[rabbitmq-discuss] LDAP backend problem

[Jensen, Kristian]

You are right, it will try the next one. But the timeout is SO long, that it makes rabbitmq useless, until all LDAP servers is back up.

So it would be nice, if we can change the default "infinity" to something like 10ms.

Med venlig hilsen / Best regards

Kristian Jensen
System Engineer | Site Operations Denmark | eBay Classifieds Group 
Phone: +45 40226333 | krjensen at ebay.com
 


-----Original Message-----
From: rabbitmq-discuss [mailto:rabbitmq-discuss-bounces at lists.rabbitmq.com] On Behalf Of Simon MacMullen
Sent: 19. februar 2014 11:50
To: Discussions about RabbitMQ
Cc: Kristian Jensen
Subject: Re: [rabbitmq-discuss] LDAP backend problem

On 19/02/14 08:55, Kristian Jensen wrote:
> Why is the LDAP module not robust againt unavailable LDAP backens? - 
> could you intruduce a sort of healtcheck?

We just pass the list of servers down to the underlying Erlang LDAP implementation (eldap). As far as I can see that then attempts to establish a TCP connection to each server in turn; on failure it tries the next one.

I assume that attempting to connect to one of your unavailable LDAP servers is taking a long time (i.e. connections are timing out rather than coming back with "connection refused" or "no route to host" or whatever).

One thing I notice is that eldap supports the idea of a connection timeout (defaulting to infinity); we don't support changing that from the default. I'll file a bug to fix that.

But if LDAP connections are timing out then we are going to struggle to handle that without incurring similar delays in AMQP / HTTP / STOMP / whatever.

Cheers, Simon

--
Simon MacMullen
RabbitMQ, Pivotal
_______________________________________________
rabbitmq-discuss mailing list
rabbitmq-discuss at lists.rabbitmq.com
https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss