[ENBD] Not enough magic ???
Julien Rennard
jrennard at cri74.org
Mon Jan 5 02:41:30 MST 2004
On Tue, 2003-12-23 at 12:06, Julien Rennard wrote:
> > > localhost:~# ps -e -o pid,wchan=WIDE-WCHAN-COLUMN -o comm | grep enbd
> > > 643 exit enbd-client <defunct>
> > > 644 exit enbd-client <defunct>
> > > 645 lock_page enbd-client
> > > 642 wait_on_page enbd-client
> >
> > Those are kernel memory management functions. The user processes are
> > apparently stuck waiting on memory. Perhaps swap. They are not called
> > from the enbd driver, well, not directly or intentionally, at least.
> >
> > Can you get any information about your memory situation? At least with
> > "free", and with anything more sophisticated if you can think of it.
> free result :
> total used free shared buffers
> cached
> Mem: 126644 120328 6316 0 2684
> 91504
> -/+ buffers/cache: 26140 100504
> Swap: 2546260 264 2545996
>
> It seem to be at the point to use swap (at the end my kernel is perhaps
> not so good...)
>
> > But principally I would like to know if that output you have above ever
> > changes.
> watch "ps -e -o pid,wchan=WIDE-WCHAN-COLUMN -o comm | grep enbd"
> always gives the same result
>
> one time, running it at hand I get :
> 643 exit enbd-client <defunct>
> 644 exit enbd-client <defunct>
> 645 - enbd-client
> 642 - enbd-client
>
>
> > Is this perhaps a shared memory issue? If one client has a locked page
> > and the other is waiting on it, it implies they are communicating
> > through it!
>
> > Have you lost /dev/zero?
> No, seems to be ok.
>
> > Can you run "make test" from the enbd suite on its own? Ie. no floppy.
> make test OK (ouff... I haven't tried it before (and I know it's bad!))
>
Happy new year ! It's starting well : I still have my problem and don't see where it comes from...
Sorry...
> _______________________________________________
> ENBD mailing list
> ENBD at lists.community.tummy.com
> http://lists.community.tummy.com/mailman/listinfo/enbd
>
More information about the ENBD
mailing list