splash: Fail to restart automatically using docker + splash

I use the following command to deamonize the process ‘docker run -d -p 8050:8050 --restart=always -v /etc/splash/proxy-profiles:/etc/splash/proxy-profiles scrapinghub/splash:1.5 --maxrss 500’ but whenever the splash process crash (it gets killed by the system because it takes too much memory even if i have 1GB on the server) docker fails to restart it with the following error:

Traceback (most recent call last):
  File "/app/bin/splash", line 3, in <module>
    from splash.server import main
  File "/usr/local/lib/python2.7/dist-packages/splash/server.py", line 12, in <module>
    from splash.qtutils import init_qt_app
  File "/usr/local/lib/python2.7/dist-packages/splash/qtutils.py", line 12, in <module>
    from PyQt4.QtCore import (QAbstractEventDispatcher, QDateTime, QObject,
ImportError: /usr/lib/python2.7/dist-packages/PyQt4/QtCore.so: cannot read file data: Input/output error

About this issue

Original URL
State: closed
Created 9 years ago
Comments: 15 (9 by maintainers)

Commits related to this issue

--cleanuprss option to clear caches when memory reaches a certain limit. See https://github.com/scrapinghub/splash/issues/216#issuecomment-97890450 — committed to scrapinghub/splash by kmike 9 years ago

Most upvoted comments

Hm, I don’t know what can cause this error. Googling for ‘Input/output error’ suggests a hardware failure, but I don’t have an experience with it myself.

One problem which may be related is that Splash can sometimes segfault on shutdown (it seems because C++ destructors order doesn’t match Python objects deallocation order), but here the error is at startup, not at shutdown.

kmike on Apr 30, 2015