dhcpd process hitting data size limit

[sthaug at nethelp.no]

> > Mar  3 14:23:26 dhcp2 dhcpd: failover: connect: no matching state.
> 
> are you sure the failover connection is entering 'normal' state on
> both sides after a restart?  if it is, it is logged to syslog.

Yes. For instance, when restarting the secondary yesterday I got the
following in the logs for the secondary:

Mar 25 22:46:54 slam2 dhcpd: failover peer dhcp1-dhcp2: I move from normal to startup
Mar 25 22:46:54 slam2 dhcpd: failover peer dhcp1-dhcp2: I move from startup to communications-interrupted
Mar 25 22:46:54 slam2 dhcpd: failover peer dhcp1-dhcp2: peer moves from normal to communications-interrupted
Mar 25 22:46:54 slam2 dhcpd: failover peer dhcp1-dhcp2: I move from communications-interrupted to normal

and in the logs for the primary:

Mar 25 22:46:54 dhcp2 dhcpd: failover peer dhcp1-dhcp2: peer moves from normal to communications-interrupted
Mar 25 22:46:54 dhcp2 dhcpd: failover peer dhcp1-dhcp2: I move from communications-interrupted to normal

I still have the logs from the March 3 event. On the primary:

Mar  3 14:23:21 dhcp2 dhcpd: dhcp_failover_put_message: something went wrong.
Mar  3 14:23:21 dhcp2 dhcpd: peer dhcp1-dhcp2: disconnected
Mar  3 14:23:21 dhcp2 dhcpd: failover peer dhcp1-dhcp2: I move from normal to communications-interrupted
...
<process restarted at 14:43:16>
Mar  3 14:43:16 dhcp2 dhcpd: failover peer dhcp1-dhcp2: I move from communications-interrupted to startup
Mar  3 14:43:16 dhcp2 dhcpd: failover peer dhcp1-dhcp2: peer moves from normal to communications-interrupted
Mar  3 14:43:16 dhcp2 dhcpd: failover peer dhcp1-dhcp2: I move from startup to normal


On the secondary (process not restarted):

Mar  3 14:23:21 slam2 dhcpd: peer dhcp1-dhcp2: disconnected
Mar  3 14:23:21 slam2 dhcpd: failover peer dhcp1-dhcp2: I move from normal to communications-interrupted
...
Mar  3 14:43:16 slam2 dhcpd: failover peer dhcp1-dhcp2: peer moves from normal to communications-interrupted
Mar  3 14:43:16 slam2 dhcpd: failover peer dhcp1-dhcp2: I move from communications-interrupted to normal

> anyway, if your failover stanzas have different names, that would
> explain what is different between your boxes and others (incl. our
> lab).

I'm afraid that's not the problem since the failover stanzas have the
same name.

I am by now reasonably certain that hitting the datasize limit (in my
case at 512 MB) is due to the memory leak that I have observed and
reported in the followup to this thread.

Steinar Haug, Nethelp consulting, sthaug at nethelp.no