Re: Jetty, thread limit -- JAVA400-L

Thanks for the response, Blair.

Since posting my question we've got somebody to admit that the message is 
coming from our layer of the code and it is not as horrible as it looks, and 
the "global limit" is within the application, not the OS (after a couple days 
of pointing fingers at the #*&$#^ iSeries guy ;-) ). 

We're now dealing with the second message about "No thread" and yes, that 
message came from an iSeries but somebody might have assumed that all machines 
are alike when they tried to display the address. We're intermittently taking 
an IllegalMonitorStateException on method File.deleteOnExit and we think that 
may ultimately end up as our "No thread" problem. I submitted a question to 
PartnerWorld today about that.

I KNOW iSeries can run a bunch of threads. I remember hearing about a test with 
a couple hundred thousand. On our system we routinely run 4 or 5 copies of our 
application with a minimum of two hundred threads each. We do have 
QMAXACTLVL=*NOMAX and I have the Max Activity Level for the relevant subsystems 
set between 600 - 2000. Moot anyway since the "global thread limit" wasn't as 
"global" as we originally thought. 

Thanks again,
-Marty

------------------------------

date: Thu, 17 Feb 2005 16:43:33 -0600
from: Blair Wyman <blairw@xxxxxxxxxx>
subject: Re: Jetty, thread limit


java400-l-bounces@xxxxxxxxxxxx wrote on 02/16/2005 10:26:57 AM:

> Is anybody aware of a thread limit within OS/400, specifically V5R3?

Um, well, there probably is a limit, but it's "large."  How many threads
are we talking about here?  I know the iSeries implementation of Java
readily supports thousands of threads at a time...

>
> We are getting some messages to the effect of:
>
> "Warning: The minimum threads exceed the global thread limit,
> meaning that the global limit will be exceeded."

I can't say for sure, but I seriously doubt that this message is coming out
of the JVM layer.  I'm guessing that jetty has some thread limits that it
tries to impose itself.

>
> Later we have some application problems and see messages to the
> effect of a thread not being available:
>
> "WARN!! No thread for com.myco.services.pshttp.jetty.
> ConduitStreamListener$1@ac0be11d"

Ewww, that looks kinda like a 4-byte address tag out there at the end of
that callout (@ac0be11d).  Is that message from an iSeries?

>
> Not sure yet exactly where these messages originate, that is, Jetty,
> IBM, etc, what layer of code.
>
> I have considered the activity level of the subsystem, but it is
> already set quite high and we are not seeing any transitions  to the
> ineligible state, so I am at a loss for where a "thread limit" is
> being imposed. I have perused the sysvals, but don't see anything there.

MAXACTLVL is the relevant gating parameter, I think.  Never hoid of that
'global thread limit' thang.

> I have not encountered this "global thread limit" before and would
> greatly appreciate hearing from anyone who has.
>
> Thanks in advance,
> -Marty

Sorry I couldn't be more help.


  ___   _           Blair Wyman                  IBM Rochester
 ( /_)  /  _  ' _   (507)253-2891            blairw@xxxxxxxxxx
__/__)_/_<_/_/_/_'  Opinions expressed may not be those of IBM