Hi,
I have checked docs and done many searches on this topic and thought I had it all figured but it just still doesn’t work out.
I have 16 shards in my cluster each is a PSA
I have set the default Write Concern to 1
db.adminCommand({
"setDefaultRWConcern" : 1,
"defaultWriteConcern" : {
"w" : 1
},
"defaultReadConcern" : {
"level" : "available"
}
})
I am insert_many with Python drivers and making sure to set the WriteConcern to 0 as with bulk operations.
curs = db_col.with_options(write_concern=WriteConcern(w=0)).insert_many(data)
Yet still when I shut down an S member of any shard the whole insert stops working until I bring the S back up again.
Any idea what else I can do?
Kobe_W
(Kobe W)
#2
what you mean by this?
timeout? exception? blocking forever?
Majority data bearing nodes should be up to acknowledge writes but Arbiter is a non data bearing
Check this link
I probably should have mentioned it is MongoDB 6.0.4
So it is supposed to work when Write Concern w=1 even with only PA up and S down, Isn’t it?
From the client the queries just stop and sit there not doing anything until S is back up again.
Here is a snippet from the current_ops which looks interesting
'numYields': 0,
'waitingForLatch': {'timestamp': '2023-05-03T08:00:09.812+00:00',
'captureName': 'FlowControlTicketHolder::_mutex'},
'locks': {},
'waitingForLock': False,
'lockStats': {},
'waitingForFlowControl': True,
'flowControlStats': {'acquireWaitCount': 1,