Getting server selection timeout error when maintenance happend on mongodb(mongodb atlas) server

Hi,

I am using mongoose api to connect to mongodb. When any maintenance happend on mongodb i am getting server selection timeout error on running application.

mongodb - 4.0.17 mongoose - 5.7.6 node - 8.12.0

mongoose connection code -

mongoose.Promise = global.Promise;

mongoose.connection.on('connected', function () {
    console.log("MongoDB event connected");
});

mongoose.connection.once('open', function () {
    console.log('Connection to mongodb database is open');
});

mongoose.connection.on('disconnected', function () {
    console.log("MongoDB event disconnected");
});

mongoose.connection.on('reconnected', function () {
    console.log('MongoDB event reconnected');
});

mongoose.connection.on('error', function (err) {
    console.log('MongoDB event error: ' + err);
});
mongoUri = 'mongodb+srv://<userid>:<password>@cluster0-OMITTED.mongodb.net/mydb?retryWrites=true&w=majority';
mongoose.connect(mongoUri , {
    useUnifiedTopology: true,
    useNewUrlParser: true,
    useFindAndModify: false, //to remove findAndModify deprication warming
    autoIndex: false, //false
    reconnectTries: Number.MAX_VALUE,
    reconnectInterval: 500, //In milliseconds
    poolSize: 80,
    socketTimeoutMS: 60000, //Close sockets after 45 seconds of inactivity
}).then(() => {
    console.log("Database connected successfully");
}).catch(err => {
    console.log('App starting error:' + err.stack);
});

[ztrange]

I can confirm that v5.9.10 no longer fails.

I reproduced it with mongo atlas M10 cluster first with mongoose v5.8.11 and got the NotMaster error 3 min after i used the Restart primary button to test failover on Atlas. I ran the test again using 5.9.10 and the failover ended without interrupting writes. My test script was constantly writing to db.

[vkarpov15]

If you’re experiencing a similar issue, please open a new issue and follow the issue template.