[Spread-users] mod_log_spread: Multiple daemons versus single daemon

[Christian Robottom Reis]

Hi,

I was up to now using a single daemon and having all mod_log_spread
instances talk directly to that daemon (using the port at host syntax for
SpreadDaemon). I messaged George and he suggested I could shift to using a
set of daemons participating in a ring, which to him sounded saner.

I've done this now, and immediately I've noticed how high spread hits the
CPU on each of the servers. I'm not entirely sure if the load is up or
down from before, since our IP balancing solution balances the loads out,
but I am wondering if this is expected or now. Right now the daemon eats
up 30% CPU of a 500MHz machine, which isn't too little IMHO.

My question is: is using a single daemon more efficient in this case?
Since it's a very CPU-bound application, and the local network between
machines can take the traffic, shouldn't avoiding the multicasting be
better? In other words, are we stressing the local loop between server and
local daemon, without making things any lighter on the network at all?

I'd like to know the opinions of veteral m_l_s users, and if there are
tweaks I should make (beyond shifting MAX_SESSIONS_MESSAGE to 5000 which
I just did)?

Take care,
--
Christian Reis, Senior Engineer, Async Open Source, Brazil.
http://async.com.br/~kiko/ | [+55 16] 272 3330 | NMFL