I’m experiencing issues reliably starting containers with network = hostnet and native/port-forwarding = true. This is manifesting with an error when starting the container:
ERROR: for boots driver failed programming external connectivity on endpoint boots (f836b9399866a2fbe6b5dc5827f7a428acd69362da4b6a66af05e155dfd9ea0f): Error starting userland proxy: write /port/0.0.0.0:4400/ctl: errno 526
I’m using docker-compose. The docker-compose.yml file has 8 containers with about 20 ports between them, all specified as 5432:5432. This error does not occur when starting containers for the first time after a reboot, but happens reliably after stopping and restarting containers a few times.
It may be worth noting that I cannot resolve docker.local.
I’ve tried using network = nat but it seems to keep reverting to hostnet when I enable port forwarding.
Expected behavior
Containers start as expected.
Actual behavior
Some containers start successfully, others (the same ones, repeatedly) fail to start with an error, ex.:
ERROR: for boots driver failed programming external connectivity on endpoint boots (f836b9399866a2fbe6b5dc5827f7a428acd69362da4b6a66af05e155dfd9ea0f): Error starting userland proxy: write /port/0.0.0.0:4400/ctl: errno 526
Steps to reproduce the behavior
- Enable
native/port-forwarding - Start the containers with
docker-compose create && docker-compse start - Stop the containers with
docker-compose down - Repeat steps 3 and 4 until step 3 has one or more containers fail with errno 526.
Information
I’ve tried reinstalling Beta 8, and am using the default beta 8 configuration with the only change being that I’ve turned on native/port-forwarding.
Rebooting fixes the issue, allowing me to start the containers for the first time. Subsequent restarts of the containers fail. pinata restart sometimes fixes the issue.
> pinata list
...
network = hostnet (docker-ipv4=192.168.65.2, host-ipv4=192.168.65.1)
...
native/port-forwarding = true
Pinata diagnose looks OK:
> pinata diagnose -u
OS X: version 10.11.4 (build: 15E65)
Docker.app: version v1.11.0-beta8.2
Running diagnostic tests:
[OK] docker-cli
[OK] Moby booted
[OK] driver.amd64-linux
[OK] vmnetd
[OK] osxfs
[OK] db
[OK] slirp
[OK] menubar
[OK] environment
[OK] Docker
[OK] VT-x
Docker logs are being collected into /tmp/20160421-152415.tar.gz
Most specific failure is: No error was detected
Your unique id is: 04F763D7-6CE6-47A4-A93B-1CF232968064
Please quote this in all correspondence.
docker-compose.yml:
boots:
image: CENSORED
ports:
- "4400:4400"
- "4700:4700"
- "4800:4800"
- "5100:5100"
- "5200:5200"
- "5400:5400"
- "5500:5500"
- "5600:5600"
- "5700:5700"
- "5900:5900"
- "6100:6100"
- "6300:6300"
- "6400:6400"
consul:
image: CENSORED
ports:
- "8500:8500"
elasticsearch:
image: CENSORED
ports:
- "9200:9200"
volumes:
- "${DOCKER_VOLUME_ROOT}/elasticsearch:/data"
memcached:
image: CENSORED
command: -m 32
ports:
- "11211:11211"
postgres:
image: CENSORED
ports:
- "5432:5432"
volumes:
- "${DOCKER_VOLUME_ROOT}/postgres:/var/lib/postgresql/data"
redis:
image: CENSORED
ports:
- "6379:6379"
volumes:
- "${DOCKER_VOLUME_ROOT}/redis:/data"
riak:
image: CENSORED
ports:
- "8087:8087"
- "8098:8098"
volumes:
- "${DOCKER_VOLUME_ROOT}/riak:/var/lib/riak"
nsqd:
image: CENSORED
command: nsqd -broadcast-address=127.0.0.1 -lookupd-tcp-address=127.0.0.1:4160 -max-body-size=163962880 -max-msg-size=40990720 -mem-queue-size=0 -data-path=/var/lib/nsqd
ports:
- "4150:4150"
- "4151:4151"
volumes:
- "${DOCKER_VOLUME_ROOT}/nsqd:/var/lib/nsqd"
nsqlookupd:
image: CENSORED
command: nsqlookupd -broadcast-address=127.0.0.1
ports:
- "4160:4160"
- "4161:4161"
nsqadmin:
image: CENSORED
command: nsqadmin -lookupd-http-address=127.0.0.1:4161
ports:
- "4171:4171"
I’m running OSX 10.11.4.