A lot of Clients with Solution <none>

[merijnp]

At the moment we are running Servoy version 5.2.8 -build 1016
Last week we had a lot of problems with the clients in our webclient solutions.

Normally we have a few clients in servoy-admin with solution <none>.
I think those clients logged out, after a few minutes the <none> clients disappear.

A few times this week the clients were not cleared.
At one time, we had more than hundred clients with solution <none>,
and also more than hundred in our solution "Webshop".
We believe a lot of those Webshop-clients were in an error state,
because when we look at our own event logs just a few people where active in the solution.
I could not shutdown the clients using solution "Webshop",
only a restart of servoy server removed those clients.
(we check which clients are active, so we can send confirmations to inactive users, so this causes a lot of problems)

What could be the problem? Why doesn't Servoy remove those clients?

Update 15 June 2011:
Yesterday we had the same problem. In the eventlog we have the following errors:
java.sql.SQLException: I/O Error: Connection reset
org.apache.wicket.protocol.http.PageExpiredException: Request cannot be processed
java.lang.IllegalArgumentException: use close method!
A few databaseservers had 5 out of 5 Idle connections.

An example of the problem:

[rgansevles]

Merijn,

This sounds like a connection problem with the db: java.sql.SQLException: I/O Error: Connection reset

There is probably more stack trace to go with that.

Rob

[merijnp]

Rob, next time I'll try to get some stack traces. Since the last time I don't have problems...

[merijnp]

We think this problem is related to our problem with <none> solutions. Every time someone logs out, we see this error message "Cannot open solution" and servoy creates a <none> client. viewtopic.php?f=9&t=16327&p=88053#p88053

[merijnp]

The problem with <none> clients is solved because we now use application.exit(); instead of security.logout();

[merijnp]

This week I had the same problem with clients that were not cleared.
This time the clients were using solution "Webshop".
We use application.exit(); to disconnects clients when they logout or when they are idle for a long time.
I really would like to know how to solve this problem, now we have to restart Servoy Application Server every time it happens.

We upgraded to Servoy version 5.2.9 -build 1020
and started using a new 64 bit server.

Server Information
Servoy version 5.2.9 -build 1020

Operating System Information
os.name=Windows Server 2008 R2
os.version=6.1
os.arch=amd64

System Information
Heap memory: allocated=506560K, used=495613K, max=2796224K
None Heap memory: allocated=93760K, used=93081K, max=180224K

At this moment I don't know how I can find out which solution and/or database server has a problem.
Also I don't know which solution uses the memory, can I find this information somewhere?

I have the following stack traces:
http://orderyourflowers.nl/stacktrace/stackdumpusers1.txt
http://orderyourflowers.nl/stacktrace/stackdumpusers2.txt
http://orderyourflowers.nl/stacktrace/stackdumpusers3.txt
http://orderyourflowers.nl/stacktrace/stackdumpusers4.txt
http://orderyourflowers.nl/stacktrace/stackdumpusers5.txt

[jcompagner]

Merijn,

you first post that it is solved by using exit()
and then you say you still have some problem, so which is it exactly?
What is exactly the problem?

Stackdumps that you attached don't show anything for hanging users because they don't do anything..

EDIT: are there now around 3 clients hanging? (But not that many with solution:<none> as above) ?
Because i do see one thing, and that is currently fixed for 6, will see if that is backported to 5

[merijnp]

Hi Johan,

The problem is that in the Servoy Application Server we have a lot of Clients hanging in solution Webshop (sometimes more than hundred) while most should be disconnected. First we thought that <none> clients caused the problem. But now I don't have <none> clients anymore.

When I look at our own logs, I can see that most of those clients should be disconnected because they logged out of our solution.
Also we throw out clients when they are idle for more then 30 minutes.
I have this problem almost every day with 1 server, and only once at another server (other customer).
I restart the servoy service two times a day as a precaution, and have to throw out clients every time.

I have a hard time finding what's the problem with the clients.
Problem is that I don't know what to look for.

We use 3 databses within the same network.
We use 19 databases that we access via the internet.
We run 3 batch processors that access the database servers.
Clients use 3 webclient solutions. The solution with the client problem is Webshop, and is most used.
I think we also have 1 headless client.
And we use 2 smartclient solutions.

Recently I installed a new Servoy Installation on a 64-Bit Server.
Since we use the new server I have more errors in Server Log.
There could also be some configuration problems.
Every day I have a lot of errors and messages in the Server Log.
("Read timeouts","Connection resets" and more)
But I cannot address those errors to a specific solution or database.

[jcompagner]

in the stackdump you gave in the forum thread, i did see 3 hanging clients but not more.
Those hanging clients problem will be fixed in the next release of 5.2 and in the final of 6.0

[merijnp]

When I have an error in the Server Log. How can I find out to which solution and database (and client) this error belongs?

[jcompagner]

no thats not that easy. If something like that is needed then the logging system must be enhanced so that we also include the clientid and the solutonname on all errors..

[merijnp]

ok thanks, hope the new version of 5.2 solves the problem

[merijnp]

Hi Johan, about the hanging clients problem: in this situation you cannot shutdown a client throught Servoy Application Server? I have that problem now.

[jcompagner]

Yes don't think that will help.

[merijnp]

When I start using final of 6.0 it should be solved?