On 12/19/2010 7:45 PM, Andrew Bennetts wrote:
> What Martin said. My first guess is that Conch is almost certainly
> doing all the crypto in the same thread that runs the event loop, so if
> that takes non-negligible time then it will delay processing of the next
> connection. Some profiling of that code will probably find the culprit
> fairly easily.
>
> I can rule out one possibility: I'm fairly sure that the XML-RPC call to
> the Launchpad database for SSH keys for that username is not blocking in
> that code; at least it wasn't when I wrote it :)
>
How do you profile the twisted server, though?
John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On 12/19/2010 7:45 PM, Andrew Bennetts wrote:
> What Martin said. My first guess is that Conch is almost certainly
> doing all the crypto in the same thread that runs the event loop, so if
> that takes non-negligible time then it will delay processing of the next
> connection. Some profiling of that code will probably find the culprit
> fairly easily.
>
> I can rule out one possibility: I'm fairly sure that the XML-RPC call to
> the Launchpad database for SSH keys for that username is not blocking in
> that code; at least it wasn't when I wrote it :)
>
How do you profile the twisted server, though?
John
=:->
-----BEGIN PGP SIGNATURE----- enigmail. mozdev. org/
iQZEACgkQJdeBCY SNAAMvggCgw9iob knbf9tHpKe/ aHgzaQRN 6lHRb9ccMMwkXgT SulEEgJoyN
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://
iEYEARECAAYFAk0
388AnR/
=lGs1
-----END PGP SIGNATURE-----