Failed to start node (12) (and some kind of workaround)

[grindelwaldus]

OK, so I made some research and there's several topics with this problem. So here's how I resolved this (this is not a panacea of course and I'm posting this mostly like some kind of a bugreport and as a possible (in some cases) workaround ).

First of all, I have EVE 2.0.3-86 on ESXI 6.0.

So I have serveral dynamips nodes (c7200-adventerprisek9-mz.152-4.S6) in my lab which worked like a charm, but one morning some of them failed to start with

Code: Select all

Failed to start node (12)

This is what I saw in logs:

Code: Select all

Jul 30 09:40:17 ovs-vsctl: no port named vunl1_6_0
Jul 30 09:40:17 ERROR: Failed to delete the TAP interface (80034).
Jul 30 09:35:34 Jul 30 09:35:34 ERROR: Failed to start node (12).
Jul 30 09:35:34 Jul 30 09:35:34 ERROR: Failed to delete the TAP interface (80034).
TUNSETIFF: Device or resource busy
Jul 30 09:35:34 ovs-vsctl: no port named vunl1_6_0

At first I tried to wipe the node but it didn't really help. (Permissions were, of course, fixed as well).

It seems that when we start node unl_wrapper tries to delete interfaces listed above, but it's not able to delete it: this is what happens if we try to perform it from cli

Code: Select all

tunctl -d vunl1_6_0 
TUNSETIFF: Device or resource busy

But it's possible to delete it this way:

Code: Select all

sudo ip link delete vunl1_6_0

After you remove all the problem interfaces listed in logs, your node will be able to start succesfully.

FYI:

sysinfo.zip

[Uldis (UD)]

Upgrade to v92

Uldis