Crash when running container with host mount

Docker Desktop for Mac
22

Having the same issue with latest Docker for Mac beta.

Console shows the following:

Sep 29 15:37:12 Fredriks-MBP-2 Docker[91284]: sending SIGTERM to com.docker.hyperkit pid 91288
Sep 29 15:37:14 Fredriks-MBP-2 Docker[91261]: dockerState = Starting
Sep 29 15:37:28 Fredriks-MBP-2 Docker[91282]: Socket.Stream: caught Uwt.Uwt_error(Uwt.ENOTCONN, "shutdown", "")
Sep 29 15:37:29 Fredriks-MBP-2 Docker[91282]: DHCPRELEASE without request ip
Sep 29 15:37:29 Fredriks-MBP-2 Docker[91282]: Socket.Stream: caught Uwt.Uwt_error(Uwt.ENOTCONN, "shutdown", "")
Sep 29 15:38:55 Fredriks-MBP-2 Docker[91261]: dockerState = Starting

Unfortunately, the daemon does not get up and running even after a restart, and this seem to affect the ability to upload diagnostics as well:

Docker for Mac: version: 1.12.1-beta26.1 (12c3e63)
OS X: version 10.11.6 (build: 15G1004)
logs: /tmp/6C609C3B-783E-4D46-9A49-C336F3A029BA/20160929-154553.tar.gz
failure: docker ps failed: Failure("docker ps: timeout after 10.00s")
[ERROR]  docker-cli
         docker ps failed
[OK]     virtualization kern.hv_support
[OK]     menubar
[OK]     moby-syslog
[OK]     dns
[OK]     disk
[OK]     system
[OK]     app
[OK]     osxfs
[OK]     virtualization VT-X
[OK]     db
[OK]     slirp
[OK]     logs
[OK]     env
[OK]     vmnetd
[OK]     moby-console
[OK]     moby
[OK]     driver.amd64-linux
Failure: Could not upload diagnostic data to remote server (docker-diagnose exit code is 1)

Also, I am too new a forum user to upload attachments. (Sorry, new users can not upload attachments.)

If a Docker developer wants the diagnose tar.gz, let me know where to send it.

I see that there is a giant issue in github tracking these kind of issues, and there is a temporary workaround on how to get Docker running again without resetting to factory defaults here: https://github.com/docker/for-mac/issues/119#issuecomment-246635432

I’ll settle for the temporary workaround until a new version comes out. they seem to have enough reported crashes in github already.

23

I have two modes of failure. Initially I got the “transport endpoint is not connected” error. Now I just get a hung process.

What I do:

#Setup
mkdir /tmp/play
docker run -v /tmp/play:/tmp/play -t -i --name play ubuntu:16.04 /bin/bash

#The heavy activity
Run this on the host:

for ((i = 0; i < 1000 * 1000 * 1000; i++)); do dd if=/dev/urandom of=/tmp/play/"$i" bs=4096 count=1 2>/dev/null; done

(I originally wanted to just see if docker is suitable for Symfony development, that’s why the large number in the loop. You’ll get the misbehaviour after about 1000 files)

#The quick death
Run this in the container a few times:

du -sh /tmp/play

It will get slow quickly, and after a few runs you should have either the error originally reported or a hung du. You should kill the heavy activity once it happens.

If it hangs you can still do this:

docker exec -t -i play /bin/bash

And if you run ps aux you’ll see this:

USER       PID %CPU %MEM    VSZ   RSS TTY      STAT START   TIME COMMAND
root         1  0.0  0.1  18240  3200 ?        Ss   19:13   0:00 /bin/bash
root        16  0.0  0.0   4732  1740 ?        D+   19:15   0:00 du -sh /tmp/play
root        17  0.3  0.1  18244  3192 ?        Ss   19:16   0:00 /bin/bash
root        28  0.0  0.1  34424  2860 ?        R+   19:16   0:00 ps aux

Notice the du process in the D state. Any attempt to access /tmp/play at this point hangs.

If it crashes
 the container exits and the daemon stops responding for a little while.

#Software
Docker version 1.12.1, build 6f9534c
OS X 10.11.6

#Hardware
iMac11,2
Intel Core i3 @ 3.06 GHz
12 GB RAM

I can’t Diagnose & Upload because Failure: Could not upload diagnostic data to remote server (docker-diagnose exit code is 1)