>From documentation [1], it looks like rabbit.tcp_listen_options.linger is
set to {linger, {true,0}} by default.1. Does anybody know if this is indeed the default option ? Unless explicitly set in config, rabbitmqclt environment does not list the option. 2. Is anybody explicitly setting it to 0 ? Without that option turned on in a rabbitmq cluster setup (mq1 and mq2 for example), If mq1 goes down and openstack service was connected to mq1 then it automatically switches to mq2 but connection to mq1 remains in close_wait state in lsof output for a while. And it does affect some nova boot requests. If linger is explicitly set to 0, we do not see any connections left in close_wait state. Anybody else observing this behavior ? We are using juno code and oslo.messaging-1.4.1. Thanks, Meghal [1] - https://www.rabbitmq.com/networking.html
_______________________________________________ OpenStack-operators mailing list [email protected] http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
