We have a staging server that I threw Phabricator on to test it out and see how it would function. I like it for the most part and despite this error it seems to be working ok.
Since this is our staging server, I already had a working Mercurial repo that is locally available. I looked around and couldn't find anywhere where I could tell Phabricator to point to a local repo, so I did a hack by using an ssh key to localhost (ssh://user@127.0.0.1//path/to/repo) as a work around and that seems to have done the job. It's only in Observe Mode, so I figured it should be safe to mess with.
After the repo fully imported, I now get the following error when I view the status of the repo:
Error updating working copy: Failed to proc_open(): proc_open(): unable to create pipe Too many open files
Here's a screenshot, just because I can :)
That error has been there ever since and hasn't gone away.
When I view what's going on with the server with htop, I just see a wall of hg commands (that's it, just hg) and a bunch of /usr/bin/python /usr/bin/hg log --template={node} --rev 'some rev hash' The cpus all peg at 100% while this happens (every 15 seconds or so like the setting in the screenshot) and then it goes away after about 5-10 seconds. I could push the code to bitbucket and then just pull from bitbucket but wanted to see if I could resolve this first this way.
My question ultimately is, is my workaround causing this or is this a bug that needs to be looked at? Is this expected behavior on a Phabricator server?
Some other details that may be helpful:
CentOS 6.8
PHP 5.6.24
Mercurial 3.9.1
Staging Repo has about 10,000 commits and 1,000 or so branches.
phabricator b521f2349e46888514ce55d06ace1d8c5fa6df03 (Mon, Aug 22)
arcanist 89e8b48523844cc3eff8b775f8fae49e85f8fc22 (Fri, Aug 19)
phutil 237549280f08feb8852857b1d109dfb0aba345f0 (Mon, Aug 22)