Running 6K hosts
ae at op5.se
Wed Feb 9 09:40:51 CET 2005
Michael Medin wrote:
> Would the CGIs really be faster?
> I implemented one of them as fastCGI a while ago to see if I could make
> it faster but no such luck. The "slowness" of the CGI (at least in my
> limited trials) was from reading / parsing the status data. And if this
> (as I read it) would decrease speed in the reading phase and increase it
> in the rendering phase I wonder if there would be any overall benefit?
That was a possibility for the nagios core, although not necessarily
true. The CGI's would get everything served on a platter, so while the
core does parse/count+index/malloc/tabulate in separate passes the CGI's
would malloc first and then do all the rest in a single pass and thus
save the time of calculating the initial hashes.
> As the CGIs do load/dostuff/close, load/dostuff/close... as opposed to
> Nagios that does load/dostuff/dostuff/dostuff...
Currently, the load phase consists of "parse, calculate hashes". If we
implement indexed tables it would be "parse, index" with the indexing
already done (for the most part) by the nagios daemon. Needless to say,
calculating hashes takes longer than indexing tables.
> // Michael Medin
> -----Original Message-----
> From: nagios-devel-admin at lists.sourceforge.net
> [mailto:nagios-devel-admin at lists.sourceforge.net] On Behalf Of Andreas
> Sent: den 8 februari 2005 23:31
> To: nagios-devel at lists.sourceforge.net
> Subject: Re: [Nagios-devel] Running 6K hosts
> Jason Martin wrote:
>>On Tue, Feb 08, 2005 at 09:16:09PM +0100, Andreas Ericsson wrote:
>>>host_id, which would reside in the host object struct (in running
>>>config, that is). The one-time load penalty would be slightly higher,
>>>but not necessarily very much since the internal tables really don't
>>>have to bother with alphabetical sorting and such.
>>Do you have any thoughts on how long a fluent C programmer would
>>need to implement that?
> A weeks worth of fulltime hacking, roughly. Perhaps a little more, and
> it would definitely require some serious testing with some seriously
> weird configurations so as to make sure it doesn't break anything.
> Ofcourse, CGI's would have to be hacked as well, but they would see the
> greatest speedups so that should only be fun.
Andreas Ericsson andreas.ericsson at op5.se
OP5 AB www.op5.se
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
More information about the Developers