[OpenAFS] Re: Fallback Time with Multiple RO Sites
Ken Elkabany
Ken@Elkabany.com
Wed, 18 Apr 2012 16:15:34 -0700
--f46d043be1d0b2c03f04bdfc3c25
Content-Type: text/plain; charset=ISO-8859-1
Yes, you're correct. Upon retrying it, it took 56.7 seconds. Thanks for the
explanation. We'll try out the sysctl, and "fs setserverprefs".
On Wed, Apr 18, 2012 at 4:04 PM, Andrew Deason <adeason@sinenomine.net>wrote:
> On Wed, 18 Apr 2012 15:45:58 -0700
> Ken Elkabany <Ken@Elkabany.com> wrote:
>
> > But, when running "ls" from within the mount point, the client hangs
> > for about 20 to 30 seconds, and then finally brings up the contents.
> > I'm assuming the 20-30 second hang is the timeout time for the first
> > server. Is there a way to decrease the timeout allowance?
>
> I would've expected closer to 50 seconds, but it's possible some other
> operation started trying to contact the server before that. The default
> 'dead' timeout (meaning, we get no response from the other end of any
> kind) is 50 seconds. You can sort of change this on Linux by altering
> the afs.rx_deadtime sysctl, but I'm not sure how often that's used / how
> well it is honored (you might need to set that before afsd starts).
>
> In addition, we may be able to detect downed peers more quickly if we
> processed ip errors from the Linux kernel. But we don't do that
> currently.
>
> --
> Andrew Deason
> adeason@sinenomine.net
>
> _______________________________________________
> OpenAFS-info mailing list
> OpenAFS-info@openafs.org
> https://lists.openafs.org/mailman/listinfo/openafs-info
>
--f46d043be1d0b2c03f04bdfc3c25
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Yes, you're correct. Upon retrying it, it took 56.7 seconds.=A0Thanks f=
or the explanation.=A0We'll try out the sysctl, and "fs setserverp=
refs".<br><br><div class=3D"gmail_quote">On Wed, Apr 18, 2012 at 4:04 =
PM, Andrew Deason <span dir=3D"ltr"><<a href=3D"mailto:adeason@sinenomin=
e.net">adeason@sinenomine.net</a>></span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div class=3D"im">On Wed, 18 Apr 2012 15:45:=
58 -0700<br>
Ken Elkabany <Ken@Elkabany.com> wrote:<br>
<br>
> But, when running "ls" from within the mount point, the clie=
nt hangs<br>
> for about 20 to 30 seconds, and then finally brings up the contents.<b=
r>
> I'm assuming the 20-30 second hang is the timeout time for the fir=
st<br>
> server. Is there a way to decrease the timeout allowance?<br>
<br>
</div>I would've expected closer to 50 seconds, but it's possible s=
ome other<br>
operation started trying to contact the server before that. The default<br>
'dead' timeout (meaning, we get no response from the other end of a=
ny<br>
kind) is 50 seconds. You can sort of change this on Linux by altering<br>
the afs.rx_deadtime sysctl, but I'm not sure how often that's used =
/ how<br>
well it is honored (you might need to set that before afsd starts).<br>
<br>
In addition, we may be able to detect downed peers more quickly if we<br>
processed ip errors from the Linux kernel. But we don't do that<br>
currently.<br>
<span class=3D"HOEnZb"><font color=3D"#888888"><br>
--<br>
Andrew Deason<br>
<a href=3D"mailto:adeason@sinenomine.net">adeason@sinenomine.net</a><br>
<br>
_______________________________________________<br>
OpenAFS-info mailing list<br>
<a href=3D"mailto:OpenAFS-info@openafs.org">OpenAFS-info@openafs.org</a><br=
>
<a href=3D"https://lists.openafs.org/mailman/listinfo/openafs-info" target=
=3D"_blank">https://lists.openafs.org/mailman/listinfo/openafs-info</a><br>
</font></span></blockquote></div><br>
--f46d043be1d0b2c03f04bdfc3c25--