Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.os.linux.development.system > #493

Re: Analysing intermitent more CPU for a process.

Show all headers | View raw

Thanks for reply Jorgen,

On Tuesday, March 19, 2013 2:32:29 PM UTC+5:30, Jorgen Grahn wrote:
> On Mon, 2013-03-18, kunal wrote:
> 
> ...
> 
> > 126sendto(65, "\t\0\0\0\0\0\0\0\305\n\0\0\0\0\0\0\0\0\0\0\0\0
> 
>   \0\0\377"..., 164, 0, {sa_family=AF_INET, sin_port=htons(7701),
> 
>   sin_addr=inet_addr("172.18.31.1")}, 16) = 164
> 
> > futex(0x2b22b8000020, FUTEX_WAIT, 2, NULL) = 0
> 
> > futex(0x2b22b8000020, FUTEX_WAIT, 2, NULL) = -1 EAGAIN (Resource temporarily unavailable)
> 
> ...
> 
> 
> 
> > Is there a way to back trace futex to specific api?
> 
> 
> 
> I usually try to combine what strace(1) tells me with what ltrace(1)
> 
> tells me.  tcpdump(1) can also be very helpful, if your application
> 
> mostly does network I/O.

Ok, i shall try ltrace.

> 
> 
> 
> The other key things for me are
> 
> - being able to test the code, i.e. feed it with realistic and
> 
>   unrealistic inputs
> 
> - understanding the code by simply reading it
> 
> 
> 
> > OR attaching any
> 
> > profile runtime, excluding perf(not present)/lstrace(got to try this)?
> 
> 
> 
> I have not tried any such tools, but perhaps I should have.
Sorry, typo error, I meant lsstack.
> 
> 
> 
> /Jorgen
> 
> 
> 
> -- 
> 
>   // Jorgen Grahn <grahn@  Oo  o.   .     .
> 
> \X/     snipabacken.se>   O  o   .

The problem here is that the process is live in network and not able reproduce this scenario in lab and most annoying is that it comes up after 2-3 days after restart so i can't run with oprofile(needs process restart) for such long time.

Back to comp.os.linux.development.system | Previous | Next — Previous in thread | Next in thread | Find similar

Thread

Analysing intermitent more CPU for a process. kunalekawde@gmail.com - 2013-03-18 13:15 -0700
  Re: Analysing intermitent more CPU for a process. Jorgen Grahn <grahn+nntp@snipabacken.se> - 2013-03-19 09:02 +0000
    Re: Analysing intermitent more CPU for a process. Kunal Ekawde <kunalekawde@gmail.com> - 2013-03-19 02:32 -0700
      Re: Analysing intermitent more CPU for a process. Jorgen Grahn <grahn+nntp@snipabacken.se> - 2013-03-19 18:58 +0000

csiph-web