|
"Stéphane Anet" wrote:
>
> Here is few information about my software configuration :
> I use RedHat 6.2
> I use Piranha-0.4.14-1
You need to grabe the latest piranha RPMs from
http://people.redhat.com/kbarrett/
> --------------------------------------------------------------------
> About Kernel I use the original kernel that was in the RedHat 6.2 CD.
Should be OK.
> question 1 :
> I read in some documentation that to install and configure Piranha, we have
> to have Redhat 6.1 or Redhat 6.2.
> About that in the documentation they speak about lvs patch.
>
> Do I need to apply this patch to my original kernel (redhat 6.2) ?
No. 6.2 includes the patch.
> I use one router, one real server and one computer to simulate a client
> (someone who wants to see a site html install on the real server )
>
> --------------------------------
> My configuration installation :
> --------------------------------
>
> |----------HUB-------------
> | |
> |eth0=192.168.40.21 |eth0=192.168.40.20
> |eth0:1=192.168.40.1 (vs1) | ---------|
> |---------- | |
> | active | | computer |
> | router | | Internet |
> | | ------------
> |----------
> |eth1=192.168.149.141
> |eth1:1=192.168.149.143 (NAT router)
> |
> HUB
> |
> |eth0=192.168.149.142
> |---------|
> | rs1 |
> |_________|
>
> ______________________________________________________________________
> At the level of my configuration (files) :
> ------------------------------------------
> In /etc/sysctl.conf I activate Defrag and Ip_forward.
> #Disables packet forwarding>
> net.ipv4.ip_forward = 1
> # Enables source route verification>
> net.ipv4.conf.all.rp_filter = 0
> # Disables automatic defragmentation (needed for masquerading, LVS)
> net.ipv4.ip_always_defrag = 1
> # Disables the magic-sysrq key
> kernel.sysrq = 0
>
> I have verify that these options are activated when the computer start .
>
> cat ip_forward (the result is : 1)
> cat ip_always_defrag (the result is : 2)
ip_always_defrag must be "1". "2" is incorrect.
Sounds like you made an error in an "echo >" statement to /proc
> --
> In /etc/rc.d/rc.local I add this line :
>
> ipchains -A forward -J MASQ 192.168.149.0/24 -d 0.0.0.0/0
> I have veify that option is activated when the computer start.
>
> ipchains -L
Good
> At the level of my real server I add the default gateway :
> the real server is a window OS.
> the gateway i configure is 192.168.149.143
>
> I do a test to see if it works.
> -> ping 192.168.40.21
> -> ping 192.168.40.20
>
> It's ok
>
> The passwd section for piranha is correct :> http://xxxxx/piranha
> - it ask me for a password , I type it, it is ok ( i type it before with the
> command piranha-passwd *****)
>
> - after i put the information concerning my cluster with the web interface,
> - all is ok . I can verify by editing the file /etc/lvs.cf
>
> - after I start the cluster, I type pulse
> the daemon pulse run after few second, 1 daemon nanny run and lvs daemon run
> as well.
>
> - When I look at the web interface on the control monitoring
> I can read at the level of current LVS routing table :
> - My Primary server IP : 192.168.40.21 with the correct port 80
>
> - The adress IP of my real server 192.168.149.142
>
> When I try to connect to my cluster puting http://192.168.40.21
> on the computer internet (IP : 192.168.40.20)
> - It doesn't work (after 5 or 6 minutes there is an indication :
> - "there is no response the server may be out of order"
I'm not sure using 192. addresses everywhere will work, since
you are trying to use NAT. Perhaps a more knowledgable LVS person
can respond to this.
I examine that defrag parameter also and make sure it's set to 1.
> - I did several test to understand the problem, my observation are :
> - at the level of the web interface i can see that there is a
> connexion when i try to reach one of the two real servers with the computer
> internet
>
> - at the level of real servers (windows) I can see that there
> is a connexion with the active router (with the log files).
If it helps, you can type "pulse -n -v" instead of using the startup script.
This will run pulse interactively and display debugging output. Make sure
you have the new RPMs installed.
A "tail -f /var/log/messages" is also useful.
--
Keith Barrett
Red Hat Inc. HA Team
kbarrett@xxxxxxxxxx
|
