search

fpco / ide-backend

ide-backend drives the GHC API to build, query, and run your code
120 stars 17 forks source link

stdin: openFile: does not exist #208

Closed snoyberg closed 10 years ago

snoyberg commented 10 years ago

Not sure if this is related to any other issues we're seeing, but since it's coming from the RPC layer, I thought it was definitely worth reporting. Occasionally when running one of our test cases, I see the following kind of message:

ide-backend-server: /tmp/isolation-runner-workdir/projects/23790/session.3063/rpc.3265/stdin: openFile: does not exist (No such file or directory)

This particular test never succeeds right now (which is why I'm running it so frequently), so I can't be certain if that ide-backend-server error message is causing any issues.

snoyberg commented 10 years ago

Actually, this does seem to be pretty serious. When I run the test suite further, I get the client side throwing this exception, killing our test run:

uncaught exception: ErrorCall (expectSome: terminated with PRBackendError "/tmp/isolation-runner-workdir/projects/23794/session.14392/rpc.14435/stdin: openFile: does not exist (No such device or address)"

Obviously you can ignore the expectSome and PRBackendError bit, those are from our code base. This is with commit 72278db59724765d9cdf052b80505e8c60abf175.

snoyberg commented 10 years ago

Actually, that seems to have just been a mismatch between ide-backend and ide-backend-server versions, ignore this.

snoyberg commented 10 years ago

Even after fixing the ide-backend-server conflict, this problem is still occurring occasionally.

snoyberg commented 10 years ago

I'm sufficiently uncertain of this one that I'm going to close it again. I think upgrading to the newest commit from fork-snippets has resolved this issue.

edsko commented 10 years ago

@snoyberg Might this be the exception that we are now throwing if the server has deadlocked?

snoyberg commented 10 years ago

Yeah, that would make sense. OK, closing, sorry for the noise.