MagicStack / MagicStack/asyncpg
LOG: could not receive data from client: Connection reset by peer
- Dominant language
- Python
- Stars
- 8.1k
- Forks
- 468
- PR merge metrics
- No merged PRs in 30d
Description
* **asyncpg version**: 0.22.0
* **PostgreSQL version**: 12
* **Do you use a PostgreSQL SaaS? If so, which? Can you reproduce
the issue with a local PostgreSQL install?**: AWS Aurora . Cannot reproduce with a local PostgreSQL.
* **Python version**: v3.7.6+
* **Platform**:
* **Do you use pgbouncer?**: No
* **Did you install asyncpg with pip?**: Yes
* **If you built asyncpg locally, which version of Cython did you use?**: N/A
* **Can the issue be reproduced under both asyncio and
[uvloop](https://github.com/magicstack/uvloop)?**: just uvloop
It seems like asyncpg's connection pooler is not gracefully terminating connections. This seems to happen only with AWS Aurora version, but not the locally installed Postgres version.
Aurora Postgre has `idle_in_transaction_session_timeout` set to `86400000`.
Here is the code to help reproduce this issue. .
```
from sanic import Sanic, Blueprint
from sanic.exceptions import NotFound
from sanic.response import json
from sanic.handlers import ErrorHandler
import asyncpg
app = Sanic(name="asyncpg-errors")
bp = Blueprint("info", url_prefix="/info")
DB_CONFIG = {
'host':,
'user': ,
'password': ,
'port': 5432,
'database': 'postgres'
}
def log_listener(connection, message):
print(message)
@app.listener('before_server_start')
async def init(app, loop):
app.pool = await asyncpg.create_pool(
**DB_CONFIG, loop=loop,
max_size=10, max_inactive_connection_lifetime=1
)
@bp.route('/', methods=['GET'])
async def info(request):
async with request.app.pool.acquire() as conn:
conn.add_log_listener(log_listener)
d = await conn.fetchrow('''SELECT * FROM mmrtest limit 1''')
return json(dict(d))
group = Blueprint.group(bp)
app.blueprint(group)
app.run(port=9091, access_log=True, debug=True)
```
```
Contributor guide
No contributing guide indexed for this repository
Research direction
Start with the supplied Sanic reproduction, focusing on asyncpg.create_pool, max_inactive_connection_lifetime, and the connection acquire/release path. Run it against AWS Aurora and compare its behavior with local PostgreSQL under uvloop. Done means the connection-reset behavior is explained and a verified fix or narrowly scoped next step is documented.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- postgresql, python
- Domain
- backend, databases
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100