[Linux-HA] Learning the CIB

[vini.bill at gmail.com]

Ok, ok... Now I think I know where is my problem...

I'd like to know how should I specify my ip, nic and broadcast parameters
for my IPaddr primitive. Any tips?

Vinicius Menezes

On 1/10/07, vini.bill at gmail.com <vini.bill at gmail.com> wrote:
>
> Ok... I followed the tips you gave me and I'm still get some errors. Take
> a look at my /var/log/messages:
>
> Jan 10 11:59:25 DB01 heartbeat: [4788]: WARN: Late heartbeat: Node db01:
> interval 1000 ms
> Jan 10 11:59:25 DB01 heartbeat: [4788]: WARN: Late heartbeat: Node
> 192.168.0.1: interval 1000 ms
> Jan 10 11:59:25 DB01 heartbeat: [4788]: WARN: Late heartbeat: Node db02:
> interval 1000 ms
> Jan 10 11:59:25 DB01 crmd: [4867]: debug:
> init_client_ipc_comms_nodispatch: ipc.c Could not init comms on:
> /var/run/heartbeat/crm/cib_rw
> Jan 10 11:59:25 DB01 crmd: [4867]: debug: cib_native_signon:cib_native.c
> Connection to command channel failed
> Jan 10 11:59:25 DB01 crmd: [4867]: WARN: cib_native_signon:cib_native.c
> Connection to CIB failed: connection failed
> Jan 10 11:59:25 DB01 crmd: [4867]: debug: cib_native_signoff:cib_native.c
> Signing out of the CIB Service
> Jan 10 11:59:25 DB01 crmd: [4867]: debug: do_cib_control:cib.c Could not
> connect to the CIB service
> Jan 10 11:59:25 DB01 crmd: [4867]: WARN: do_cib_control: cib.c Couldn't
> complete CIB registration 25 times... pause and retry
> Jan 10 11:59:25 DB01 crmd: [4867]: debug: crm_timer_start:utils.c Started
> Wait Timer (I_NULL:500ms), src=30
> Jan 10 11:59:25 DB01 crmd: [4867]: debug: register_fsa_input_adv:
> messages.c Stalling the FSA pending further input: cause=C_FSA_INTERNAL
> Jan 10 11:59:25 DB01 crmd: [4867]: debug: s_crmd_fsa:fsa.c Exiting the
> FSA: queue=0, fsa_actions=0x180021000000006, stalled=true
>
> the cib I'm using is the one attached and the ha.cf is also attached. I'm
> completely lost now. All the cib.xml seem to be well done but I still
> can't get the virtual IP up.
>
> I'm investigating what's wrong here. Any ideas or suggestions are welcome.
>
>
> Vinicius Menezes
>
> On 1/10/07, John Hearns <john.hearns at streamline-computing.com> wrote:
> >
> > vini.bill at gmail.com wrote:
> > > Well It's a 10 Gigabit connection so... I tought I could try 100ms,
> > but
> > > anyway. I'll That tomorrow ;).
> >
> > You are confusing bandwidth and latency.
> > Sticking a wet finger in the air, you might get 80ms latency on a
> > standard gigabit connection when passing MPI traffic, using a good
> > switch.
> > Using specially tuned drivers, we get down to 20ms which allows us to
> > build some highly cost effective and efficient gigabit Beowulf clusters.
> >
> > Don't know about standard 10Gbps,
> > but if you're looking for low latency at 10Gbps Myrinet have a new line
> > of NICs
> > http://www.myricom.com/Myri-10G/overview/
> >
> > But in short, stick with the 1 second recommendation.
> >
> >
> >
> > --
> >       John Hearns
> >       Senior HPC Engineer
> >       Streamline Computing,
> >       The Innovation Centre, Warwick Technology Park,
> >       Gallows Hill, Warwick CV34 6UW
> >       Office: 01926 623130 Mobile: 07841 231235
> > _______________________________________________
> > Linux-HA mailing list
> > Linux-HA at lists.linux-ha.org
> > http://lists.linux-ha.org/mailman/listinfo/linux-ha
> > See also: http://linux-ha.org/ReportingProblems
> >
>
>
>
> --
> ... Vinicius Menezes ...
>



-- 
... Vinicius Menezes ...