[stgt] tgtd crash when more than 40 LUNs per target (and a way to reproduce)

[Ming Zhang]

On Fri, 2008-09-26 at 00:12 +0900, FUJITA Tomonori wrote:
> On Thu, 25 Sep 2008 17:05:13 +0200
> Tomasz Chmielewski <mangoo at wpkg.org> wrote:
> 
> > FUJITA Tomonori schrieb:
> > > On Thu, 25 Sep 2008 16:17:02 +0200
> > > Tomasz Chmielewski <mangoo at wpkg.org> wrote:
> > > 
> > >> tgtd crashes if we want to use more than ~43 luns in a target.
> > >>
> > >> To reproduce, use this bash script:
> > >>
> > >> #!/bin/bash
> > >>
> > >> i=1
> > >>
> > >> tgtadm --lld iscsi --op new --mode target --tid 7 -T iqn.2007-02.com.example:test.break
> > >>
> > >> while [ $i -ne 1024 ] ; do
> > >>
> > >>     tgtadm --lld iscsi --op new --mode logicalunit --tid 7 --lun $i -b /dev/sda
> > >>     [ $? -ne 0 ] && exit
> > >>     echo lun $i
> > >>     i=$((i+1))
> > >>
> > >> done
> > > 
> > > You might fail to create pthreads with both problems.
> > > 
> > > This patch enables you to create more luns (but eventually, tgtd
> > > crashes) ?
> > 
> > Yes, with this patch it crashes after ~110 luns or ~60 targets.
> 
> I see, thanks.
> 
> 
> The first problem is that tgtd can't handle the failure to create
> pthreads. I'll try to fix this soon.
> 
> The second problem is seems that glibc has the limit of the maximum
> number of pthreads per process.

reduce the stack size by using attr, you can create much more threads.



> 
> I guess that we create too many pthreads per lun. Probably, it would
> be better to have some pthreads per target (regardless of the number
> of luns). But it takes some time to fix the second problem.
> --
> To unsubscribe from this list: send the line "unsubscribe stgt" in
> the body of a message to majordomo at vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

--
To unsubscribe from this list: send the line "unsubscribe stgt" in
the body of a message to majordomo at vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html