1

I need to setup a VIP / floating IP that network devices will send their syslog messages to. The idea behind the VIP is to have a highly available setup in an attempt to minimize the amount of logs lost as a result of the syslog server being down.

This is going to be done in a CentOS 7 environment so I figure I can either have rsyslog listen for remote messages on the floating address or setup syslog-ng on an alternate port (such as 5514) solely for receiving the network messages.

My research tells me that Pacemaker should be able to do this but I am not quite sure how. Any tips on the actual how part would be greatly appreciated.

0

We have this configuration along with syslog-ng, which I believe is needed as syslog itself doesn't allow receiving remote messages.

There is an ocf primitive called IPaddr2 whose syntax is similar to this:

primitive virtual_ip ocf:heartbeat:IPaddr2 \
        params ip="192.168.1.100" cidr_netmask="32" nic="eth0" \
        op monitor interval="30s" timeout="20s" \
        meta target-role="Started"

It is self explanatory, however, to see all the parameters you can check the help within the crm command.

  • So, do you just tell syslog-ng to listen for messages on 192.168.1.100 on both nodes? Is there any health checking of syslog-ng being done or is it based on the node being available? I am new to clustering like this so please bear with me. Thanks for the help! – GeneBean Aug 30 '15 at 19:12
  • Yes, exactly, but not just listen, but also configure it as a cluster resource. This way you make sure that it only runs where the VIP is currently configured. You can configure syslog-ng as a LSB resource but be careful, the LSB script that installs the package on Ubuntu is not Pacemaker LSB compliant, I don't know in CentOS. In any case you can customize the LSB script to make it LSB compliant. – nKn Aug 30 '15 at 19:17
  • That makes sense. One last question: would I be right in assuming that I would need to make a group out of the IPaddr2 and syslog-ng resources? It seems both are available via the heartbeat provider ( github.com/ClusterLabs/resource-agents/blob/master/heartbeat/…) – GeneBean Aug 30 '15 at 20:27
  • Yep, that would be a good idea since both will be running on the same machine and one without the other doesn't make sense, so a group is convenient here – nKn Aug 30 '15 at 20:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.