WebApp Error: <type 'exceptions.SystemExit'>

Issue #240 resolved
Former user created an issue

I got this error a few times. The callstack is always:

{{{ #!python Module weberror.errormiddleware:242 in next

return self.app_iterator.next() Module mercurial.hgweb.protocol:59 in groupchunks chunk = cg.read(4096) Module mercurial.changegroup:143 in read return self._stream.read(l) Module mercurial.util:825 in read for chunk in self.iter: Module mercurial.util:803 in splitbig for chunk in chunks: Module mercurial.localrepo:1670 in gengroup filerevlog = self.file(fname) Module mercurial.localrepo:579 in file return filelog.filelog(self.sopener, f) Module mercurial.filelog:35 in init "/".join(("data", path + ".i"))) Module mercurial.revlog:251 in init }}}

I haven't updated RhodeCode for about 4 weeks.

Comments (9)

  1. Marcin Kuzminski repo owner

    probably process gets kileld because of large cpu usage ? is it on vps ? how much ram does it have, what repo, how many changesets, on what view it crashes ?

  2. Marc Sanfaçon

    It is during a pull operation, I think. Here is other information:

    CGI Variables
    AUTH_TYPE	'basic'
    CONTENT_LENGTH	'0'
    HTTP_ACCEPT	'application/mercurial-0.1'
    HTTP_ACCEPT_ENCODING	'identity'
    HTTP_AUTHORIZATION	'Basic bnBlbGxldGllcjo4N1pvcmdsdWI='
    HTTP_CONNECTION	'Keep-Alive'
    HTTP_HOST	'mercurial'
    HTTP_USER_AGENT	'mercurial/proto-1.0'
    HTTP_VARY	'X-HgArg-1'
    HTTP_VIA	'1.1 mercurial'
    HTTP_X_FORWARDED_FOR	'192.168.70.59'
    HTTP_X_FORWARDED_HOST	'mercurial'
    HTTP_X_FORWARDED_SERVER	'mercurial'
    HTTP_X_HGARG_1	'common=0000000000000000000000000000000000000000&heads=0ae96b82879153f8489542192d59fbda8d25fb80+655ccbcb68380b85f3ac0103e2c8fe5c78d92f3d'
    PATH_INFO	'/'
    QUERY_STRING	'cmd=getbundle'
    REMOTE_ADDR	'192.168.64.78'
    REMOTE_USER	'user1'
    REQUEST_METHOD	'GET'
    SERVER_NAME	'192.168.64.78'
    SERVER_PORT	'5000'
    SERVER_PROTOCOL	'HTTP/1.1'
    WSGI Variables
    application	<rhodecode.lib.middleware.simplegit.SimpleGit object at 0x390fbd0>
    paste.httpserver.thread_pool	<paste.httpserver.ThreadPool object at 0x7fe2d3f95c10>
    paste.parsed_querystring	([('cmd', 'getbundle')], 'cmd=getbundle')
    paste.registry	<paste.registry.Registry object at 0x7fe2c462a050>
    paste.throw_errors	True
    pylons.status_code_redirect	True
    wsgi process	'Multithreaded'
    

    It is not running in VPS. The repo is about 3.2GB and has 18,000 changesets.

    paster memory ranges from 150MB to 1024MB when I check it.

  3. Marcin Kuzminski repo owner

    does it raises same exception on pure hgweb or pure paster without http server ? And after aprox what time it happens I seen this error on my server but it was related to OS that killed process that was running for long time, don't know if this is the case here.

  4. Marc Sanfaçon

    Good question. We don't have hgweb setup. We only use Rhodecode for our Mercurial server. We would need to setup hgweb and see if it causes the same problem. Not sure we want to do that... We actually moved away from hgweb because of the additional features of Rhodecode.

  5. Marcin Kuzminski repo owner

    you don't need to setup hgweb. Running hgserver on a repo is enought to make hgweb instance running on port 8000(default)

  6. Former user Account Deleted

    I finally had time to work on this a bit. I think I pinpointed the problem.

    Looks like the Apache proxy rules forget, sometimes, to pass the exact path of the repo. In the above error, the PATH_INFO is '/', but there are no repo in this path. It misses the repo path.

    Maybe it is my setup that is wrong. Is there another way to setup Rhodecode in https?

  7. Log in to comment