Transient commands should clean up after themselves

[matburt]

Some of the cli commands are transient (ping, send, status, etc). These commands create and persist a directory and manifest record in the run directory for receptor when they don't need to. We could use tmp or just force cleanup the directories that receptor normally creates so that it doesn't pollute the system.

[ghjm]

This was fixed by https://github.com/project-receptor/receptor/pull/128

[matburt]

Closing since it's been 2 weeks.

[Ichimonji10]

Re-opening because we don't actually know whether this issue is fixed.

Where is the run directory? I'm unclear on how to test this issue.

[matburt]

The run directive is either in /var/run or it's wherever you specify with the -d command.

[Ichimonji10]

In one pane:

$ poetry run receptor --data-dir="$(mktemp --directory)" --node-id=my-node node

And in another:

$ ls -d ~/ping
ls: cannot access '/home/ichimonji10/ping': No such file or directory
$ poetry run receptor --data-dir=~/ping ping my-node --peer 127.0.0.1:8888 --count 2
{"initial_time": {"_type": "datetime.datetime", "value": 1584397276.60994}, "response_time": {"_type": "datetime.datetime", "value": 1584397276.712162}, "active_work": []}
{"initial_time": {"_type": "datetime.datetime", "value": 1584397277.611923}, "response_time": {"_type": "datetime.datetime", "value": 1584397277.714023}, "active_work": []}
$ ls -d ~/ping
/home/ichimonji10/ping
$ rm -rf ~/ping
$ ls -d ~/ping
ls: cannot access '/home/ichimonji10/ping': No such file or directory

The continued presence of the "ping" directory indicates that this issue has not been resolved.

[matburt]

So, actually... that directory we expect to stay around as that would be the persistent data root. Underneath that directory we create per-node directories. We couldn't actually delete that root directory without jeopardizing other running Receptor nodes.

[ghjm]

It's not supposed to remove ~/ping - the way you're running these commands, it's supposed to remove ~/ping/my-node. The --data-dir command specifies the top level directory (like /tmp/receptor), but the thing we're cleaning up is the node-specific directory under this.

[Ichimonji10]

Gotcha! I'll re-verify with this in mind.

[Ichimonji10]

I installed receptor-sleep, then executed the following in one pane:

$ poetry run receptor --data-dir="$(mktemp --directory)" --node-id=my-node node

And in another pane:

$ run_dir="$(mktemp --directory)"
$ poetry run receptor --data-dir="${run_dir}" ping my-node --count 2
{"initial_time": {"_type": "datetime.datetime", "value": 1584400344.662021}, "response_time": {"_type": "datetime.datetime", "value": 1584400344.663873}, "active_work": []}
{"initial_time": {"_type": "datetime.datetime", "value": 1584400345.663792}, "response_time": {"_type": "datetime.datetime", "value": 1584400345.766154}, "active_work": []}
$ poetry run receptor --data-dir="${run_dir}" status
Nodes:
  Myself: d5e187d7-2b2f-4d2c-ab11-e35a06efc6c6
  Others:
  - 18976285-5058-4bb5-b03e-a06235fa7a5c
  - my-node

Route Map:
- ('d5e187d7-2b2f-4d2c-ab11-e35a06efc6c6', 'my-node', 1)
- ('18976285-5058-4bb5-b03e-a06235fa7a5c', 'my-node', 100)

Known Node Capabilities:
  my-node:
    worker_versions: {}
    max_work_threads: 16
$ poetry run receptor --data-dir="${run_dir}" send --directive=receptor_sleep:execute my-node '{"duration": 5}'
No ID: iteration 1, response 1
---
$ ls -lA "${run_dir}"
total 0

This issue appears to be resolved.