Si, pero "el ojo tambien quiere su parte"  ;))

Thanks for the explanation about failover time. Ten seconds are ok for me I was 
just wondering (as an ip address moves from a mac to another) if the failover 
process sends gratuitous arp to speed up the discovery of the new mac...

Ciao.
stefano.

________________________________

Da: Emilio Campos [mailto:[email protected]] 
Inviato: giovedì 18 agosto 2011 21.47
A: [email protected]
Oggetto: Re: [Zenloadbalancer-support] Habemus Cluster!


Hi, Stefano,  first of all thanks about give your time for test the zen cluster.

And yes, It could be possible test all interfaces and not one, could be 
implemented on the future  with no problem.

In spain, we say about food : "No es bueno llenar el ojo antes que la barriga! 
" lol lol

And about the 10 second: Zen cluster checks the node status every 2 second, if 
a change is detected, you have to add time for configure the interfaces and 
time for run the farms, secuently,

2 seconds + seconds for configure networking + seconds for configure farms : I 
worked with clusters which you can expect more time, I accept  suggestions for 
down the time.

Thanks and regards!


2011/8/18 Chiesa Stefano <[email protected]>


        Hi Laura and Emilio.
        
        This morning, after a brief struggle with zen2 interface configuration, 
I managed to set up the cluster in the "production" environment.
        As I said before I had to substitute the cross cable on cluster 
interface with a patch connected to a switch (in a private vlan) otherwise the 
RSA communication wouldn't work.
        
        I tried few times the failover process and everithing went good. I 
noticed that the time needed to see the website up again is about 10 seconds.
        Is it a time needed by the sync process or an ARP matter? Does ZEN send 
gratuitous arp during the failover process?
        
        In Italy we say: "el apetito viene comiendo". So another request about 
cluster:
        * would be possible in the future have ZEN checking not only the 
cluster interface to decide a failover?
        
        Anyway, enjoy your vacation, I'll let you know about our stress tests 
in september.
        
        For the moment thanks a lot for your work.
        
        Ciao.
        Stefano.
        
        ________________________________
        
        Da: Emilio Campos [mailto:[email protected]]
        Inviato: venerdì 12 agosto 2011 19.52
        A: [email protected]
        Oggetto: Re: [Zenloadbalancer-support] R: (no subject)
        
        
        Hi Stefano, Thanks about your test, and still not on vacation, soon, 
the end of this next week!!!! lol
        
        We are working for solve the issues reported by you and Dave, will be 
added on the new release, with improvements like a new concept: "profiles", for 
advanced configuration with http and https protocols.
        
        And about the backup commented on your mail, you're right, Im going to 
modify the process of "backup import", adding a "return to fabric status" 
before import the configuration.
        
        For all members of the list; thanks about your tests and suggestions.
        
        
        2011/8/11 Chiesa Stefano <[email protected]>
        
        
                       Hello again.
        
               I tried the failover and it worked good. I made zen servers 
change their role many times and they never messed anything up.
        
                               So I learnt/think:
               * the MAC mess is just a display problem (and only on the 
passive device)
               * never use a cross cable on cluster interface
        
                       It seems rc4 it's a big jump forward...
        
                       Now I'll try the upgrade on the two production server 
and then the cluster configuration.
        
               In relation to the restore procedure it seems that it simply 
copies  the previously backed up files but not restores "a situation".
               To explain: if I made a backup before a cluster config, then I 
configure the cluster and later restore the previous backup the cluster 
configuration remain configured.
               I'd find more useful if the files missing before the backup (and 
created later) were deleted by the restore program.
        
               Matter for the next rc's
        
               Anyway, go on holyday!
        
               Ciao.
               Stefano.
        
        ________________________________
        
               Da: Chiesa Stefano [mailto:[email protected]]
               Inviato: giovedì 11 agosto 2011 15.42
               A: [email protected]
               Oggetto: [Zenloadbalancer-support] (no subject)
        
        
        
               Hello Laura and Emilio, still not on vacation?
               I'm just came back and as promised I put up the test lab.
        
               I built two zen server with a simple configuration:
        
               * one "farm" interface
               * one gui if
               * one cluster if  (cross cable)
        
               And then I tried to create the cluster. As before I didn't 
manage to create it. It generated always an error during the RSA connection 
creation.
               Reading Dave posts I decided to change the connection type on 
the cluster if, I removed the cross cable and connected the two if through a 
switch.
        
               And it worked.
        
               Now this is the cluster page on ACTIVE host ZENLAB-1:
        
               Zen latency is UP on zenlab-1 172.16.99.1 | Zen latency is UP on 
zenlab-2 172.16.99.2
               Cluster IP 172.16.99.3 is active on zenlab-1
               Zen Inotify is running on zenlab-1
               Global status: (ok)
        
        
        
               But on ZENLAB-2:
        
               Zen latency is UP on zenlab-2 172.16.99.2 | Zen latency is DOWN 
on zenlab-1 172.16.99.1  
<https://10.39.160.242:444/img/icons/small/exclamation.png>
               Host key verification failed.  
<https://10.39.160.242:444/img/icons/small/exclamation.png>
               Zen Inotify is running on Host key verification failed.  
<https://10.39.160.242:444/img/icons/small/exclamation.png>
               Global status:  
<https://10.39.160.242:444/img/icons/small/exclamation.png>
        
               And, as Dave, I have a strange display of MAC addresses on GUI 
on ZENLAB-2 :
        
        
        
        
               Settings::Interfaces
        
               Table Interfaces
               Name     Addr    HWaddr  Netmask         Gateway         Status  
Actions
                       eth0     1.1.1.241       00:50:da:81:db:d7       
255.255.255.0   -       down     
<https://10.39.160.242:444/index.cgi?id=3-2&action=upif&if=eth0>   
<https://10.39.160.242:444/index.cgi?id=3-2&action=editif&if=eth0&toif=eth0&source=files&status=down>
   <https://10.39.160.242:444/index.cgi?id=3-2&action=addvip&toif=eth0>   
<https://10.39.160.242:444/index.cgi?id=3-2&action=addvlan&toif=eth0>   
<https://10.39.160.242:444/index.cgi?id=3-2&action=deleteif&if=eth0>
               eth1   
<https://10.39.160.242:444/img/icons/small/application_home.png>         
10.39.160.242    00:0f:fe:13:f7:95       255.255.224.0   -       up       
<https://10.39.160.242:444/index.cgi?id=3-2&action=addvip&toif=eth1>   
<https://10.39.160.242:444/index.cgi?id=3-2&action=addvlan&toif=eth1>       
<https://10.39.160.242:444/img/icons/small/lock.png>
               eth2   <https://10.39.160.242:444/img/icons/small/databases.png> 
       172.16.99.2      00:10:5a:be:f5:69       255.255.255.248         -       
up       <https://10.39.160.242:444/index.cgi?id=3-2&action=addvip&toif=eth2>   
<https://10.39.160.242:444/index.cgi?id=3-2&action=addvlan&toif=eth2>       
<https://10.39.160.242:444/img/icons/small/lock.png>
        
               eth2:cl   
<https://10.39.160.242:444/img/icons/small/databases.png>     172.16.99.3       
      00:10:5a:be:f5:69        255.255.255.248                down          
<https://10.39.160.242:444/img/icons/small/lock.png>
        
               eth0.1   1.1.1.100              00:10:5a:be:f5:69        
255.255.255.0          down      
<https://10.39.160.242:444/index.cgi?id=3-2&action=upif&if=eth0.1>   
<https://10.39.160.242:444/index.cgi?id=3-2&action=editif&if=eth0.1&toif=eth0.1&source=files&status=down>
   <https://10.39.160.242:444/index.cgi?id=3-2&action=addvip&toif=eth0.1>   
<https://10.39.160.242:444/index.cgi?id=3-2&action=deleteif&if=eth0.1>
        
               Now (as maybe it's just a GUI problem) I'll try the failover and 
I'll let you know.
        
               Ciao.
        
               Stefano.
        
        
        
               ----------------------------------------
               Stefano Chiesa
               Wolters Kluwer Italia
               Strada 1, Palazzo F6
               20090 Milanofiori Assago (Mi) - Italia
               Phone +39 0282476279 (20279 Voip)
               Fax +39 0282476633
        
        
        
               
------------------------------------------------------------------------------
               Get a FREE DOWNLOAD! and learn more about uberSVN rich system,
               user administration capabilities and model configuration. Take
               the hassle out of deploying and managing Subversion and the
               tools developers use with it.
               http://p.sf.net/sfu/wandisco-dev2dev
               _______________________________________________
               Zenloadbalancer-support mailing list
               [email protected]
               
https://lists.sourceforge.net/lists/listinfo/zenloadbalancer-support
        
        
        
        
        
        
        --
        Load balancer distribution - Open Source Project
        http://zenloadbalancer.sourceforge.net
        Distribution list (subscribe): 
[email protected]
        
        
        
        
------------------------------------------------------------------------------
        Get a FREE DOWNLOAD! and learn more about uberSVN rich system,
        user administration capabilities and model configuration. Take
        the hassle out of deploying and managing Subversion and the
        tools developers use with it. http://p.sf.net/sfu/wandisco-d2d-2
        _______________________________________________
        Zenloadbalancer-support mailing list
        [email protected]
        https://lists.sourceforge.net/lists/listinfo/zenloadbalancer-support
        




-- 
Load balancer distribution - Open Source Project
http://zenloadbalancer.sourceforge.net
Distribution list (subscribe): [email protected]



------------------------------------------------------------------------------
Get a FREE DOWNLOAD! and learn more about uberSVN rich system, 
user administration capabilities and model configuration. Take 
the hassle out of deploying and managing Subversion and the 
tools developers use with it. http://p.sf.net/sfu/wandisco-d2d-2
_______________________________________________
Zenloadbalancer-support mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/zenloadbalancer-support

Reply via email to