Smart Clients not starting for 3 hours

Classic Servoy
1

Sometimes, the smart clients of our users do not start up, i.e. the login screen does not appear.
The users that are already logged in can work as usual.

It it not a matter of an insufficient number of licenses, I checked that already.

There are no messages to the user. The duration of the startup failure is different, but recently it was up to three hours for all users.
Startup method is by Java WebStart, e.g.
http://deXXX.YYY.ZZZ.net:8080/servoy-client/bob.jnlp

Did anybody experience something similar or has someone a tip what that might be?

2

can you enable the webstart console for that
and then start up the client with:

http://deXXX.YYY.ZZZ.net:8080/servoy-cl … acing=true

then you should get a lot of logging to see what happens, it could be that for some reason it just can’t connect,
on what version of servoy are you?

3

Thank you for the tip, we will prepare that.

We are on Servoy 7.4.9
The issue appeared already at begin of 2017, but as it got better since then, we stopped to examine it.
But now it is back again.

I have already asked the network experts to install a port sniffer to check if another system blocks port 1099.

Another idea would be to switch from a virtual server to a dedicated server - who knows what the other virtual servers do on the one shared physical server, and if a virtual server is really 100% separate…

4

Could it be a license problem?
Because we have two licenses integrated, one for 150 clients and one for 50 clients.

One theory could be that the Servoy server is not properly adding them inside, but refusing more connections already for client number 151.

5

could be but you should see this then at the admin page.

6

I asked the admin of my customer to get more than 150 users to login. Then I should know more.

7

The license number was not the problem, we succesfully could login more users than the number that one of the licenses allowed, so adding up license numbers works well.

I had a look to the Servoy Server log now and could identify four error message types that are quite frequent there.
Could one of them give a clue why login is not possible?

(What exactly happens is: only the startup logo appears, but no login screen appears)

Because the host IP address is always different in the error messags below,
and also the server id is different, I assume that “host” means in this context the servoy client, is that correct?

java.rmi.ConnectException: Connection refused to host: 10.116.xxx.yyy; nested exception is:
java.net.ConnectException: could not connect to port=5139 of server with id=1110522c-fcbf-41c7-bd4e-e7cb8XXX
at sun.rmi.transport.tcp.TCPEndpoint.newSocket(Unknown Source)

java.rmi.ConnectException: Connection refused to host: 10.110.xxx.yyy; nested exception is:
java.net.ConnectException: could not connect to port=4833 of server with id=d23f3ab3-2fa0-4164-b0d5-9a2a78cXXXX

java.rmi.ConnectException: Connection refused to host: 10.116.xxx.yyy; nested exception is:
java.net.ConnectException: could not connect to port=4794 of server with id=9182f6af-9f70-4cef-a5c3-84857XXX

java.rmi.ConnectException: Connection refused to host: 192.168.xxx.yyy; nested exception is:
java.net.ConnectException: no multiplexer found for server with id=e070aad6-96c2-4dcf-a02a-781b1a1XXXXX
at sun.rmi.transport.tcp.TCP

8

yes stuff in the log files of the server are just that the server can’t reach those clients anymore (because the are suddenly gone or not nicely terminated)

you need loging from the client side. And maybe then test if you can telnet to the 1099 port

Also you could test to have a fallback if you place the tunnel mode on Http&Socket (socket is then default) and make a profile for starting a smart client in http mode. Does that then do work…

9

Is there something that can be seen from the following startup log?

<?xml version="1.0" encoding="windows-1252" standalone="no"?>
<!DOCTYPE log SYSTEM "logger.dtd">
<log>
<record>
  <date>2019-10-22T14:51:21</date>
  <millis>1571748681494</millis>
  <sequence>0</sequence>
  <logger>com.sun.deploy</logger>
  <level>FINE</level>
  <class>com.sun.deploy.trace.LoggerTraceListener</class>
  <method>print</method>
  <thread>8</thread>
  <message>CacheEntry[http://de19aXXX.XXX.XXX.net:8080/servoy-client/bob.jnlp]: updateAvailable=false,lastModified=Wed Oct 16 19:46:15 CEST 2019,length=-1
</message>
</record>
<record>
  <date>2019-10-22T15:03:37</date>
  <millis>1571749417449</millis>
  <sequence>1</sequence>
  <logger>com.sun.deploy</logger>
  <level>FINE</level>
  <class>com.sun.deploy.trace.LoggerTraceListener</class>
  <method>print</method>
  <thread>8</thread>
  <message>bit32or64? 32
</message>
</record>
10

that log doesn’t say anything has almost no data.

11

Yes right, but that is what we got. And it is not unlikely that no data is produced when login is not working right at the start.

I will continue in analyzing which other systems are using port 1099.

Regarding the tunnel mode, I am currently a bit reluctant to consider a change to that, because I remember that performance dropped very much when I changed it during the beginning of the system.