feat: Remove automate in main

[gretelliz]

Closes #77

Summary

According to main.py, as well as TRIGGER_SCHEDULE_TRACES, TRIGGER_IMPORTANT_CIRCUITS and FIND_CIRCUITS_IN_FLOWS set to False, automate.py does not seem to be needed. So, this is to remove the dependency of main.py to automate.py.
Then, automate.py won't be used anymore and issue #39 is not needed.
Instead, a timeout has been added to the requests from stored_flows to flow_manager.

Local Tests

Everything seems to continue working well.

2023-04-28 14:47:37,787 - INFO [kytos.napps.kytos/mef_eline] (mef_eline) EVC(6875c026f20a4b, epl) was deployed.
2023-04-28 14:47:37,788 - INFO [kytos.napps.kytos/mef_eline] (mef_eline) EVC(5b1147f173c84a, epl) enabled but inactive - redeploy

/traces call example:

[
    {
        "trace": {
            "switch": {
                "dpid": "00:00:00:00:00:00:00:01",
                "in_port": 1
            },
            "eth": {
      "         dl_vlan": 10
            }
        }
  }
]

{
    "result": [
        [
            {
                "dpid": "00:00:00:00:00:00:00:01",
                "port": 1,
                "time": "2023-04-28 14:42:30.554977",
                "type": "starting"
            },
            {
                "dpid": "00:00:00:00:00:00:00:02",
                "out": null,
                "port": 2,
                "time": "2023-04-28 14:42:30.555051",
                "type": "incomplete"
            }
        ]
    ]
}

End-to-End Tests

N/A

[viniarck]

It's almost there @gretelliz, asked minor changes.

Also, when locally exploring it, make sure that you've seen the @retry decorator work at least once. The easiest way you can keep a local code change raising one of the retried exceptions like raise Timeout(...) and observing. Finally, please, add a unit test simulating a side effect raising http 424, and then ensuring that /traces and /trace will return the expected response.

[viniarck]

@gretelliz nicely done, looks good to me.

Did you run a local exploratory test to see the retry decorator working at least once? It looks good, but waiting for this confirmation before approving. Also, unfortunately you'll need to solve some git conflicts since chore/starlette has landed #94.

[gretelliz]

@viniarck, I temporarily modified flow_manager to sleep(100) before returning stored_flows.
Then, with /traces:

{
     "description": "It couldn't get stored_flows",
     "code": 424
}

2023-05-04 12:51:00,490 - WARNING [kytos.core.retry] (AnyIO worker thread) Retry #1 for get_stored_flows, args: (), kwargs: {}, seconds since start: 20.01
2023-05-04 12:51:20,660 - WARNING [kytos.core.retry] (AnyIO worker thread) Retry #2 for get_stored_flows, args: (), kwargs: {}, seconds since start: 40.18

I had to change mef_eline to catch Timeout, because it just caught ConnectTimeout and it failed in the local test.
Now:

2023-05-04 13:14:41,933 - WARNING [kytos.core.retry] (AnyIO worker thread) Retry #1 for get_stored_flows, args: (), kwargs: {}, seconds since start: 20.00
2023-05-04 13:14:51,901 - ERROR [kytos.napps.kytos/mef_eline] (mef_eline) Request has timed out: HTTPConnectionPool(host='localhost', port=8181): Read timed out. (read timeout=30)
2023-05-04 13:15:02,080 - WARNING [kytos.core.retry] (AnyIO worker thread) Retry #2 for get_stored_flows, args: (), kwargs: {}, seconds since start: 40.15

Note that I merged master and PR#94 is considered.
Now, when I start kytos I get this error:

2023-05-04 13:29:08,907 - ERROR [kytos.napps.kytos/mef_eline] (MainThread) Could not load EVC: dict={...

Do you have any ideas?

[viniarck]

2023-05-04 13:29:08,907 - ERROR [kytos.napps.kytos/mef_eline] (MainThread) Could not load EVC: dict={...

Do you have any ideas?

@gretelliz, overall looks good.

Do you have the rest of error content? It's coming from _evc_from_dict, I wouldn't expect it to fail.

[viniarck]

Nicely done @gretelliz.

It turns out the mentioned error wasn't related to this PR, it was the order that the NApp was loading, so running a touch kytos.json solved it.

[gretelliz]

Thank you, @viniarck