Incorrect error message `MasterNotFoundError: No master found for 'pre'` when timeout.

Version: 2.10.6 Platform: 3.6 on win 10

Description:

if socket_timeout=0.1 :

sentinel = Sentinel([("192.168.0.10", 26379),("192.168.0.11", 26379),("192.168.0.12", 26379)], password='xxxxxx', socket_timeout=0.1)
sentinel.discover_master('pre')

got error:

C:\Anaconda3\lib\site-packages\redis\sentinel.py in discover_master(self, service_name)
    221                     sentinel, self.sentinels[0])
    222                 return state['ip'], state['port']
--> 223         raise MasterNotFoundError("No master found for %r" % (service_name,))
    224
    225     def filter_slaves(self, slaves):

MasterNotFoundError: No master found for 'pre'

But remove socket_timeout :

sentinel = Sentinel([("192.168.0.10", 26379),("192.168.0.11", 26379),("192.168.0.12", 26379)], password='xxxxxx')
sentinel.discover_master('pre')

got ('192.168.0.12', 6316)

This misleading message waste me lot of time .

[Cimmanuel]

Getting the same error when i try to configure with airflow celery AIRFLOW__CELERY__BROKER_URL=“sentinel://$SENTINEL_HOST:$SENTINEL_PORT” AIRFLOW__CELERY_BROKER_TRANSPORT_OPTIONS={‘master_name’:“mymaster”}

Resolved!

How did you solve this please?

[andymccurdy]

Setting socket_timeout configures the socket to raise a timeout after socket_timeout seconds have elapsed. Setting this value too low for your environment will cause every command to a Redis or Sentinel server to timeout.

In this particular case, the redis-py Sentinel code catches the TimeoutError and tries the next Sentinel server. If none of the Sentinel servers respond, the MasterNotFoundError is raised.