Re: bind9 thread problems?

[Harry Waddell <waddell%caravan-epub.com@localhost>]

On Thu, 31 Jul 2014 14:57:39 +0200 (CEST)
Havard Eidnes <he%uninett.no@localhost> wrote:

> > I have a mid sized server ( E5 with 12 cores )
> > running netbsd 6.1.4 amd64 with pkgsrc from the pkgsrc-2014Q2 branch
> > on which named hangs soon after startup. This happens with the bundled named
> > and with bind98 and bind99 packages. I pulled bind910 from pkgsrc HEAD and
> > it hangs as well. 
> 
> I suspect this is related to this change, which has been pulled up to
> netbsd-6 and netbsd-6-1, but which isn't part of the NetBSD 6.1.4
> release (will therefore be in the 6.2 and 6.1.5 releases, though,
> whenever they come out):
> 
> lib/libpthread/pthread_cond.c                   1.62
> lib/libpthread/pthread_mutex.c                  1.57, 1.59
> 
>         Partial fix for thread deadlock commonly observed with
>         named. Also address PR/44756.
>         [prlw1, ticket #1029]
> 
> You could apply those patches, and get a working threaded BIND; I did,
> and suffered this problem before and it's now gone for me.
>

I can and will apply these patches. Thanks!
 
> > I compiled 9.10 from source with threads disabled and it
> > doesn't seem to hang.  It's also noticably faster.
> 
> By what measurement factor, may I ask?  I have a similar install with
> the above patch installed, which is showing somewhat elevated response
> times to queries compared to other installations outside of our
> domain, and I've been wondering what the reason might be.
> 

Well, I haven't tested with the patch installed yet. To test on the same 
system — which is very much in production — so it can't happen today. 

What I used to test was I grepped and awked a list of all my non-GENERATED A 
records
and then appended the list to itself 2000 times. I then fed that list to 
queryperf.

Once the threading issues are resolved I'll send you the new benchmarks for 
comparison.

> Best regards,
> 
> - Håvard
> 

Thank you for your help. 

Harry Waddell