> hello: > > > I use uwsgi backlog alarm option --alarm-listen-queue, and set > listen option to 1024, but backlog alarm trigger again and again, I > use different alarm action to try, for example 'cmd' and > uwsgi-sentry pulgin. but the backlog is not full obviously . the > sentry report is > > > listen queue of socket "127.0.0.1:8883" (fd: 3) full !!! (0/1024) > > > or I use cmd action for example > > > alarm = lqfull cmd:curl "http://test.server"; > alarm-listen-queue = lqfull > > > and the test.server also receive many alarms > > > my config like this > > > [uwsgi] > auto-procname = true > procname-prefix-spaced = xxx > socket = xxx.sock > #http-socket = 127.0.0.1:8883 > master = true > wsgi-file = app.py > touch-reload = app.py > uid = mine > gid = mine > processes = 4 > gevent = 20 > max-requests = 10000 > listen = 1024 > daemonize = xxx.log > disable-logging = true > env = PYTHON_EGG_CACHE=/tmp > env = FP_ENV=local > callable = app > plugins-dir = /usr/lib/uwsgi > plugin = python27,gevent,sentry > > > alarm = lqfull cmd:curl "http://xxxx"; (or sentry config) > alarm-listen-queue = lqfull > > > so what is the reason? thanks > > _______________________________________________ >
hi for unix socket backlog monitoring you need the uwsgi-netlink plugin: https://github.com/unbit/uwsgi-netlink unfortunately it looks like the alarm system is not smart enough to detect the listen queue is not correctly managed -- Roberto De Ioris http://unbit.com _______________________________________________ uWSGI mailing list [email protected] http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi
