[DGD] Uninterruptible Power Supply

[bart at wotf.org]

Indeed, but there is a huge difference between a planned reboot and an
unplanned one.

When aiming for a persistent mud like both Shentino and me are doing, the
'uptime' of the mud is independent of the uptime of your machine, as you can
continue from a statedump, but that does require being able to make that
statedump before rebooting the machine. Planned reboots are no problem for
that, unplanned reboots however can be quite a problem, resulting in restoring
the mud to an older state then it was in when the reboot happened (and hence,
causing losses to players which were online at that time)

So while you are right about long uptimes of a system, this actually doesn't
apply as such to the uptime of a dgd based persistent mud.

aidil at Way of the Force: [43] /cmd/arch> uptime -v
It is 16:58:01, total virtual uptime 8 years 103d10:57, 2 players.
The system was first started 8 years 128d18:27 ago.
The last restore was on dec 21 21:35.
Memory: 291377392 bytes used, 298074216 bytes allocated.

So that is 8 1/3 years 'virtual' uptime for the mud, as you can see, I lost
almost 25 days of uptime over those 8 1/3 years due to things like having to
restore from a slightly older statedump or downtime of the server, or the mud
not running due to being busy with an upgrade of the hardware or operating system.

Bart.

On Thu, 22 Dec 2016 10:33:30 -0500, Littlefield, Tyler wrote
> Long uptimes are something that people gloat and cheer about for all 
> the wrong reasons. a UPS isn't to allow you to stay up (since if 
> your power is down, chances are internet will be too), but to allow 
> you to properly save data before you drop it on the floor. A few 
> years uptime is generally not a great idea since it means kernels 
> haven't been updated, you haven't updated main libs which some 
> services run on (and it's really realy hard to restart everything 
> once you update and know for sure they're all restarted), etc. I 
> take every server down once every 2 weeks or so, unless there's a 
> big security update. BSD has hell with OpenSSL. Usually it's just a 
> quick reboot and there we are again, up and running. On 12/22/2016 
> 1:06 AM, Raymond Jennings wrote:
> > I just got my desktop (and thus the server processes hosted on it) hooked
> > up to a UPS...and I now have the great rich feeling of being immune to
> > brownouts.
> > 
> > Do any of you guys use a UPS to keep uptime going?
> > 
> > Both commercial and hobbyist usages count.
> > ____________________________________________
> > https://mail.dworkin.nl/mailman/listinfo/dgd
> >
> 
> -- 
> Take care,
> Ty
> Twitter: @sorressean
> Web: https://tysdomain.com
> Pubkey: https://tysdomain.com/files/pubkey.asc
> ____________________________________________
> https://mail.dworkin.nl/mailman/listinfo/dgd


--
http://www.flickr.com/photos/mrobjective/
http://www.om-d.org/