[rabbitmq-discuss] Strange behaviour of RabbitMQ or client when network problems occurs

[Michael Klishin]

Haster:

> I don't have a possibility to get trace, but I think that new connection is
> started before the server
> delete old connection. 

It takes time to detect dead connections and TCP will try redelivery first. This is why
RabbitMQ has heartbeats, to detect that the peer has disconnect earlier.

> And how I can avoid this situation? I don't know when server find connection
> error and close old connection. 

If you use REJECT for TCP traffic on a firewall, it does not immediately indicate that the connection
is closed to RabbitMQ. Either find a different way of testing what you are trying to test,
or set heartbeat interval to a low value (e.g. a few seconds).

MK



-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 495 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20130913/9ad4d295/attachment.pgp>