Periodic log message in case of interface in "down" state

Lars Kruse lists at sumpfralle.de
Sat May 4 22:24:30 CEST 2019


Hello,

I am running a VPN with around 30 nodes. One of these (a server for monitoring)
is only connected to the VPN - but without any configured interface state
("tinc-up" is empty).

# ip address show dev foo
7: foo: <POINTOPOINT,MULTICAST,NOARP> mtu 1500 qdisc pfifo_fast state DOWN group default qlen 500
    link/none

In this situation, I noticed that periodically the following message appears in
the log:

 Can't write to Linux tun/tap device (tun mode) /dev/net/tun: Input/output error

These single messages appear with varying periods (between 60 seconds and five
minutes).

These messages stop appearing after the interface is brought up (without
configured addresses):

 ip link set foo up

The log messages will start again after setting it to "down".

The log messages are usually no real problem.
But recently I noticed, that sometimes a flood of these messages appears.
This can be problematic, since it can easily fill some hundred megabytes of
disk storage in less than an hour.

For now I will simply bring the interface "up" in order to avoid these messages.
But it would certainly be nice, if tinc would stop emitting these messages even
with the interface being in "down" state.
Or maybe there is something else wrong with my setup?

Cheers,
Lars


More information about the tinc mailing list