Hi Riccardo, So when it worked before it was by chance that all the ipv4 addresses were selected? That sounds too much luck ;)
I will test the fix but before this fix it was working correctly most of the times so my testing won't be 100% reliable. I have created this issue in github to continue the discussion https://github.com/gc3-uzh-ch/elasticluster/issues/478 Pablo. 2017-08-16 17:34 GMT+02:00 Riccardo Murri <riccardo.mu...@gmail.com>: > Hi Pablo (or anyone with IPv6 connectivity), > > can you please check if branch "fix/inventory-with-ipv6" on > https://github.com/riccardomurri/elasticluster fixes the issue? > > The reason why only *some* nodes were affected is likely this: once the > VMs have been started, ElastiCluster probes SSH connection to the nodes, > trying all the returned IP addresses in a round-robin fashion. Now, if > *both* the IPv4 and the IPv6 address are responding, then which one is > selected as the actual connection address is just a matter of chance. > (Still, one would expect a 50/50 split, whereas your inventory shows > more of a 80/20 but I do not have an explanation for that.) > > Ciao, > R > > -- > Riccardo Murri / Email: riccardo.mu...@gmail.com / Tel.: +41 77 458 98 32 > -- You received this message because you are subscribed to the Google Groups "elasticluster" group. To unsubscribe from this group and stop receiving emails from it, send an email to elasticluster+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.