bergman on 18 Apr 2016 14:40:29 -0700

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: [PLUG] >32K concurrent processes

In the message dated: Mon, 18 Apr 2016 16:20:30 -0400,
The pithy ruminations from "K.S. Bhaskar" on 
<Re: [PLUG] >32K concurrent processes> were:

Hellooooooo 2003!

	Cannot create more than 32K threads?

Seriously, the program referenced there
maybe useful as a means of testing whether the limit of 32K processes you're
seeing is really an intrinsic limit (possibly adjustable via kernel tuning)
or whether this is a limit related to your application. 

Don't worry, I'm not accusing your database itself of being unable to scale, but
perhaps there's a resource used by the database that is limiting the process
count, not that the server is limited to 32K of any kind of process.

For example, is the application hitting a shared-memory limit at 32K client
instances, where another test (ie. "while i < 32K ( fork(sleep()))")
wouldn't hit that limit?

Also, I'd do the really simple test of:
	(ulimit -a ; sysctl -a  )| grep 3276

and see if there's anything relevent.

I know you said that the test isn't network-based, but this might be useful:

Their description of failure modes might sound similar:

	CPU usage spiked to 100% in the Java process, with nothing output to
	the console. It appeared that an exception was being thrown and silently
	caught in a tight infinite loop.

Perhaps the problem isn't with the system, but with the program used to
create the clients?


Mark Bergman    Biker, Rock Climber, SCUBA Diver, Unix mechanic, IATSE #1 Stage
'94 Yamaha GTS1000A^2

I want a newsgroup with a infinite S/N ratio! Now taking CFV on:
15+ So Far--Want to join? Check out: 
Philadelphia Linux Users Group         --
Announcements -
General Discussion  --