SystemHandler action refused - Problem

[lomriilod]

Hi,

we are having serius problems with the server right now. What we are experiencing (and the users are reclaiming) is simply described 'lagging'.
We are intensivly searching, but can't realy track the problem.

I found some SystemHandler messeges I can not explain.

Code: Select all

2008/09/11 16:47:28.516 - [ FINE ] [id: 16] (SystemHandler.processEvent): SystemHandler action refused: autoJoin, from: 127.0.0.1

Nobody tried to connect from the local IP.

Code: Select all

2008/09/11 16:48:04.104 - [ FINE ] [id: 16] (SystemHandler.processEvent): SystemHandler action refused: roundTrip, from: xx.xx.xx.xxx

This happens lots of times, but none of our extensions has disabled the round trip system action.

What seems to help right now is to increase the ExtHandlerThreads.

Could a lot roundtrip requests lead to a processEvent error when using 1 ExtHandlerThreads?

Thanks

Dirk

[lomriilod]

I know know, that localhost connections are from the BueBox.

Now we are receiving the roundTrip errors as well from the BlueBox.

Code: Select all

(SystemHandler.processEvent): SystemHandler action refused: roundTrip, from: 127.0.0.1

I listed zone.isSysActionDisabled("roundTripBench") from all zone running on the server. None of them has round trip disabled.

I am still stucked.

[Lapo]

Could a lot roundtrip requests lead to a processEvent error when using 1 ExtHandlerThreads?

No, especially because it is "System Handler" (SysHandler) that is responsible for those messages.
Also the "lagging" is unlikely to be related with those refused messages.

The "lag" could be cause by many messages being queued up in the SysHandler Queue ... are you monitoring the AdminTool? What does it say?
(Btw, what version do you use?)

[lomriilod]

Version 1.6.3

The admin panel indicates that there is no problem with neither of the message handlers. All og them show 0 messeges waiting and the load is arround 0 %. The most dropped messages we had were 203 - but not right when this happended.

Dirk

[lomriilod]

...no more ideas?

[Lapo]

The two issues you have reported don't seem to be related.

As regards the "lagging" issue, it's pretty generic and we would need more details to understand what might be going on.

Is it lagging for everyone? Or it's just someone complaining?
If you test on your side, do you have lag problems?

In general I would recommend to take a look at your bandwidth usage, and also monitor the CPU and RAM usage of the server machine.

As regards the second problem, we are not aware of a similar problems. Those messages should be discarded only if the configuration is set so.
It would be probably better to analyze the log files and see if there's something else going on.

[lomriilod]

Hi,

lets ignor any lagging problems.

What could cause the roundTrip error? It is flooding the logs and I hav no idea where it comes from.

Is there any specific info I could give you that might help at least to point in a direction where to look further ?

Thanks.

Dirk

[Lapo]

First of all, to avoid having huge log files it would be better to set a INFO level, instead of FINE or even lower. These levels are more for debugging with fewer users than in production...

Anyways about the refused actions, it sounds pretty strange.
I would suggest two possible options, both have to be verified...

1. maybe the message is really not recognized because of some non-printable character in the command name? Would be quite odd... it could be done by a custom made client, created for flooding/cracking purposes

2. maybe there's a problem in your client side code, which floods the server? This one is as strange as the previous.
Do you use an interval in your code to fire those events?
How fast do you send those messages?
Are you sure you're clearing the interval/timer correctly?

We have seen problems with Timers when developers forget to clear intervals/timer listeners and these objects get duplicated creating a small "monster flooder"

Another question. Can you replicate the problem in a test environment or is it erratic?