Hi folks,
we just ran into a strange problem in our test environment (Spectrum 10.2.1 / CentOS 6.9):
When attempting to start AlarmNotifier, it failed saying that it could not connect to port 0xbeef, although the SpectroSERVER process was running:
[spectrum@spe2-vm Notifier]$ ./AlarmNotifier
AlarmNotifier 10.2.1.000
Multiple Landscapes Mode
SPECTRUM Alarm Notification Manager Enabled
CA Technologies
One CA Plaza
Islandia, NY 11749 USA
Copyright 2017 CA Technologies All Rights Reserved
AlarmNotifier: Warning: 10:01:09 05/24/17 - Landscape on spe2-vm: failed to connect, not listening on port 0xbeef. Check that the server is started and ready.
AlarmNotifier: 10:01:29 05/24/17 - Error: Landscape 0x1000000: Connection lost
After some investigation, it turned out that the AlarmNotifier tried to connect to the port via IPv6:
[spectrum@spe2-vm Notifier]$ telnet spe2-vm 48879
Trying 2001:db8::98:134...
telnet: connect to address 2001:db8::98:134: Connection refused
Trying 192.168.98.134...
Connected to spe2-vm.
I have the following entries in /etc/hosts:
[spectrum@spe2-vm Notifier]$ grep spe2-vm /etc/hosts
2001:db8::98:134 spe2-vm
192.168.98.134 spe2-vm
When I remove the IPv6 entry or change the associated hostname, AlarmNotifier works. Has anybody been facing the same issue and could convince the AlarmNotifier to fall back to IPv4?
Ideally, the SpectroSERVER process would just be listening on IPv6 as well...
[spectrum@spe2-vm Notifier]$ netstat -an | grep 48879
tcp 0 0 0.0.0.0:48879 0.0.0.0:* LISTEN
As far as I know, this was not possible because of some Visibroker libraries, but this iformation is also a couple years old.
Regards
Maik