ERROR! [pid 28651] 18:23:43.174203 E mitogen.ctx.local.28788: mitogen: <IoLogger stderr> crashed #323
Comments
The fix for this (f7e288f) is on master branch and will make it into the next release. To be updated when a new release is made, subscribe to https://networkgenomics.com/mail/mitogen-announce/ Thanks for reporting this! |
Sorry -- that fix is for a separate issue. Sorry for closing this prematurely. I'll investigate in the coming days. |
No sweat, thanks for looking into it at all. Honestly, the playbook appears to have executed successfully. It just seems to be hitting a snag while cleaning up connections. If I can provide any additional information to make troubleshooting easier then let me know. |
You might try a re-run with the latest master, but that stdio fix was for a probably unrelated issue -- subprocesses generating a huge amount of IO could receive errors because the file descriptors were in the wrong mode. This looks more like WSL is reporting an incorrect errno when the write end of the stdio forwarder pipe is shut down, but I'll need to reproduce it under WSL to be sure. Thanks again for the report! |
I attempted this using the latest master this morning and got the same result. |
Now I've had time to think about it, this looks suspiciously like a Linux variant of https://github.com/dw/mitogen/issues/320 Both OS X and Linux pollers have an asynchronous event queue decoupled from FD lifetime, so it's very likely the same bug. If it is simple for you, would you mind trying this: edit "mitogen/parent.py" and delete line ~667, where it reads " 'Linux': EpollPoller," and rerun your job. Otherwise just wait a few more days and I'll get around to this. If this is indeed the case, then the bug is basically a manifestation of #26, which entails a poller interface change. |
Hi there, How many targets are part of your Ansible run? |
There is only one actual target, though the target belongs to a number of groups. IE
|
Commenting out line 659 of master
Did not resolve the error
|
Thanks for trying it out! So that would seemingly absolve the poller of blame. I'll try reproducing this over the weekend locally on WSL. Can you tell me which version of WSL/Windows you have? I'm not too familiar with WSL -- but at least include "uname -a" output. There should be a build ID visible somwhere, but I can't find anything on Google about where to locate it. |
Linux {hostname} 4.4.0-17134-Microsoft #137-Microsoft Thu Jun 14 18:46:00 PST 2018 x86_64 x86_64 x86_64 GNU/Linux I keep my WSL pretty up to date, and am running the very latest version of windows 10. Windows seems to update the subsystem with windows updates, because the subsystem always behaves a tiny bit different (usually more like linux) each time I get a major windows patch. A little more information about what I'm doing.
|
Sorry the very long delay on this one - I've finally got a WSL setup :) This is a WSL bug, and a duplicate of #333, for which I've already started generating commits. So closing this one off as a duplicate. Thanks for reporting this! |
This is now on the master branch and will make it into the next release. To be updated when a new release is made, subscribe to https://www.freelists.org/list/mitogen-announce Thanks for reporting this! |
Thank You! I'll definitely pull the master branch sometime next week (currently out of the office) and give it a go. |
I noticed today, after making some modifications to my playbook to include the XML module instead of lineinfile, that after the playbook has finished its play recap it was printing error messages out of mitogen.
The attached message is printed when running the playbook with -vvvv, and is printed AFTER my prompt has already returned. In fact, the
zamrok@laptop -> ~/rigors/ansible-scripts/ansible-install:
portion is my prompt.Environment:
Ansible Host: Windows Subsystem Linux
Ansible Target: Centos 7 running on Virtualbox
The text was updated successfully, but these errors were encountered: