-
-
Notifications
You must be signed in to change notification settings - Fork 427
Errno::EPIPE: Broken pipe #68
Comments
Hi, I see that too (ruby192, osx1067, guard, spork084) - primarily the very first time after I restarted the spork server. So it's a timing problem probably. |
Do you have |
I don't have any JS - I'm testing the following rack application: class MyApp
def call(env)
if env['PATH_INFO'] == '/home'
[200, {'Content-Type' => 'text/plain'}, ["Hello world!"]]
else
[200, {'Content-Type' => 'text/plain'}, [""]]
end
end
end |
I'm seeing a similar problem. Repeatbly, in exactly the same place every time in my test suite. When I bypass webkit and use selenium for the failing test, the broken pipe error occurs on the next webkit test. And if I bypass webkit on that one, the error occurs on the next webkit test.
|
Any updates on this? |
Curious about the updates too... |
just a thought, but how does parallelizing your tests affect capybara webkit, if it forks after capybara webkit server starts, you'll have two clients sharing the same socket, that would result in this failure? |
I still see this, just tried switching out selenium for capybara-webkit master on a project (one I'm working on with @hgimenez):
@halogenandtoast @jferris do you guys ever see this on an app? |
A broken pipe happens when the webkit_server binary crashes, so it always indicates a bug in capybara-webkit. Unfortunately, it's frequently difficult to find out what actually caused it to crash. What actually happens when you follow "Assign Now" in the above scenario? |
"Assign Now" is a placeholder link for a set of unassigned attributes on a Patient model. When you click it, the placeholder markup is replaced with an editable form in a Backbone.js view. The link is "#edit-family-history" and |
I have the same problem. If i run my I tried now for one hour to nail it down to a as simple as possible example. If there is any way how I can assist you to find the error, I'm happy to help. |
when using :webkit_debug we don't get these broken pipe errors.. (newly installed ubuntu 11.04 machine, mac os lion no such errors even with the normal :webkit driver) |
Thank you @carlhoerberg for this hint. I tried it. The first one still passes, starting from the second one it fails.
|
Also seeing these issues, but there's no rhyme or reason. The test suite's been running for a couple hours fine but then different steps start failing. Running the scenarios individually don't result in the exception. Running against thoughtbot/capybara-webkit HEAD.
|
I think this could be a duplicate of #39. Can you guys try again on 0.7.1? If you keep running into issues, please open a new issue and include output from :webkit_debug. |
I was having this issue on 0.6.1 (especially with devise paths (e.g.: visit login_path)). After upgrading to 0.7.2, the problem was solved. Just an fyi for the people coming to this page like I did. Thank you for making testing easier. |
We are still having the same issue with capybara-webkit 0.7.2 Any ideas greatly appreciated. Thanks. |
@mhoran Thanks, I'll try to give it a go :) |
Im running
then on the next example
|
+1 |
@kylewelsby, @haarts, please try out the latest master, bca84f9. If you continue to experience segfaults, please follow the instructions in #380. |
Rock on. Commenting out font-awesome worked for me. I used http://spin.atomicobject.com/2012/09/05/font-awesome-capybara-webkit/ |
My setup now
A few tests pass then,
followed by
|
@kylewelsby, could you open a new issue? It looks like you're experiencing something different than the others, and I don't want to clutter this thread. Please confirm that you are not using font-awesome. Then, please provide output from the failing scenario with the |
i might add a possible solution that seems to work for me for the moment. I discovered that there are times after running some specs that fail or crash that there might be lingering(orphaned?) webkit/rspec processes. I continue to get broken pipe while these processes exist even though they're not in the foreground anymore of where I ran them. I use the following command to clear everything out:
there's probably a more slick unix command, so feel free to update with a better way... |
please disregard issue as it is relevant to puma as well |
I'm getting this issue on my OS X Mavericks setup. Could this be because homebrew suggested to install QT using |
I had a similar issue as described above that I was able to solve by upgrading Qt |
I sporadically get the following error:
I am running my acceptance specs in parallel, if that is relevant.
The text was updated successfully, but these errors were encountered: