[
https://issues.apache.org/jira/browse/DISPATCH-952?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16423764#comment-16423764
]
Robbie Gemmell commented on DISPATCH-952:
-----------------------------------------
{quote}...and connections to brokers, which could have > 32k links. I think the
fix is to use a single shared session on such connections. Since traffic on a
single connection gets serialized in any case, there's no real benefit to
multiple sessions on a connection.
{quote}
It can to some peers, e.g. broker-j will see improved performance from multiple
sessions when persistent messages are involved.
Its also worth noting that the protocol actually allows 65k sessions, so
proton-c is artificially limiting things there it seems (presumably due to
magic-bit usage).
> qdrouterd seg fault after reporting "too many sessions"
> -------------------------------------------------------
>
> Key: DISPATCH-952
> URL: https://issues.apache.org/jira/browse/DISPATCH-952
> Project: Qpid Dispatch
> Issue Type: Bug
> Components: Container
> Reporter: Alan Conway
> Assignee: Ganesh Murthy
> Priority: Major
> Fix For: 1.1.0
>
>
> Reported at [https://bugzilla.redhat.com/show_bug.cgi?id=1561876]
>
> {code:java}
> Currently running Satellite 6.3 with 5K clients. The clients are managed by 2
> capsules:
> Capsule 1: 3K clients
> Capsule 2: 2K clients
> Logs from Capsule 1:
> [root@c02-h10-r620-vm1 ~]# journalctl | grep qdrouterd
> Mar 26 03:00:47 c02-h10-r620-vm1.rdu.openstack.engineering.redhat.com
> groupadd[19140]: group added to /etc/group: name=qdrouterd, GID=993
> Mar 26 03:00:47 c02-h10-r620-vm1.rdu.openstack.engineering.redhat.com
> groupadd[19140]: group added to /etc/gshadow: name=qdrouterd
> Mar 26 03:00:47 c02-h10-r620-vm1.rdu.openstack.engineering.redhat.com
> groupadd[19140]: new group: name=qdrouterd, GID=993
> Mar 26 03:00:47 c02-h10-r620-vm1.rdu.openstack.engineering.redhat.com
> useradd[19145]: new user: name=qdrouterd, UID=996, GID=993,
> home=/var/lib/qdrouterd, shell=/sbin/nologin
> Mar 28 10:39:06 c02-h10-r620-vm1.rdu.openstack.engineering.redhat.com
> qdrouterd[16084]: [0x7fe3f0016aa0]:pn_session: too many sessions: 32768
> channel_max is 32767
> Mar 28 10:39:06 c02-h10-r620-vm1.rdu.openstack.engineering.redhat.com kernel:
> qdrouterd[16087]: segfault at 88 ip 00007fe40b79d820 sp 00007fe3fd5f9298
> error 6 in libqpid-proton.so.10.0.0[7fe40b77f000+4b000]
> Mar 28 10:39:07 c02-h10-r620-vm1.rdu.openstack.engineering.redhat.com
> systemd[1]: qdrouterd.service: main process exited, code=killed,
> status=11/SEGV
> Mar 28 10:39:07 c02-h10-r620-vm1.rdu.openstack.engineering.redhat.com
> systemd[1]: Unit qdrouterd.service entered failed state.
> Mar 28 10:39:07 c02-h10-r620-vm1.rdu.openstack.engineering.redhat.com
> systemd[1]: qdrouterd.service failed.
> Mar 29 01:02:09 c02-h10-r620-vm1.rdu.openstack.engineering.redhat.com
> /usr/sbin/katello-service[1740]: *** status failed: qdrouterd ***
> Logs from Capsule 2:
> [root@c02-h10-r620-vm2 ~]# systemctl status qdrouterd
> ● qdrouterd.service - Qpid Dispatch router daemon
> Loaded: loaded (/usr/lib/systemd/system/qdrouterd.service; enabled; vendor
> preset: disabled)
> Drop-In: /etc/systemd/system/qdrouterd.service.d
> └─limits.conf
> Active: failed (Result: signal) since Wed 2018-03-28 10:58:02 EDT; 14h ago
> Process: 1158 ExecStart=/usr/sbin/qdrouterd -c
> /etc/qpid-dispatch/qdrouterd.conf (code=killed, signal=SEGV)
> Main PID: 1158 (code=killed, signal=SEGV)
> Mar 28 07:38:46 c02-h10-r620-vm2.rdu.openstack.engineering.redhat.com
> systemd[1]: Started Qpid Dispatch router daemon.
> Mar 28 07:38:46 c02-h10-r620-vm2.rdu.openstack.engineering.redhat.com
> systemd[1]: Starting Qpid Dispatch router daemon...
> Mar 28 10:58:02 c02-h10-r620-vm2.rdu.openstack.engineering.redhat.com
> qdrouterd[1158]: [0x7f36a000a170]:unable to find an open available channel
> within limit of 32767
> Mar 28 10:58:02 c02-h10-r620-vm2.rdu.openstack.engineering.redhat.com
> qdrouterd[1158]: [0x7f36a000a170]:process error -2
> Mar 28 10:58:02 c02-h10-r620-vm2.rdu.openstack.engineering.redhat.com
> qdrouterd[1158]: [0x7f36a000a170]:pn_session: too many sessions: 32768
> channel_max is 32767
> Mar 28 10:58:02 c02-h10-r620-vm2.rdu.openstack.engineering.redhat.com
> systemd[1]: qdrouterd.service: main process exited, code=killed,
> status=11/SEGV
> Mar 28 10:58:02 c02-h10-r620-vm2.rdu.openstack.engineering.redhat.com
> systemd[1]: Unit qdrouterd.service entered failed state.
> Mar 28 10:58:02 c02-h10-r620-vm2.rdu.openstack.engineering.redhat.com
> systemd[1]: qdrouterd.service failed.
> {code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
