[EnMasse] Disconnect from time to time
Ulf Lilleengen
lulf at redhat.com
Wed Mar 27 11:45:43 UTC 2019
On Wed, Mar 27, 2019 at 11:49 AM Gordon Sim <gsim at redhat.com> wrote:
> On 27/03/2019 10:42 am, Bob Claerhout wrote:
> > I do have 2 router pods. This is the logging from the second pod:
> > https://pastebin.com/tMzghGsS
>
> It *looks* like the broker connection was timed out for lack of heartbeat:
>
> [0x7f37c815e940]:0 <- @close(24) [error=@error(29)
> [condition=:"amqp:resource-limit-exceeded",
> description="local-idle-timeout expired"]]
>
>
>
>
At present, this is hardcoded to 5 seconds in the broker connector service.
I wonder if we should make this configurable (on both ends?), though I find
it unusual that it would not be able to receive heartbeat within 5 seconds.
Bob, how much memory is the broker running with?
One possibility is that the broker fails to send the heartbeat either due
to a bug in the broker amqp/connection code, or broker is running with low
resources causing it to be 'stuck' in GC/load on broker is above what
resource settings permit.
Best regards,
Ulf
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/enmasse/attachments/20190327/6d25f44f/attachment.htm>
More information about the enmasse
mailing list