing. Could you try adding the repos despite the failed tests and check whether they begin indexing?

Best,

Themis

[Tamas Csabina]

Have you guys ever figured this out? I have a similar (same?) problem while testing a newly added mercurial repository.

The error message in my case:

Error testing connection.
process timed out
at com.atlassian.utils.process.ExternalProcessImpl.wrapUpProcess(ExternalProcessImpl.java:506)
at com.atlassian.utils.process.ExternalProcessImpl.finish(ExternalProcessImpl.java:423)
at com.atlassian.utils.process.ExternalProcessImpl.execute(ExternalProcessImpl.java:378)
at com.atlassian.fisheye.dvcs.DvcsScmConfig.executeCommand(DvcsScmConfig.java:139)
at com.atlassian.fisheye.hg.HgScmConfig.testConnection(HgScmConfig.java:125)

Connection failed.

If I ignore the error and just Save the settings, the repository clone is at the first changeset:

Info

Indexing commit 8ebd8aaa9bfc5e469a1ff5fce93c9c8bebd3ce80

The "Test" under "SCM details" for other repositories that were added sooner, works fine. No timeout, but "Connection succeeded".

If I clone the repository manually on the Fisheye server (windows command line) with the same username and password, it works. So it must be the Fisheye front/backend, in my opinion.

Any tips?

[Tamas Csabina]

This might help (though not really making sense to me).

After several failed attempts to Test the connection, I just saved the configuration.

As mentioned, there was no indexing happening, as the indexer just stuck at the first commit.

I did "Re-clone and Re-index:" in the "Maintenance" menu, and all of a sudden the indexer started...

[Tamas Csabina]

False alarm...

I accidentally saved a different URL (different repository) in this new repo. And that is an URL of a working repo.

So the "Re-clone and Re-index" trick is irrelevant. I' am still struggling to add this new repo...

[Tamas Csabina]

I finally managed to solve this (at least, partially).

The repository I tried to configure in Fisheye contained couple of thousand files!

After removing those (also from the history, by using the 'convert' extension of mercurial), I was able to clone and index the repository.

I hope this helps someone.