PyMongo fails to read from the secondary in a replica set setup when the primary is down

Working with Data Python Frameworks
,
1

I have configured a replica set with primary ‘n1’ (default port), second ‘n2’ and arbiter n1:27018. n1 and n2 are configured in /etc/hosts and are resolved as IPs. My replica set also uses those names with no issue.
When both nodes are up all is working as expected. However if n1 goes down , PyMongo can’t read the DB and fails with timeout. I’ve read everything I managed to find on the topic and appears the proper way to read from the secondary in such case is to set ReadPreference to ‘secondary’, but this approach also times out and fails.
Never mind my code , I’m trying very basic test directly in a python shell - here is what I did

Blockquote
from pymongo import ReadPreference

from pymongo import MongoClient

uri = “mongodb://n1:27017,n2:27017,n1:27018/?replicaSet=rs0” # n1:27018 is my arbiter

client = MongoClient(uri, socketTimeoutMS=6000,connectTimeoutMS=6000,readPreference=‘secondary’)

configdata = client.get_database(‘audiobridge’, read_preference=ReadPreference.SECONDARY)

Blockquote
When I print configdata it shows:

Blockquote

Blockquote
Database(MongoClient(host=[‘n1:27017’, ‘n2:27017’, ‘n1:27018’], document_class=dict, tz_aware=False, connect=True, replicaset=‘rs0’, sockettimeoutms=6000, connecttimeoutms=6000, readpreference=‘secondary’), ‘audiobridge’)
Blockquote
And yet when I try :

Blockquote

Blockquote

configdata.ha.find_one({‘ha_exists’: True})

Blockquote

it crashes with :

Blockquote
pymongo.errors.ServerSelectionTimeoutError: No replica set members match selector “Secondary(tag_sets=None, max_staleness=-1, hedge=None)”, Timeout: 30s, Topology Description: <TopologyDescription id: 6756cc9f09429c36ed83449d, topology_type: ReplicaSetNoPrimary, servers: [<ServerDescription (‘n1’, 27017) server_type: Unknown, rtt: None, error=NetworkTimeout(‘n1:27017: timed out (configured timeouts: socketTimeoutMS: 6000.0ms, connectTimeoutMS: 6000.0ms)’)>, <ServerDescription (‘n1’, 27018) server_type: Unknown, rtt: None, error=NetworkTimeout(‘n1:27018: timed out (configured timeouts: socketTimeoutMS: 6000.0ms, connectTimeoutMS: 6000.0ms)’)>, <ServerDescription (‘n2’, 27017) server_type: RSOther, rtt: 0.0010362059028441717>]>

Blockquote

Any suggestions , ideas are much appreciated

2

The will be no secondary available in that setup if one of the data bearing members fail. Just a Primary and the Arbiter. Using secondaryPreferred will work.

3

Thank you,
This really was the case. My secondary was not recognised as a ‘secondary’ , but ‘other’ as I have added it with option ‘hidden’. Once I created changed this configuration , pymongo started working as expected.

1 Like