[systemd-devel] networkd failing sporadically failing to bring up network device

[Tom Gundersen]

On Tue, Apr 7, 2015 at 5:46 PM, Richard Maw <richard.maw at codethink.co.uk> wrote:
> On Sat, Apr 04, 2015 at 12:08:04PM +0100, Sitsofe Wheeler wrote:
>> Hi,
>>
>> I've noticed that networkd occasionally fails to bring up one of two
>> network interfaces on boot (this happens about once every 70 or so
>> boots). The machine in question is a VMware ESXi 5.5 guest with two
>> VMXNET3 network adaptors. When this issue occurs the message
>>
>> rtnl: received address for a nonexistent link (1), ignoring
>>
>> is present in the journal.
>>
>> Does anyone know if this is a known issue in systemd 216 (I'm using
>> Fedora 21)? I've attached a journal log below:
>
> There's been a few post-217 changes related to parsing messages in rtnl,
> one of which was related to accidentally interpreting change notifications
> as responses under rare circumstances, so there are some known bugs in
> 216, but I haven't encountered yours before.
>
> Re-testing with the latest systemd snapshot would let us know if it's
> one of the bugs fixed post 217.
>
> If it comes to debugging what netlink is doing, one option is to
> use the nlmon interface, capture it with tcpdump as described in
> http://seclists.org/tcpdump/2014/q4/32 and then use wireshark to parse
> the dump.

Seems I missed parts of this thread (I cannot find the original email
in my archive), so in case I missed something important: could this be
reproduced with current git, or is this seemingly resolved now?

Cheers,

Tom