Hi Pablo,
(Pablo Escobar, Wed, Aug 16, 2017 at 01:02:08PM +0200:)
> we got few errors like this and the cluster couldn't boot properly:
>
> fatal: [compute017]: FAILED! => {"failed": true, "msg": "the field 'port'
> has an invalid value (620:5ca1:2f0:f816:3eff:fedb:34ae), and could not be
> converted to an int. Error was: invalid literal for int() with base 10:
> '620:5ca1:2f0:f816:3eff:fedb:34ae'"}
>
> We were booting a cluster with 30 nodes. 25 nodes booted correctly but the
> other 5 nodes got this error.
>From the looks of it, this error comes from Ansible (can you pls post
the task name that produced it?). It looks like there's some Jinja
expression `{{ something.split(':')|int }}` where `something` is clearly
expected to be an IPv4:port pair. Instead you get an IPv6 address,
which breaks the `.split(':')` part: what should be a simple decimal
integer, is now the last 112 bits of the IPv6 address.
Since this only happens on a few nodes, my guess would be that there is
some slow response from DHCPv4 (or whatever mechanism SWITCHengines uses
to assign IPv4 addresses to VMs) so that when Ansible connects and
gathers the node information, the IPv4 address has not yet been assigned
and only the IPv6 is available. Does this make sense?
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.
