[Pacemaker] crm_failcount would take some irrelevant values

[Andrew Beekhof]

but 0 is a valid value - and possibly the most likely one to be set
and that strcmp looks wrong to me


i'll apply a modified version of this, but i don't want to get too
hung up trying to detect every possible value a hostile admin could
throw at the cluster (especially when they can just as easily use
cibadmin to achieve the same result).

for 1.0 we'll be using relax-NG schemas (instead of DTDs) for
validation which will allow us to catch more things like this
centrally (on the server) and reject them (and defining what is a
valid 'score' is one thing I've already done :-).

On Fri, May 16, 2008 at 10:45 AM, Junko IKEDA <ikedaj at intellilink.co.jp> wrote:
> Hi,
>
> Using Pacemaker stable-0.6, the following operation works well.
>
> # crm_failcount -r <resource-id> -U <nodename> -v INFINITY
> fail-count would be increased to 10000.

should be higher than that, INFINITY is 1,000,000

>
> This one would fail,
> # crm_failcount -r <resource-id> -U <nodename> -v ERROR
>
> fail-count wouldn't be changed.
> But crm_failcount command updates CIB.
>
> cibadmin -Q sais;
> # cibadmin -Q | grep fail-count
> <nvpair id="status-db8f2da4-a7fb-40bf-bf14-befe4af11db7-fail-count-dummy"
> name="fail-count-dummy" value="ERROR"/>
>
> It would be convenient if crm_faicount can ignore some string value given by
> -v option other than "INFINITY".
> See atached.
>
> Best Regards,
> Junko Ikeda
>
> NTT DATA INTELLILINK CORPORATION
>
>
> _______________________________________________
> Pacemaker mailing list
> Pacemaker at clusterlabs.org
> http://list.clusterlabs.org/mailman/listinfo/pacemaker
>
>