apache / apache/pulsar-client-cpp

[Bug] Getting ConnectError when expecting AuthenticationError

Open
#227 1 comment 0 reactions 0 assignees View on GitHub
Dominant language
C++
Stars
71
Forks
90
Avg merge
2h 33m
Merged PRs (30d)
3

Description

### Search before asking

- [X] I searched in the [issues](https://github.com/apache/pulsar/issues) and found nothing similar.

### Version

pulsar-all image version 2.10.32 and pulsar-all image version 2.11.0
Token auth enabled
Python client version 2.10.1

### Minimal reproduce step

Token auth is correctly set up and tested to be working with a good token.

Again, with token auth enabled, try to connect with a (python) client and provide a broken token (try various ways to break it, including a malformed token).

I tried this with a standard token mechanism, but I also tried this with a custom Auth plugin where other token fields can be broken in similar ways.

### What did you expect to see?

Expect a ConnectError when there are network reasons for not being able to establish connection.

Expect an AuthenticationError with a well-formed token without permission to use Pulsar resources.

Expect an AuthenticationError when there is a bad token; but get ConnectError.

We got reports from users that ConnectError was confusing when debugging issues.

### What did you see instead?

ConnectError when a bad token is provided.
```
bad_admin_token = "eyJhbGciOiJIUzI1aaaaaaaNiJ9.eyJzdWIiOiJhZG1pbiJ9.1Eu3ph4RneevmKIKlE9gglAAyFBJMeN7GM-ZnHLZNsM" # ConnectError
pulsar_service_url = 'pulsar://localhost:6650/'
client = pulsar.Client(pulsar_service_url, authentication=pulsar.AuthenticationToken(bad_admin_token))

try:
producer = client.create_producer('my-topic')
for i in range(10):
producer.send(('Hello-%d' % i).encode('utf-8'))
client.close()
except Exception as e:
print(e.args)
print(dir(e))
print()
raise
```
Get this result:
```
2023-03-21 10:53:02.074 INFO [0x306688000] ExecutorService:41 | Run io_service in a single thread
2023-03-21 10:53:02.075 INFO [0x202b38600] ClientConnection:189 | [ -> pulsar://localhost:6650/] Create ClientConnection, timeout=10000
2023-03-21 10:53:02.086 INFO [0x202b38600] ConnectionPool:96 | Created connection for pulsar://localhost:6650/
2023-03-21 10:53:02.129 INFO [0x306688000] ClientConnection:375 | [[::1]:64418 -> [::1]:6650] Connected to broker
2023-03-21 10:53:02.953 INFO [0x306688000] ClientConnection:1560 | [[::1]:64418 -> [::1]:6650] Connection closed
2023-03-21 10:53:02.954 ERROR [0x306688000] ClientImpl:194 | Error Checking/Getting Partition Metadata while creating producer on persistent://public/default/my-topic -- ConnectError
2023-03-21 10:53:02.954 ERROR [0x306688000] ClientConnection:597 | [[::1]:64418 -> [::1]:6650] Read operation failed: Bad file descriptor
2023-03-21 10:53:02.955 INFO [0x306688000] ClientConnection:263 | [[::1]:64418 -> [::1]:6650] Destroyed connection
('Pulsar error: ConnectError',)
['__cause__', '__class__', '__context__', '__delattr__', '__dict__', '__dir__', '__doc__', '__eq__', '__format__', '__ge__', '__getattribute__', '__gt__', '__hash__', '__init__', '__init_subclass__', '__le__', '__lt__', '__module__', '__ne__', '__new__', '__reduce__', '__reduce_ex__', '__repr__', '__setattr__', '__setstate__', '__sizeof__', '__str__', '__subclasshook__', '__suppress_context__', '__traceback__', '__weakref__', 'args', 'with_traceback']

2023-03-21 10:53:02.967 INFO [0x306688000] ExecutorService:47 | Event loop of ExecutorService exits successfully
Traceback (most recent call last):
File "/Users/******/PycharmProjects/pythonProject/local.py", line 36, in
print_hi('PyCharm')
File "/Users/******/PycharmProjects/pythonProject/local.py", line 24, in print_hi
producer = client.create_producer('my-topic')
File "/Users/thors/PycharmProjects/pythonProject/venv/lib/python3.9/site-packages/pulsar/__init__.py", line 603, in create_producer
p._producer = self._client.create_producer(topic, conf)
_pulsar.ConnectError: Pulsar error: ConnectError

Process finished with exit code 1
```

Again, our users complain that this error is hard to debug in logs because it would have you troubleshoot connectivity rather than problems with authentication setup.

### Anything else?

I have tested this with both an admin token and lesser permissions tokens.
I have also tested breaking the token in various parts that would fail to parse it.
Also tested against a custom Auth plugin where we implement token checks.

### Are you willing to submit a PR?

- [ ] I'm willing to submit a PR!

Contributor guide

Open the contributing guide

Research direction

Start by reproducing the malformed-token case with the Python client against the Pulsar service versions described, then trace the C++ client's authentication failure handling. Done means malformed or unauthorized tokens surface as AuthenticationError while network failures continue to surface as ConnectError.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp, python
Domain
authentication
Issue type
Bug
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
42/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.