Closed coneybeare closed 10 years ago
did you tried running just the command on server:
/usr/local/rvm/bin/rvm ruby-2.0.0-p353@[REDACTED] do ruby --version
I just ran it and only 2 wrapper instances were spawned, not the same behavior as when running through cap
and in case it wasn't clear, the version string was printed to console
what is surprising is that it's called at all, check if you have any executable hooks in rvm:
ls -l $rvm_path/hooks
those with executable bit are enabled (x
), most of the hooks is disabled by default, but some users report that running chmod -R +x $rvm_path
does it .. the easiest is to disable all the hooks with: chmod -x $rvm_path/hooks/*
They are all default installed hooks. The executables are after_install_auto_gem
, after_install_codesign
and after_use_maglev
.
I have been tinkering around with the server side and reinstalled rvm, reinstalled ruby-2.0.0-p353, manually redid a bundle install and now the error is gone. The bug, if exists, could still be there, but I can no longer reproduce it. If the source is not able to be determined, perhaps a local check for the symptoms could be added. I only noticed because it slowed down every other site on my server, then I realized there were tens of thousands of rvm wrapper instances happening. It seems like it could have been prevented with a default timeout on the local side.
maybe this was older version of rvm?
I don't think that is the case, I ran a rvm get head yesterday
if anyone can reproduce it then please report a ticket for rvm => https://github.com/wayneeseguin/rvm/issues
@coneybeare I would recommend you switching to rbenv which is super lightweight and don't have such many issues like overriding cd
command.
@kirs since when overriding cd
is a problem? it is totally unrelated to this problem, and the rvm override of cd
works perfectly fine, you could find a lot of issues in rvm but you had to mention not-an-issue.
I am generally pretty comfortable running RVM so I will stick with it, but thanks
@coneybeare there is alternative (better) implementation of the integration that should work better https://github.com/rvm/rvm1-capistrano3 ... I'm missing time to open a PR to merge it over here.
I am running a simple
cap production rvm:check
.Locally, everything looks good until it hits here:
ruby-2.0.0-p353
is installed on the remote machine, as is the gemset.On the server side, hundreds of rvm wrapper instances are spawned per second with no end. Only a force kill locally, and a pkill remotely will cause it to stop.
I never had a problem with cap 2. What is going on here?