[ILUG] (WAS desktoplinux..) now RHEL5 HA

[John Moylan]

I used CLVM in the labs for the RH Storage course last Sep. .. everyones
systems kept locking up and had to be rebooted. I believe the Kernel
changes were blamed by the trainer at the time.

We were also told that Lucli was not yet as mature as the X-Windows
app... And never to mix the two.

I'll be testing 5.1 HA with a view to deploying Q2, so's it's good to
hear things are working the way they should.

J


On Tue, 2008-02-19 at 21:36 +0000, Conor Wynne wrote:
> 
> On 19 Feb 2008, at 21:26, Conor Wynne wrote:
> 
> > On Feb 19, 2008 7:25 PM, John Moylan <john at nuatech.net> wrote:
> >         Hi Conor,
> >         
> >         Are you having any RHEL5 issues with CLVM? Maybe 5.1 has
> >         sorted these
> >         out, but 5.0 introduced some issues when that were not in
> >         4.6 and may
> >         have been a side effect of taking some of the cluster code
> >         out of the
> >         Kernel.
> 
> 
> Removing stuff from the kernel is a good idea. 
> What issues have you come across? 
> 
> 
> I don't remember the release we used in training, but I do remember
> getting frustrated as I couldn't figure out why my NFS wouldn't fail
> over. 
> For some reason, when using the GUI to manually failover, it just
> borked, but when I did it on the command line it worked. 
> 
> 
> Then when I re-tried from the luci, it magically worked again. Never
> figured out why as we were moving very quickly through labs. 
> It was for the RHCA storage course. 
> 
> 
> I have not experienced this on 5.1. 
> 
> > Have not come across any issues to speak of, although I initially
> > couldn't get the cluster IP going, and a respawn of the daemons
> > magically fixed that. Only happened the once though. I also had not
> > setup the virtual machine fencing correctly, and once sorted, no
> > more woes. I noticed logs from each node stated they vm's were
> > rebooted as the cluster was not quorate, but I hardly noticed it
> > happened at all. I suspect the system was running low on resources
> > as I had lots of apps running. 
> >  
> > This is what should happen, as the cluster thinks that the other one
> > is overloaded, or not receiving keep alives (whatever the term is),
> > so reboots em! 
> >  
> > The box (and vm's) were updated to 5.1 prior to doing the cluster
> > configs. I tend to patch first, then config. I find 5.x vastly
> > better thn any previous incarnation. Conga simply rulez. 
> > very impressive bit of software. I need more RAM before I add more
> > services though. I really should have boughthat lttle dell server
> > with 4GB. 
> > 
> > J
> > 
> > Conor. 
> 
> Conor.