Tor 0.1.2.2-alpha eventdns timeouts

Fabian Keil freebsd-listen at fabiankeil.de
Thu Oct 26 16:49:57 UTC 2006


Nick Mathewson <nickm at freehaven.net> wrote:

> On Mon, Oct 23, 2006 at 12:48:48PM +0200, Fabian Keil wrote:
>  [...]
> > Oct 23 12:09:06.074 [warn] eventdns: Nameserver 10.0.0.4 is back up
> > 
> > This always seems to happen if there are more than a few (<5)
> > DNS requests per second. Am I the only one seeing this?
> 
> No, this seems to be a (relatively harmless) bug with our eventdns
> code.  I'm hoping to have it fixed soon; please let us know if it
> appears in later versions.
> 
> Also, you can check to see whether bugs are already reported by
> checking our bugtracker at:
> 
>    http://bugs.noreply.org/flyspray/index.php?tasks=all&project=4

I'll keep that in mind, thanks.

> This is bug 326:
> 
>    http://bugs.noreply.org/flyspray/index.php?do=details&id=326

Is "timeout threshold is set too low, and the threshold for declaring
a server down is set too low" still the current analysis?

I increased global_max_retransmits and global_max_nameserver_timeout
from 3 to 5 and changed global_timeout from {3, 0} to {20, 0}.

As eventdns used to complaint from time to time about
"Bad response 2 (server failed)" and I didn't find a threshold
for these cases, I also added {0, 0} at the beginning of
global_nameserver_timeouts[], to make sure the server wouldn't
be ignored for ten seconds.

As a result I see indeed a lot less timeout warnings,
and the servers are declared active right away, but it
also looks as if the number of timeouts isn't reset
correctly:

Oct 26 18:00:26.467 [warn] eventdns: Nameserver 81.169.148.34 has failed: request timed out.
Oct 26 18:00:26.469 [warn] eventdns: Nameserver 81.169.148.34 is back up
[ten DNS requests successful]
Oct 26 18:00:35.962 [warn] eventdns: Nameserver 81.169.148.34 has failed: request timed out.
Oct 26 18:00:36.017 [warn] eventdns: Nameserver 81.169.148.34 is back up
[thirteen DNS requests succesful]
Oct 26 18:00:55.992 [warn] eventdns: Nameserver 81.169.148.34 has failed: request timed out.
Oct 26 18:00:56.008 [warn] eventdns: Nameserver 81.169.148.34 is back up

I assumed that there should be at least global_timeout seconds
between the last successful DNS requests, and a DNS server being
marked as down because of timeouts?

Fabian
-- 
http://www.fabiankeil.de/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
URL: <http://lists.torproject.org/pipermail/tor-talk/attachments/20061026/09d81248/attachment.pgp>


More information about the tor-talk mailing list