Hide Forgot
Description of problem: When using the new nodelist, it seems unnecessary to have an interface/bindnetaddr setting. One of the principles behind the nodelist idea was that when started, corosync would go through the node entries, figure out which entry was its own, and take it's ip addr from that. This should then tell you what the bindnetaddr would be. If you have a config file with both an interface/bindnetaddr and a nodelist, then the startup code could use the interface/bindnetaddr as an easy method of figuring out its own node entry by matching them. Without interface/bindnetaddr, identifying your own nodelist entry is a bit more work, but shouldn't be too difficult (cman has had code to do this reliably for a long time, so you should be able to get some ideas there.) Also, if bindnetaddr becomes optional, then the entire interface section could also be optional. Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
Dave, patches sent to ML. [PATCH 1/6] Add generic implementation of getifaddrs upto [PATCH 6/6] Allow autoconfiguration of interface section will allow following config file to work as expected (note cluster_name and NO interface section): ### SNIP #### compatibility: whitetank totem { version: 2 secauth: off threads: 0 cluster_name: testcluster } logging { fileline: off to_stderr: yes to_logfile: no logfile: /var/log/cluster/corosync.log to_syslog: yes debug: on timestamp: on } nodelist { node { ring0_addr: 10.34.38.107 nodeid: 1 } node { ring0_addr: 10.34.38.108 nodeid: 2 } node { ring0_addr: 10.34.38.109 } } ### - SNIP #### Hopefully this satisfy your needs.
Yes, it looks great, thanks.