Sometimes it takes a long time for getInt to get the value

[paulaegido]

I have a program that, depending on the value of a variable 'control', has to update the value of another field 'data' per second. I read 'control' every second, and when it goes to 1, I send 'data' every second and I read 'control' every 3 seconds, because if it goes back to 0 I have to stop sending 'data'.

Everything works fine, but the operation of reading 'control' sometimes takes longer than normal, I don't know why, and prevents me from writing 'data' in time, causing that, once I have been able to read 'control', all the dependent registers are written.

Is there a way to improve the reading efficiency of 'control' to avoid the queuing of 'data'? Is it possible to write to 'data' and not be affected by the getInt operation of 'control'?

[mobizt]

If you want to manually reading the data while storing the data, you can use different FirebaseData objects.

One FirebaseData object can use for storing the data or fastest storing as in this example

Another FirebaseData object can use for reading the data.

The time for first connection with server depends on the SSL handshake operation time (almost 1 second or more ) and time to send the request (depends on payload and auth token size).

Normally the TCP (http) session will keep open to reuse, the subsequent request can be performed faster in few hundreds ms without waiting for the SSL handshake as in new connection.

The library will close the TCP session that exceeds 3 minutes, the subsequent request needs to do SSL handshake again, in this case you may feel that the the data reading or storing is slower.

But polling read the data is still not efficient as it involved sending the request and wait for server response which the request payload may be large as 1k depending on the authentication used and can be slow and costs your network bandwidth, you can read this for details.

I recommend to use Stream instead as why it called Realtime database.

See this examples for how to use it.
https://github.com/mobizt/Firebase-ESP32/tree/master/examples/DataChangesListener

[paulaegido]

OK, thank you very much for the answer, it will certainly be very useful. I will try to create two different FirebaseData objects, which is what I know, and also the Stream, but first I must understand the examples.

[mobizt]

Please remember, the library works on REST API instead of web socket which not supported in this environment and platform, which the time for sending each request to read/store data will always depending on the payload and auth token size.

[paulaegido]

Copy that!

[paulaegido]

Ok, I realised that I already had two different FirebaseData objects, so I thought maybe we could try modifying these two parameters:

fbdo_read.setResponseSize(4096); 
Firebase.setReadTimeout(fbdo_read, 1000 * 60);

What would be the optimal values to perform this operation in 1s?

[paulaegido]

Thanks

[mobizt]

This timeout applied when server is already connected and request is being sent and not in the case internet connection issue when trying to connect to server.

[paulaegido]

And there is no timeout in case there is an internet connection problem when trying to connect to the server?

[mobizt]

It wrote in the examples already

Firebase-ESP32/examples/Basic/Basic.ino

Lines 103 to 125 in 408883c

	/** Timeout options, below is default config.
	//WiFi reconnect timeout (interval) in ms (10 sec - 5 min) when WiFi disconnected.
	config.timeout.wifiReconnect = 10 * 1000;
	//Socket begin connection timeout (ESP32) or data transfer timeout (ESP8266) in ms (1 sec - 1 min).
	config.timeout.socketConnection = 30 * 1000;
	//ESP32 SSL handshake in ms (1 sec - 2 min). This option doesn't allow in ESP8266 core library.
	config.timeout.sslHandshake = 2 * 60 * 1000;
	//Server response read timeout in ms (1 sec - 1 min).
	config.timeout.serverResponse = 10 * 1000;
	//RTDB Stream keep-alive timeout in ms (20 sec - 2 min) when no server's keep-alive event data received.
	config.timeout.rtdbKeepAlive = 45 * 1000;
	//RTDB Stream reconnect timeout (interval) in ms (1 sec - 1 min) when RTDB Stream closed and want to resume.
	config.timeout.rtdbStreamReconnect = 1 * 1000;
	//RTDB Stream error notification timeout (interval) in ms (3 sec - 30 sec). It determines how often the readStream
	//will return false (error) when it called repeatedly in loop.
	config.timeout.rtdbStreamError = 3 * 1000;

[paulaegido]

Oh great!! Thanks again! Didn't see that