[Buildbot-devel] buildbot 0.7.7 slave on RHEL5 not sending keepalive events

[Jannich Thorsen]

Hello,

We have a buildbot 0.7.7 setup here with master on Ubuntu7.10 and slaves 
on Ubuntu7.10, Ubuntu8.04, RHEL4 and all works fine.

But when adding a RHEL5 slave to the setup, it keeps 
disconnecting/connecting about once every hour. Looking at twistd.log, 
as shown below, one can see that the keepalive events are not happening 
every 600s as supposed.

2008/08/11 13:59 CEST [Broker,client] sending application-level 
keepalives every 600 seconds
2008/08/11 14:42 CEST [-] sending app-level keepalive
2008/08/11 14:42 CEST [-] BotFactory.checkActivity: nothing from master 
for 2559 secs
2008/08/11 14:42 CEST [Broker,client] BotFactory.keepaliveLost
2008/08/11 14:42 CEST [Broker,client] lost remote
2008/08/11 14:42 CEST [Broker,client] <twisted.internet.tcp.Connector 
instance at 0xb7b0514c> will retry in 2 seconds
2008/08/11 14:42 CEST [Broker,client] Stopping factory 
<buildbot.slave.bot.BotFactory instance at 0xb7b0506c>
2008/08/11 14:42 CEST [-] Starting factory 
<buildbot.slave.bot.BotFactory instance at 0xb7b0506c>
2008/08/11 14:42 CEST [Broker,client] message from master: attached
2008/08/11 14:42 CEST [Broker,client] 
SlaveBuilder.remote_print(rhel5-i386-prodtest): message from master: 
attached
2008/08/11 14:42 CEST [Broker,client] sending application-level 
keepalives every 600 seconds

On the other slaves in our setup, the twistd.log shows each slave 
sending it every 600s:

2008/08/11 13:52 +0200 [-] sending app-level keepalive
2008/08/11 14:02 +0200 [-] sending app-level keepalive
2008/08/11 14:12 +0200 [-] sending app-level keepalive

It's an out of the box RHEL5 install, with Python 2.4.3.

What is going on? Any help appreciated.

Regards
Jannich Thorsen
Teklatech A/S