Cfservd keeps exiting.

[Sean Everson]

Hi all,

I have been using Cfengine 2.0.3 on Redhat 7.3 for a few weeks now.  I have 
been experiencing sporadic failures where cfservd has been exiting without 
leaving any errors in my logfiles.

I am running cfservd in daemon mode out of an init script.  My config files 
all seem to be working fine. I have poured over the list and the docs but 
have not seen this particular problem.  As I am new to cfengine I am not sure 
if this is a result of a bad config file or a feature :)

The main aim of my current config is to copy a number of system files from the 
masterhost to my clients. This is done with a number of copy commands.  This 
works well most of the time.  However, once in awhile the cfservd on the 
masterhost simply exits silently.  

On the cfservd side I get the following strace when the daemon exits:

select(4, [3], NULL, NULL, NULL)        = 1 (in [3])
accept(3, {sin_family=AF_INET, sin_port=htons(2002), 
sin_addr=inet_addr("192.1.1.23")}}, [16]) = 4
time(NULL)                              = 1029774058
rt_sigprocmask(SIG_SETMASK, NULL, [RTMIN], 8) = 0
write(6, "\240\200\31@\0\0\0\0\200\250\f\10\4\303\4\10\200\337\16"..., 148) = 
148
rt_sigprocmask(SIG_SETMASK, NULL, [RTMIN], 8) = 0
rt_sigsuspend([] <unfinished ...>
--- SIGRTMIN (Real-time signal 0) ---
<... rt_sigsuspend resumed> )           = -1 EINTR (Interrupted system call)
sigreturn()                             = ? (mask now [RTMIN])
stat64("/var/cfengine/inputs/cfservd.conf", {st_mode=S_IFREG|0600, 
st_size=1108, ...}) = 0
select(4, [3], NULL, NULL, NULL)        = ? ERESTARTNOHAND (To be restarted)
--- SIGRT_1 (Real-time signal 1) ---
wait4(21754, NULL, __WCLONE, NULL)      = 21754
_exit(1)                                = ?


The important bit seem to be this line:

select(4, [3], NULL, NULL, NULL)        = ? ERESTARTNOHAND (To be restarted)

The client will see something like this when the server exits:

Connect to rdu.mydomain.com = 192.1.1.70 on port cfengine
Loaded /var/cfengine/ppkeys/root-192.1.1.70.pub
jfk.mydomain.com: Couldn't recv
jfk.mydomain.com: recv
jfk.mydomain.com: Challenge response from server rdu.mydomain.com/192.1.1.70 
was incorrect!

This is not a simple misconfig since if I restart cfservd on the masterhost 
and re-run cfagent on the client it will complete successfully.

Any help on this would be greatly appreciated!

-Sean Everson
severson@combinenet.com