lpd forking infinitely after HACMP failover.
Two RS/6000, when the second machine takes over and attempts to
establish the remote print queue it seems to go into an infinite
loop of forking and starting lpd daemon.
D: Customer is using a single name from machine to machine to indicate
a single physical printer. For the machine that the printer is
attached to, that name is a local queue. For all other machines that
name is a remote queue pointing to the machine with the local queue.
This is working.
When the HA failover occurs, this leaves the backup system with a
remote queue pointing to itself (but using the secondary ip address).
As soon as the customer does an lpstat, they end up with multiple
instances of the lpd.
A: I set up a test case on, pointing a remote queue back to
itself over the secondary address. An lpstat command returned
"host does not have line printer access". Once I put the host
system into its own /etc/hosts.lpd, I was able to recreate the
spawning of lpd daemons. These lpd processes continued to stay
suspended after the lpstat had been stopped. They did eventually
go away once the lpd subsystem was brought down.
ACT: The customer did have his own host id in his own /etc/hosts.lpd.
From my own recreation, if this is taken out (take away the privelege
to remote print to itself) then the infinite loop will be cut short by
a "host does not have line printer access" message. This would
include any entries in the /etc/hosts.equiv for the host itself.
Support Line: lpd forking infinitely after HACMP failover. ITEM: T9755L
Dated: April 1995 Category: N/A
This HTML file was generated 99/06/24~13:30:35
Comments or suggestions?