I've got a couple of virtual machines (running Scientific Linux 6.2) that publish Workstation records so that they can easily be found on the net. Some of the virtual machines stop publishing after a few minutes, i.e. I can reach them via hostname.local for a few minutes and then any discovery will time out. They are still reachable via their IP address. Restarting avahi-daemon
results in the same behaviour.
There is nothing in /var/log/messages that would indicate a problem:
Nov 27 08:07:33 vm01 avahi-daemon[2836]: Got SIGTERM, quitting.
Nov 27 08:07:33 vm01 avahi-daemon[2836]: Leaving mDNS multicast group on interface eth0.IPv4 with address 10.100.200.226.
Nov 27 08:07:33 vm01 avahi-daemon[2880]: Found user 'avahi' (UID 70) and group 'avahi' (GID 70).
Nov 27 08:07:33 vm01 avahi-daemon[2880]: Successfully dropped root privileges.
Nov 27 08:07:33 vm01 avahi-daemon[2880]: avahi-daemon 0.6.25 starting up.
Nov 27 08:07:33 vm01 avahi-daemon[2880]: WARNING: No NSS support for mDNS detected, consider installing nss-mdns!
Nov 27 08:07:33 vm01 avahi-daemon[2880]: Successfully called chroot().
Nov 27 08:07:33 vm01 avahi-daemon[2880]: Successfully dropped remaining capabilities.
Nov 27 08:07:33 vm01 avahi-daemon[2880]: No service file found in /etc/avahi/services.
Nov 27 08:07:33 vm01 avahi-daemon[2880]: Joining mDNS multicast group on interface eth0.IPv4 with address 10.100.200.226.
Nov 27 08:07:33 vm01 avahi-daemon[2880]: New relevant interface eth0.IPv4 for mDNS.
Nov 27 08:07:33 vm01 avahi-daemon[2880]: Network interface enumeration completed.
Nov 27 08:07:33 vm01 avahi-daemon[2880]: Registering new address record for fe80::216:3eff:fe73:df12 on eth0.*.
Nov 27 08:07:33 vm01 avahi-daemon[2880]: Registering new address record for 10.100.200.226 on eth0.IPv4.
Nov 27 08:07:33 vm01 avahi-daemon[2880]: Registering HINFO record with values 'X86_64'/'LINUX'.
Nov 27 08:07:34 vm01 avahi-daemon[2880]: Server startup complete. Host name is vm01.local. Local service cookie is 320168790.
There is nothing further in the log that would indicate a problem. The SIGTERM during startup seems to be normal, a cursory google'ing showed that.
I only modified entries in the publish
section of the configuration; everything else was left as-is:
[publish]
disable-publishing=no
disable-user-service-publishing=no
#add-service-cookie=no
publish-addresses=yes
#publish-hinfo=yes
publish-workstation=yes
#publish-domain=yes
#publish-dns-servers=192.168.50.1, 192.168.50.2
#publish-resolv-conf-dns-servers=yes
#publish-aaaa-on-ipv4=yes
#publish-a-on-ipv6=no
The avahi-daemon does not produce any significant cpu load. The process is running:
# ps auxww |grep avahi
avahi 2880 0.0 0.0 27660 1488 ? S 08:07 0:00 avahi-daemon: running [vm01.local]
avahi 2881 0.0 0.0 27660 520 ? Ss 08:07 0:00 avahi-daemon: chroot helper
I have some virtual machines with CentOS 5.8 and they do not have this problem. They have avahi-daemon 0.6.16 and the same configuration. Any ideas?
Best Answer
Just realized I'm about to respond to a really old question, but maybe someone will find this useful.
Have a look at your network equipment, specifically any managed switches. I've run into issues with IGMP snooping, where multicast requests might suddenly stop being received by clients after a few minutes because the switch was filtering it out.
Food for thought.