[
https://issues.apache.org/jira/browse/KAFKA-1512?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14051951#comment-14051951
]
Jay Kreps commented on KAFKA-1512:
----------------------------------
Yes, I hadn't thought of that. Disabling connections could potentially be
useful. The intended use was actually the other way around, basically default
most things to something reasonable like 10 but have a way to whitelist some
IPs to have unlimited connections.
The background here is that we were previously having clients bootstrap
metadata through a VIP (which appears to the kafka nodes as a single ip). We
just had an issue where a 200 node cluster that uses Kafka started creating and
leaking connections through the vip which brought down a big shared cluster. So
we thought we should have some limits. The hope was to change the VIP to DNS
round-robin and gradually migrate the clients to that. In the meantime we
thought it would be useful to be able to enforce the limit but whitelist the
VIP with unlimited connections.
Thinking about this, maybe it is a little crazy hard coding ip/host names in
config?
> Limit the maximum number of connections per ip address
> ------------------------------------------------------
>
> Key: KAFKA-1512
> URL: https://issues.apache.org/jira/browse/KAFKA-1512
> Project: Kafka
> Issue Type: New Feature
> Reporter: Jay Kreps
> Assignee: Jay Kreps
> Attachments: KAFKA-1512.patch, KAFKA-1512.patch
>
>
> To protect against client connection leaks add a new configuration
> max.connections.per.ip
> that causes the SocketServer to enforce a limit on the maximum number of
> connections from each InetAddress instance. For backwards compatibility this
> will default to 2 billion.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
