|
Real servers are running normally and configured "for a big website". :)
I'm telling you.. it seems like the loadbalancer is the bottleneck..
I'll do some perfirmance tests soon with and w/o a loadbalancer.
On 10/16/07, Joseph Mack NA3T <jmack@xxxxxxxx> wrote:
>
> On Tue, 16 Oct 2007, Janar Kartau wrote:
>
> > Hi there,
> > We recently moved several sites from their own dedicated servers to a
> shared
> > Apache cluster. All was fine till the last site was migrated...
> > Now, we see random lockups (loading of the page hangs for a while) which
> > last up to ~5 seconds.
>
> haven't a clue. Have you done all the right things for a big
> website? eg enough instances of httpd running?
>
> > Seems this only happens under heavier load... which
> > leads me to the question if it hits some limit or something?
>
> I have a link in the HOWTO to the squid performance webpage
> (which you can probably find directly with google) which
> tells you how to tweek apache.
>
> > I haven't encountered this problem if i browse the real
> > server directly. Loadbalancer is CentOS 5.0 and Apache
> > nodes (3 of them) are CentOS 4.5. All sites use LVS-NAT,
> > persistent connections (1800 secs) and firewall marks (for
> > ports 80 and 443).
>
> nothing to do with your problem, but you could try the -SH
> scheduler instead of persistence.
>
> > No errors in dmesg. Only errors i see right now are a LOT
> > of "ERR!" state connections in ip_vs_conn (all of which
> > are IP btw, TCP is fine).
>
> Horms (or Graeme) might know what they're about.
>
> Joe
>
> --
> Joseph Mack NA3T EME(B,D), FM05lw North Carolina
> jmack (at) wm7d (dot) net - azimuthal equidistant map
> generator at http://www.wm7d.net/azproj.shtml
> Homepage http://www.austintek.com/ It's GNU/Linux!
>
> _______________________________________________
> LinuxVirtualServer.org mailing list - lvs-users@xxxxxxxxxxxxxxxxxxxxxx
> Send requests to lvs-users-request@xxxxxxxxxxxxxxxxxxxxxx
> or go to http://lists.graemef.net/mailman/listinfo/lvs-users
>
|
