are all agents on one server? Could be hit the fs.file-max in sysctl.
Please check:

1. In /etc/security/limits.conf
* soft nofile 1024
* hard nofile 65535
2. Increase ulimit by "ulimit -n 65535"
3. echo 65535 > /proc/sys/fs/file-max
4. In /etc/sysctl.conf
fs.file-max=65535


best,
 Alex 

2011/11/17 Mingjie Lai <mjlai09@gmail.com>
Hi.

> (1) How many agents are assured ?

Tell the truth, I'm not sure how many people ever tried so many logical
nodes for one physical node. :(

> 2011-11-16 11:38:00.687 WARN com.cloudera.flume.agent.LivenessManager:
> Heartbeats are backing up, currently behind by 189 heartbeats

Did you see it with one logical node? It could be caused by other issue,
such as https://issues.apache.org/jira/browse/Flume-808


On 11/16/2011 09:45 AM, Y. Sakamoto wrote:
> Hi,
> I tried to configure 53 agents (logical agents, on 1 physical server) by
> "flume shell" all at one.
> After this, heartbeats failed and following message was written in the
> collector log :
>
> 2011-11-16 11:38:00.687 WARN com.cloudera.flume.agent.LivenessManager:
> Heartbeats are backing up, currently behind by 189 heartbeats
>
>
> I wonder that the interval of heartbeats is too short to process many
> agents.
> So I have questions,
>
> (1) How many agents are assured ?
>
> (2) If my assumption is correct, which approach is better ?
>       a. Wait until one agent configuration is finished,
>           then next configuration continue.
>       b. Change the interval of heartbeats longer (about 1 minutes) .
>
>
> Thanks,
>    Y. Sakamoto.
>
>



--
Alexander Lorenz
http://mapredit.blogspot.com

Think of the environment: please don't print this email unless you really need to.