Kontena FAQ

how is the deploy calculated

from the last deploy, it triggers a regular deploy so for example min_health is respected

what's the encryption scheme used by Kontena overlay network? (weave)

IPSec/ESP using AES in GCM mode more details here: https://github.com/weaveworks/weave/blob/master/docs/fastdp-crypto.md

Is the traffic encrypted between nodes in the same region that communicate with their private IP ?

by default yes. you can disable encryption for a certain private subnet using kontena grid trusted-subnet add 1.2.3.4/24

weave

2018-01-25T07:53:48.665796695Z Unable to find image 'weaveworks/weavedb:latest' locally
2018-01-25T07:53:48.679919775Z latest: Pulling from weaveworks/weavedb
2018-01-25T07:53:48.679950835Z 14fea4d3be35: Pulling fs layer
2018-01-25T07:53:48.679971883Z 14fea4d3be35: Verifying Checksum
2018-01-25T07:53:48.679997638Z 14fea4d3be35: Download complete
2018-01-25T07:53:48.680012354Z 14fea4d3be35: Pull complete
2018-01-25T07:53:48.680027006Z Digest: sha256:ee08afa34af8773ed4afe122822dad65e713833e59dc7a8e0864fc8bc8fb4a18
2018-01-25T07:53:48.680041966Z Status: Downloaded newer image for weaveworks/weavedb:latest
2018-01-25T07:53:48.680056767Z The weave container has died. Consult the container logs for further details.
2018-01-25T07:53:48.680071696Z
2018-01-25T07:53:58.838112088Z W, [2018-01-25T07:53:58.837783 #1]  WARN -- Kontena::NetworkAdapters::Weave: timeout after waiting 10.1s of 10.0s until: weave started
2018-01-25T07:54:00.855737259Z E, [2018-01-25T07:54:00.854903 #1] ERROR -- Kontena::NetworkAdapters::WeaveExecutor: weaveexec exit 1: ["--local", "reset"]
2018-01-25T07:54:00.855864791Z ERROR: weave is not running; unable to remove from cluster.
2018-01-25T07:54:00.870385826Z Re-launch weave before reset or use --force to override.
2018-01-25T07:54:00.870412063Z
2018-01-25T07:54:08.435244982Z E, [2018-01-25T07:54:08.434921 #1] ERROR -- Kontena::NetworkAdapters::WeaveExecutor: weaveexec exit 1: ["--local", "launch-router", "--ipalloc-range", "", "--dns-domain", "kontena.local", "--password", "<redacted>", "--conn-limit", "0", "--trusted-subnets", ""]
2018-01-25T07:54:08.435412281Z The weave container has died. Consult the container logs for further details.

Check that weave can bind ports like 53.

etcd

can kontena run a cluster without etcd?

What is the consequence of the majority of the initial nodes going down ?

Internal networking or kontena/lb won't work and new deployments will fail.

What does IpamError mean?

Kontena::NetworkAdapters::IpamCleaner: Kontena::NetworkAdapters::IpamError: 503 "Service Unavailable"

Unable to connect to the etcd.

kontena/lb

Can I see the haproxy config?

kontena service exec lb/ingress-lb cat /etc/haproxy/haproxy.cfg

Can I run have network_mode: host and still link to lb?

no, only containers with a io.kontena.container.overlay_cidr get registered, and network_mode: host containers don't get that label kontena/kontena#2741

Why isn't my link working with a custom image named my-kontena-lb

The magic relies on the name kontena/lb, if you want to wrap the Kontena LB or bring your own loadbalancer you should put special env variable KONTENA_SERVICE_ROLE=lb for the service.

Agent

How to restart all agents?

for i in `seq 1 3`; do kontena container exec -it node-$i.grid.net/kontena-agent kill 1; done

Agent logs warn about LogWorker queue size

WARN -- Kontena::Workers::LogWorker: queue size is 1600

Agent can not flush logs fast enough to master, queue size is capped to 1600

waited Ns of 30.0s until...

Kontena::RpcClient: waited 3.2s of 30.0s until: request /containers/save has response wth id=172071412 yielded Array

Master is slowing down

Can I disable logs completely from choking my master?

Yes, just put:

logging:
  driver: none

If errors in agent logs: <@jakolehm> ok, it seems that none is only available in more recent docker versions

CLI

How can I delete multiple nodes / stacks / services ?

With xargs, see kontena/kontena#2410

Accidentally removed ~/.kontena_client.json

The file is generated in login:

When self-hosted: $ kontena master login ... When using cloud: $ kontena cloud platform use ...

Docker

Does kontena support the current latest docker version?

Probably not, see: See https://www.kontena.io/docs/getting-started/system-requirements.html

No logs, but running (until killed because not healthy)

Missing ethwe?

$ docker exec -it e8400c1a57cd ipaddr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
162938: eth0@if162939: <BROADCAST,MULTICAST,UP,LOWER_UP,M-DOWN> mtu 1500 qdisc noqueue state UP
    link/ether 02:42:ac:11:00:08 brd ff:ff:ff:ff:ff:ff
    inet 172.17.0.8/16 scope global eth0
       valid_lft forever preferred_lft forever
    inet6 fe80::42:acff:fe11:8/64 scope link
       valid_lft forever preferred_lft forever

should look like:

docker exec -it ca3867233285 ipaddr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
686: eth0@if687: <BROADCAST,MULTICAST,UP,LOWER_UP,M-DOWN> mtu 1500 qdisc noqueue state UP
    link/ether 02:42:ac:11:00:04 brd ff:ff:ff:ff:ff:ff
    inet 172.17.0.4/16 scope global eth0
       valid_lft forever preferred_lft forever
    inet6 fe80::42:acff:fe11:4/64 scope link
       valid_lft forever preferred_lft forever
690: ethwe@if691: <BROADCAST,MULTICAST,UP,LOWER_UP,M-DOWN> mtu 1376 qdisc noqueue state UP
    link/ether 4a:60:5e:42:7e:cd brd ff:ff:ff:ff:ff:ff
    inet 10.81.138.108/16 scope global ethwe
       valid_lft forever preferred_lft forever
    inet6 fe80::4860:5eff:fe42:7ecd/64 scope link
       valid_lft forever preferred_lft forever

fix: ??

Stacks

Can I expose multiple services in the same stack?

No, but you can install a LB in the stack and expose that.

Alternatively when you expose a service (eg. "node"), all instances are accessible as stack-N

How can I remove something, like secrets from stack?

Set it to something empty like secrets: []

links vs depends_on

I don't understand what "links" (https://kontena.io/docs/using-kontena/stack-file.html#linking-kontena-services) really does for a service. The only use case that is explained in the docs (that I could find) is linking to a load balancer. What else can it do?

It sets the deploy order like depends_on, nothing else. links is also being deprecated in docker-compose.

Services

Does min_health work for services that do not have a healthcheck?

yes.

Master

If you have 3 masters how do the nodes connect to all of them?

Through a LB, see https://www.kontena.io/docs/getting-started/installing/ha-master.html

Registry

Can I declare an insecure Docker private registry (HTTP) in a Kontena YML stack file?

No, Docker requires registries to use HTTPS.

Containers

Connect to a specific instance of docker container?

kontena service exec --instance NUM

Users / Rights

Can a user with "grid_admin" role create grids and nodes?

Can't create grids, but can create nodes.

matti / kontena-faq Goto Github PK

kontena-faq's Introduction