Read down on my lines, I added a possible solution for your problema about
VIP, try next time, and tell us if this problem is solved.
And by other hand, The only one way for delete configuration on zen is with
a bad configuration cluster or manualy (but this isn't detected on logs), if
you configure your cluster on master node, but the configuration isn't
replicated to backup node, if the service switch to node backup, this new
node haven't the configuration, and this situation is replicate to node1, i,
all configuration on node backup (node1) will be the same that node2 ( now
master).
We are checking the logs and all is right, could you send us the
/var/log/messages file (2 nodes)?
Would be possible to know the test that your colleagues was running for try
on our development environment?
If you are studing your environment and you see something more that could
help for solve this, please, comment us.
We follow studing the situation, and we wait more information.
Ciao Señor Iglesia ;)
2011/9/3 Chiesa Stefano <[email protected]>
> Hello Emilio.
> Last week I wasn't at the office (last little week of vacation) but some
> collegues tested their balanced application with, they said, heavy traffic,
> I can tell you only this...
>
> They reported that after a while form the start of the test the website
> behind the VIP address didn't respond any more.
>
About this point, could happend:
backend overloaded.
farm overloaded, we detected if the farm use more than 1024 file
descriptors, or the same, 512 concurrent connections, the farm is stopped,
for solve this:
On /etc/sysctl.conf file would be insterting add this line:
fs.file-max = 100000
After add this lines you have to restart the web gui:
/etc/init.d/minihttpd stop
/etc/init.d/minihttpd start
It allows more than 1024 files open for all user, needed for tcp
connections, and this is a bug detected on rc4, that will be solved on rc5
>
> When I checked the master zen server status I noticed that all the farms
> disappeared from the web gui while the interface configuration was ok. This
> has been reported twice.
>
> The first time I restored the "last known good" configuration on ZEN-1 and
> this resolved the issue.
>
> The second time (today) I rebooted the zen server, the cluster perfomed the
> failover, the two real server started answering and also when the primary
> zen server returned master the balancing went well.
>
> This is all I can say. Attached a .rar file with the CONFIG and LOGS
> folders of the two server.
>
> I don't know what is notified to the backup sevrer during normal operation
> (each new connection? other?), but for the moment I will disable the cluster
> and I'll see what happen.
>
> Ciao.
> Stefano.
>
> Stefano Chiesa
> WOLTERS KLUWER ITALIA SRL
>
>
>
> ------------------------------------------------------------------------------
> Special Offer -- Download ArcSight Logger for FREE!
> Finally, a world-class log management solution at an even better
> price-free! And you'll get a free "Love Thy Logs" t-shirt when you
> download Logger. Secure your free ArcSight Logger TODAY!
> http://p.sf.net/sfu/arcsisghtdev2dev
> _______________________________________________
> 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]
------------------------------------------------------------------------------
Special Offer -- Download ArcSight Logger for FREE!
Finally, a world-class log management solution at an even better
price-free! And you'll get a free "Love Thy Logs" t-shirt when you
download Logger. Secure your free ArcSight Logger TODAY!
http://p.sf.net/sfu/arcsisghtdev2dev
_______________________________________________
Zenloadbalancer-support mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/zenloadbalancer-support
