Insufficient data for recovery on bind_session.js call

torquyme

Hello,

I've been using Lightstreamer 7.0.3 for a long time, but recently after i increased the size of the message sent by the adapter the following issue appeared:

As soon as the message is sent from the java data adapter to the lightstreamer server, on the browser i get this

The browser tries indefinitely times to reconnect and to create/bind to a session but the response is the following:

Could you please help to figure it out? I'm getting crazy

p.s. no logs are sent to the server or are written by the lightstreamer server, basically i don't have any info about this issue

Alessandro.Carioni

Hello,

What server version are you using? It is strange that the server doesn’t log anything. Please make sure that, when you are debugging the server in a test environment, the logger named LightstreamerLogger.requests(in the file lightstreamer_log_conf.xml ) is set at INFO at least.

Anyway, the best way to investigate such kind of problems is to activate the client internal logger and send us an extract of the generated log.
To enable the logging, you should add to your code the following lines:

var log = new SimpleLoggerProvider();
log.addLoggerAppender(new ConsoleAppender("DEBUG", "*"));
LightstreamerClient.setLoggerProvider(log);

For instance, you can add them before the initial creation of the instance of LightstreamerClient.
Keep in mind that the log is very verbose and it should be activated only in test environment.

Regards

torquyme

Hello Alessandro,

Thank you for your reply.
I managed to attach the ConsoleAppender to the LightstreamerClient and collected some logs about the issue.
Please find it attached.

Regards

Alessandro.Carioni

Please, can you send me the entire log as a text file at support@lightstreamer.com?

Thank you

Dario Crivelli

Hello,

We have received your logs. They show the frequent outages that your client experiences.
In practice, some time after a streaming session is established, the client receives no more bytes from the Server.
This is a situation that can happen for various causes, but usually it is related with poor connectivity.
So, Lightstreamer tries to recover the existing streaming session with a new connection.
However, the attempt may either succeed or fail depending on how much data was lost during the interruption.
If the recovery attempt fails, Lightstreamer has to resume by opening and reconfiguring a new session.

Since you say that the recovery attempts started to fail after you increased the amount of data, this is consistent.
In practice, our suspect is that connectivity issues have always occurred, but before the recoveries were successful and the malfunctionings went unnoticed.
You can increase the recovery capability of the Server by leveraging the <max_recovery_length> configuration flag.
By default, it is low; you can try to set, for instance, 5000 bytes and see if things improve.
On the other hand, if you feel that connectivity issues are unexpected and need to be investigated separately, please let us know.

torquyme

Hello,

Thanks for your reply. I think it might not be related to connectivity issue, because during the first production deploy we noticed the issue was happening with a lot of clients; it is also important to say that the issue happens only on production environment and not on our test environment.
I checked the configuration and we already have the <max_recovery_length> param set to 5000 bytes.

Maybe some other kind of logs can help? Do you have any suggestion?

Thanks,

Regards

Dario Crivelli

Assuming that your setting of <max_recovery_length> was left as the default, we suggested to increase it to 5000.
Since it was already 5000, then try 15000.
The aim is to see if the recovery capabilities of Lightstreamer can be increased, so that when you currently see "Insufficient data for recovery" you can see a successful recovery.
This is related to what you wrote: "recently after i increased the size of the message sent by the adapter". If the messages in transit are now big, the amount of data that should be kept to ensure recovery should be determined accordingly.
The hope is then that successful recovery from temporary outages can change the client experience significantly.

About the cause of the communication outages, for now there is no evidence that it can be caused by a Lightstreamer malfunctioning.
The Server log sent is very short (300 ms of activity), yet it shows normal activity on various sessions.
So, the interruption on a single session seems a client connectivity issue.
Only if you notice that many clients experience problems at the same time can we suspect a Server or local network problem.

For a preliminary check, you can send us a log of the Internal Monitor.
Please set the LightstreamerMonitorText logger at TRACE level in lightstreamer_log_conf.xml and send us the log produced in a few hours. But please, for now, take care of extracting only the log produced by this logger, otherwise the log could be very long.
With the current settings, you should grep for the "streamerMonitorText" pattern.