Closed kopax closed 5 years ago
Yea we have discussed that here https://github.com/eBayClassifiedsGroup/PanteraS/issues/171, as you remember problem is that different distribution require different additional files to bind. That would require some additional checks, to probe where is docker / what additional libraries are needed . hmm
This could be added as an extra feature.
I actually tried to map the docker libraries/bin and remove it from PanteraS image, but I still got an error:
mesos-slave stderr | E0629 19:52:10.022145 39 slave.cpp:3252] Failed to update resources for container 537227fd-b020-4b6b-9388-2b889a559dda of executor 'activemq.f3dde0ed-3e32-11e6-b02e-0242d1b97839' running task activemq.f3dde0ed-3e32-11e6-b02e-0242d1b97839 on status update for terminal task, destroying container: Failed to 'docker -H unix:///tmp/docker.sock inspect mesos-91b3a462-272d-4c1a-8190-82c88d58544f-S5.537227fd-b020-4b6b-9388-2b889a559dda': exit status = exited with status 1 stderr = Error: No such image or container: mesos-91b3a462-272d-4c1a-8190-82c88d58544f-S5.537227fd-b020-4b6b-9388-2b889a559dda
Mesos stderr shows:
ABORT: (../../../../../..//tmp/mesos-build/mesos-repo/3rdparty/libprocess/src/subprocess.cpp:322): Failed to os::execvpe on path 'docker': No such file or directory
*** Aborted at 1467230874 (unix time) try "date -d @1467230874" if you are using GNU date ***
PC: @ 0x7fda8ab5dc37 (unknown)
*** SIGABRT (@0x26b6) received by PID 9910 (TID 0x7fda815a7700) from PID 9910; stack trace: ***
@ 0x7fda8aeff330 (unknown)
@ 0x7fda8ab5dc37 (unknown)
@ 0x7fda8ab61028 (unknown)
@ 0x41c0bc _Abort()
@ 0x41c0fc _Abort()
@ 0x7fda8cc137fb process::childMain()
@ 0x7fda8cc1564c std::_Function_handler<>::_M_invoke()
@ 0x7fda8cc1358c process::defaultClone()
@ 0x7fda8cc1440b process::subprocess()
@ 0x7fda8c336a4f Docker::run()
@ 0x443091 mesos::internal::docker::DockerExecutorProcess::launchTask()
@ 0x7fda8cbe2451 process::ProcessManager::resume()
@ 0x7fda8cbe2757 _ZNSt6thread5_ImplISt12_Bind_simpleIFSt5_BindIFZN7process14ProcessManager12init_threadsEvEUlRKSt11atomic_boolE_St17reference_wrapperIS6_EEEvEEE6_M_runEv
@ 0x7fda8b6e0a60 (unknown)
@ 0x7fda8aef7184 start_thread
@ 0x7fda8ac2137d (unknown)
W0629 20:07:54.461889 9900 logging.cpp:88] RAW: Received signal SIGTERM from process 13 of user 0; exiting
It's like it can't find docker
. When I get into the container I can run docker commands as expected.
Here's what I mapped:
- "/etc/resolv.conf:/etc/resolv.conf.orig"
- "/var/spool/marathon/artifacts/store:/var/spool/store"
- "/var/run/docker.sock:/tmp/docker.sock"
- "/var/lib/docker:/var/lib/docker"
- "/sys:/sys"
- "/tmp/mesos:/tmp/mesos"
- "/proc:/host/proc:ro"
- "/sys:/host/sys:ro"
- "/bin/docker:/usr/local/bin/docker"
- "/usr/lib/libdevmapper.so.1.02:/usr/lib/libdevmapper.so.1.02"
I have open that, so make it with a proper way https://github.com/docker/compose/issues/3979
we now use only docker client
I am trying to understand why do we install docker inside PanteraS again. Isn't it possible to use the docker daemon from the host ?
I did it with jenkins by mounting theses volume :
This way it will be easier to upgrade/downgrade the docker version.
By the way, current installed marathon version can't rollback configuration because of a JS error by.