Hello,
Yes, it still doesn't explain the overloading limits - multiplying the
weight by 2 seems very arbitrary, although in itself there is no real
problem (as far as I can tell) with limiting the connections like that so
long as it's documented.
The problems are that activeconns don't define connections (I'm actually
just these days writing an essay on L4 session counters for LVS) as they
are seen by a RS and the implementation significantly differs between
2.4.x and 2.6.x kernels as far as I can tell.
Also is_overloaded() should actually be reserved for a different
purpuse, namely of the threshold limitation feature. I'll skim the LVS
mailinglist archive for further references regarding this issue.
Comment: Public key available at http://linux.navaho.co.uk/pubkey.steve.txt
<offtopic sidenote>
I should like to inform you of the impracticality of trying to fetch
your key at the above URL. You seem to have configured a slightely
overcautious .htpasswd or .htaccess file in your DocumentRoot. All I get
is an 'internal server error':
An error occured whilst trying to generate the original error page.
The orginal error was:
(401) Authentication Required
WWW-Authenticate: Basic realm="Navaho Web Server access on upgrade.navaho"
Secondary error occured whilst trying to parse `auth'
Please inform the system administrator of this error.
</offtopic>
Take care,
Roberto Nibali, ratz
--
echo
'[q]sa[ln0=aln256%Pln256/snlbx]sb3135071790101768542287578439snlbxq' | dc
|