LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: How is the best way to tell if my director is be over run

To: "'LinuxVirtualServer.org users mailing list.'" <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: How is the best way to tell if my director is be over run
From: Todd Lyons <tlyons@xxxxxxxxxx>
Date: Fri, 29 Apr 2005 09:34:43 -0700
Randy Paries wanted us to know:

>I have a LVS-DR setup with 1 director and 2 real servers
>I average about 2.5 million hits a day
>I have been noticing lately that when I go to may main URL (to the
>director), sometimes there is a noticeable delay. Before I go to the page.

Check hostname resolution.  Something somewhere is timing out either in
a forward lookup or a reverse lookup (the reverse lookup sounds more
likely by the description).  Make sure that apache is not logging
hostnames (requires a reverse lookup).

>CPU usage on all three is pretty low. But I would assume there is something
>else I should look at to see how the director is holding up. It is not a
>very powerful box

It doesn't have to be a powerful box.  This is from the past hour on my
box:
# iptables --list -n -v | grep INPUT
Chain INPUT (policy ACCEPT 6111K packets, 735M bytes)

6 million packets per hour, and we're not to our daily peak yet (around
noon our time).  This is running on the following old hardware (in
reality, it's way more than we need):

# egrep 'model.name|MHz' /proc/cpuinfo
model name      : Pentium III (Coppermine)
cpu MHz         : 696.618
model name      : Pentium III (Coppermine)
cpu MHz         : 696.618
# uptime
 09:23:21 up 52 days,  1:24,  1 user,  load average: 0.06, 0.03, 0.00
# vmstat 1
procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
 r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
 0  0      0 1619368 144300 126884    0    0     0     0    5     4  2  1 97  0
 0  0      0 1619368 144300 126884    0    0     0     0    0    58  0  3 97  0
 0  0      0 1619368 144300 126884    0    0     0     0    0    61  1  2 98  0
 0  0      0 1619368 144300 126884    0    0     0     0    0    59  0  2 97  0
 0  0      0 1619368 144300 126884    0    0     0     0    0   105  8  3 89  0
 0  0      0 1619304 144300 126884    0    0     0     0    0    55  1  3 97  0
 0  0      0 1619304 144300 126884    0    0     0     0    0    59  1  3 97  0
 0  0      0 1619304 144300 126884    0    0     0     0    0    63  0  3 97  0
 0  0      0 1619304 144300 126884    0    0     0     0    0    53  0  3 98  0
 0  0      0 1619304 144300 126884    0    0     0     0    0    99  8  2 90  0
 0  0      0 1619304 144300 126884    0    0     0     0    0    43  0  2 97  0
 1  0      0 1619304 144300 126884    0    0     0     0    0    48  1  2 98  0
 0  0      0 1619304 144300 126884    0    0     0     0    0    48  0  3 98  0
 0  0      0 1619304 144300 126884    0    0     0     0    0    49  0  2 98  0
 1  0      0 1619304 144300 126884    0    0     0     0    0    90  4  3 93  0
 0  0      0 1619304 144300 126884    0    0     0     0    0    64  4  4 93  0
 0  0      0 1619304 144300 126884    0    0     0     0    0    49  0  3 96  0

Every 4 seconds or so, the userspace %CPU usage jumps a bit, that's from
ldirectord polling the real servers.
-- 
Regards...              Todd
They that can give up essential liberty to obtain a little temporary 
safety deserve neither liberty nor safety.       --Benjamin Franklin
Linux kernel 2.6.11-6mdksmp   3 users,  load average: 0.07, 0.05, 0.06

<Prev in Thread] Current Thread [Next in Thread>