endless loop while reconnecting IPMI connection

[GoogleCodeExporter]

What steps will reproduce the problem?

Compile conman with freeipmi support, establish a connection with a console configured to use freeimpi. Make sure that the console connects. Now disconnect the console by disabling the switch port, pulling the cable, etc. Once conman has notified you that the the console was disconnected and that it was unable to reconnect (timeout), re-enable the connection.

What is the expected output? What do you see instead?

The expected output would be that conman manages to reconnect to the disconnected console once the connection is available again.

What is witnessed instead is an endless loop of connections/disconnections:

2010-06-02 17:26:36 NOTICE:    Starting ConMan daemon 0.2.5 (pid 12345)
2010-06-02 17:26:36 INFO:      IPMI SOL engine started with 1 thread for 1 console
2010-06-02 17:26:36 INFO:      Open file limit set to 1024
2010-06-02 17:26:36 INFO:      Console [host1] connected to <bmc1>
2010-06-02 17:27:36 NOTICE:    Console [host1] disconnected from <bmc1>
2010-06-02 17:28:17 INFO:      Unable to connect to <bmc1> via IPMI to console [host1]: connection timeout
2010-06-02 17:29:17 INFO:      Console [host1] connected to <bmc1>
2010-06-02 17:29:17 NOTICE:    Console [host1] disconnected from <bmc1>
2010-06-02 17:29:17 INFO:      Console [host1] connected to <bmc1>
2010-06-02 17:29:18 NOTICE:    Console [host1] disconnected from <bmc1>
2010-06-02 17:29:18 INFO:      Console [host1] connected to <bmc1>
2010-06-02 17:29:19 NOTICE:    Console [host1] disconnected from <bmc1>
2010-06-02 17:29:19 INFO:      Console [host1] connected to <bmc1>
2010-06-02 17:29:20 NOTICE:    Console [host1] disconnected from <bmc1>
2010-06-02 17:29:20 INFO:      Console [host1] connected to <bmc1>
2010-06-02 17:29:21 NOTICE:    Console [host1] disconnected from <bmc1>

What version of the software are you using? On what operating system?

Last released conman-0.2.5
Last released FreeIPMI 0.8.6
Fedora core 12.

Please provide any additional information below.

This problem has been tracked down to the gotEOF flag of the ipmi object which is set to 1 the first time the connection goes down but is never reset to 0 in the server-ipmi.c file, as it is the case, however, in the other back-ends. The attached patch solves this problem.

gotEOF.ipmi.reinit.patch.txt

Original issue reported on code.google.com by matthieu.perotin@bull.net on 2 Jun 2010 at 3:43