[LON-CAPA-dev] problems starting after update to 1.0.99.1

[Gabriel Friedman]

Greetings.  After running the update script from v1.0.2 to update 
to 1.0.99.1, i can no longer successfully start lon-capa

The startup process hangs at this point: 
----------------------------
[root@lon-capa loncapa-1.0.99.1]# /etc/init.d/loncontrol stop 
Stopping LON-CAPA
lonsql     stopped
lond       stopped
lonc       stopped
lonhttpd   not running
[root@lon-capa loncapa-1.0.99.1]# /etc/init.d/loncontrol start
Starting LON-CAPA
Starting LON-CAPA client and daemon processes (please be patient)
lonsql     started
lond       started
lonc       new started
lonhttpd   started
checking logs
testing connections
. BCHSLib down
...
---------------------------


lonc_error reports:

Event: trapped error in ?: send: Cannot determine peer address at 
/home/httpd/lib/perl//LONCAPA/LondConnection.pm line 565
Attempt to free unreferenced scalar at 
/usr/lib/perl5/site_perl/5.6.1/i386-linux/Event.pm line 148.
Attempt to free unreferenced scalar at 
/usr/lib/perl5/site_perl/5.6.1/i386-linux/Event.pm line 148.



If i break out of the startup process (^C), and attempt to start 
again, it seems to get one step further:
---------------
[root@lon-capa loncapa-1.0.99.1]# /etc/init.d/loncontrol start
Starting LON-CAPA
Starting LON-CAPA client and daemon processes (please be patient)
lonsql     running
lond       running
lonc       running
lonhttpd   started
checking logs
testing connections
. BCHSLib down
... capa1 down
..
--------------

lonc_error will report the same three lines as it did before.

Perhaps lonc is not handling the 'unable to connect' case 
correctly?


Some background on my setup:

A couple days ago , lonc was just beginning to show connections to 
s17.lite.msu.edu (the four domains hosted there?)   all other 
connections were still reporting conn_lost status in the reports.  
I woudl guess that this is due to the cmich domain not being in 
other servers domain tabs?   I am a bit shakey on this point.

I thought that you might perhaps wish to know about this error 
case?

-- Gabe Friedmann