[Nut-upsdev] Multipathed monitoring of an UPS

[Baruch Even]

Hi,

Once you have multiple paths to the same device you need to have logic to
decide what source is more correct at any point or to decide per-piece what
to take from which input. One thing to note is that the ups drivers are not
reading data atomically and so one reader can have partially stale data and
you need to figure it out. Also, network latencies and packet loss may
cause additional confusion.

It is a good idea to have multipath to the ups but it needs to be done
carefully.

Baruch


On Mon, Feb 17, 2014 at 9:33 PM, Jim Klimov <jimklimov at cos.ru> wrote:

> Hello all,
>
>   Given that some UPSes and other devices manageable by NUT may have
> several means of connectivity simultaneously, such as USB/Serial and
> SNMP, it may be useful to utilize several connection methods to the
> same UPS.
>
>   Moreover, a couple of master servers with connection to the
> engineering VLAN (where an SNMP UPS is present) might provide
> equivalent "access" to the UPS data for the slave systems without
> access to that VLAN.
>
>   I believe the key idea here is to have a grouping of MONITOR
> rules (or perhaps a new similar keyword) that would state that
> a number of "ups at host login pass" tuples correspond to the same
> set of "N powersources fed". As long as at least one "data path"
> provides non-stale data, we consider that the UPS information
> is available for this set of powersources.
>
>   Does this idea sound sane and/or useful? ;)
>
> Thanks,
> //Jim Klimov
>
>
> _______________________________________________
> Nut-upsdev mailing list
> Nut-upsdev at lists.alioth.debian.org
> http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/nut-upsdev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/nut-upsdev/attachments/20140218/92ebd2ec/attachment.html>