[MESOS-9025] The container which joins CNI network and has checkpoint enabled will be mistakenly destroyed by agent - ASF JIRA

Attach files

Attach Screenshot

Voters

Watch issue

Watchers

Link

Clone

Update Comment Author

Replace String in Comment

Update Comment Visibility

Delete Comments

XML

Word

Printable

JSON

Details

Type: Bug
Status: Resolved
Priority: Blocker
Resolution: Fixed
Affects Version/s: 1.6.0
Fix Version/s: 1.6.1, 1.7.0
Component/s: containerization
Labels:
- cni

Target Version/s:

1.6.1, 1.7.0
Sprint:
Mesosphere Sprint 2018-23
Story Points:
3

Description

Reproduce steps:

1) Run mesos-execute to launch a task which joins a CNI network net1 and has checkpoint enabled:

$ cat task_cni.json
{
  "name": "test1",
  "task_id": {"value" : "test1"},
  "agent_id": {"value" : ""},
  "resources": [
    {"name": "cpus", "type": "SCALAR", "scalar": {"value": 0.1}},
    {"name": "mem", "type": "SCALAR", "scalar": {"value": 32}}
  ],
  "command": {
    "value": "sleep 1000"
  },
  "container": {
    "type": "MESOS",
    "network_infos": [
      {
        "name": "net1"
      }
    ]
  }
}

$ mesos-execute --master=192.168.56.5:5050 --task=file:///home/stack/workspace/config/task_cni.json --checkpoint

2) After task is in the TASK_RUNNING state, restart the agent process, and then in the agent log, we will see the container is destroyed.

...
I0622 17:30:00.792310  7426 containerizer.cpp:1024] Recovering isolators
I0622 17:30:00.798740  7430 cni.cpp:437] Removing unknown orphaned container faf69105-e76f-49c7-8e56-964c2f882cff
...
I0622 17:30:01.025600  7433 cni.cpp:1546] Unmounted the network namespace handle '/run/mesos/isolators/network/cni/faf69105-e76f-49c7-8e56-964c2f882cff/ns' for container faf69105-e76f-49c7-8e56-964c2f882cff
I0622 17:30:01.026211  7433 cni.cpp:1557] Removed the container directory '/run/mesos/isolators/network/cni/faf69105-e76f-49c7-8e56-964c2f882cff'
I0622 17:30:02.935093  7429 slave.cpp:5215] Cleaning up un-reregistered executors
I0622 17:30:02.935221  7429 slave.cpp:5233] Killing un-reregistered executor 'test1' of framework dc2b3db0-953c-47a4-8fd4-f6d040e9d10e-0002 at executor(1)@192.168.11.7:33719
I0622 17:30:02.935900  7429 slave.cpp:7311] Finished recovery
I0622 17:30:02.937409  7427 containerizer.cpp:2405] Destroying container faf69105-e76f-49c7-8e56-964c2f882cff in RUNNING state

And mesos-execute will receive a TASK_GONE for the task:

$ mesos-execute --master=192.168.56.5:5050 --task=file:///home/stack/workspace/config/task_cni.json --checkpoint
I0622 17:29:50.538630  7246 scheduler.cpp:189] Version: 1.7.0
I0622 17:29:50.548589  7261 scheduler.cpp:355] Using default 'basic' HTTP authenticatee
I0622 17:29:50.550348  7263 scheduler.cpp:538] New master detected at master@192.168.56.5:5050
Subscribed with ID dc2b3db0-953c-47a4-8fd4-f6d040e9d10e-0002
Submitted task 'test1' to agent 'dc2b3db0-953c-47a4-8fd4-f6d040e9d10e-S0'
Received status update TASK_STARTING for task 'test1'
  source: SOURCE_EXECUTOR
Received status update TASK_RUNNING for task 'test1'
  source: SOURCE_EXECUTOR
Received status update TASK_GONE for task 'test1'
  message: 'Executor did not reregister within 2secs'
  source: SOURCE_AGENT
  reason: REASON_EXECUTOR_REREGISTRATION_TIMEOUT