Containers Frequently Disappearing - osxfs problem?

Expected behavior

Containers to persist on the system for more than 5 minutes.

Actual behavior

Containers are running, then in an instant are gone and not visible even with docker ps -a

Information

  • the output of:
    • pinata diagnose -u on OSX

pinata diagnose -u
OS X: version 10.11.3 (build: 15D21)
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/20160426-084540.tar.gz
Most specific failure is: No error was detected
Your unique id is: 588A8E5D-109A-443C-B9C2-85C2364BF012
Please quote this in all correspondence.

Only error logs I can find:
com.docker.osxfs.log

2016-04-26T08:36:32.847723-05 Debug - stdout and stderr have been redirected to /dev/null
2016-04-26T08:36:32.847776-05 Notice - Logging to Apple System Log
2016-04-26T08:36:39.512419-05 Debug - received init from client
2016-04-26T08:36:39.513125-05 Debug - Event thread is 1077
2016-04-26T08:36:39.513693-05 Error - transfused: mount
2016-04-26T08:36:39.513847-05 Error - transfused: /bin/fusermount
2016-04-26T08:36:39.513972-05 Error - transfused: -o
2016-04-26T08:36:39.514104-05 Error - transfused: allow_other,max_read=1048576,subtype=osxfs
2016-04-26T08:36:39.514242-05 Error - transfused: /Mac
2016-04-26T08:36:39.514391-05 Error - transfused:
2016-04-26T08:36:39.514526-05 Debug - max_readahead = 131072 max_write = 1048576
2016-04-26T08:36:39.514614-05 Notice - osxfs tracing: OFF
2016-04-26T08:36:39.514681-05 Debug - sending continue to client
2016-04-26T08:36:39.834898-05 Debug - setting atime
2016-04-26T08:36:39.845201-05 Debug - setting atime

Steps to reproduce the behavior

  1. My best guess is that it is a throughput issue that is breaking the mount. The last 2 times it crashed it was while I was doing a large import into a container based MySQL instance (500MB mysql import). I was using docker stats to monitor the load, and half way through all the containers disappeared and were not listed in history (docker ps -a)

Hi Bryce,

We are investigating a load-related VSOCK crash that you may have run across. I don’t think a fix will make it into Beta 9 but we are hopeful that it can be resolved for Beta 10.

Thanks for your report and please accept my apologies for this frustrating defect.

Best regards,

David Sheets