Hi all,
We've been developing a CNI plugin for Mesos. During testing the plugin
seems to perform fine, and I actually see Mesos calling it and getting an
IP from it. However, just afterwards the isolate() returns the Executor,
the executor reports to have exited, hence destroying the container.
I0702 02:38:45.207439 209 cni.cpp:1069] Got assigned IPv4 address '
10.1.1.100/24' from CNI network 'CNI-test-onboard' for container
e67c7678-f6b4-4de7-85c1-2b40a80fc80d
I0702 02:38:45.208228 208 cni.cpp:805] Unable to find DNS nameservers for
container e67c7678-f6b4-4de7-85c1-2b40a80fc80d. Using host
'/etc/resolv.conf'
// +5sec
I0702 02:38:50.459604 206 containerizer.cpp:1821] Executor for container
'e67c7678-f6b4-4de7-85c1-2b40a80fc80d' has exited
I0702 02:38:50.459709 206 containerizer.cpp:1585] Destroying container
'e67c7678-f6b4-4de7-85c1-2b40a80fc80d'
Q: Is this expected? Currently our implementation does not provide DNS, can
this be the cause? Consider that the slave has no visibility/access to the
network where the container/s are being onboarded on (aiming to be a
network overlay between containers only).
We are running Mesos (latest master), we create the container through
Marathon (latest master) with:
{
"id": "basic-0",
"cmd": "while [ true ] ; do echo 'Hello Marathon' ; sleep 5 ; done",
"cpus": 0.1,
"mem": 10.0,
"instances": 1,
"ipAddress": {
"networkName": "CNI-test-onboard"
}
}
