So, what happened:
1. Apparently, the migration of OpenLayers from my personal everydns acocunt
(where I moved it after the MetaCarta switch) to OSGeo didn't actually
happen like I thought it had.
2. EveryDNS has been planning migrations to DynDNS, and apparently starting
today, was forcing them by intentional planned downtime (that I wasn't told
about).
3. At 11am, it went down because of this.
At 11:05 I got told it was down (I was in meetings since then, which is why I
haven't
mentioned anything since then). I went through Dyns' DNS migration plan, but
that
doesn't actually work instantly, it takes a couple hours to take effect. At the
same time, I edited the PairNIC DNS information to OL to point to the new
Dyn servers.
So:
1. It shouldn't break again tomorrow during the next 'planned migration
outage'.
2. We should get it out of my personal acocunt as soon as possible: I'll send
an
additional email later today about the hosts we need to configure.
3. I am sorry that this happened, because I Thought we had done this migratino
before.
The biggest problem with doing the migration is, I believe, that when you flip
the
switch on pairnic from "host remotely" (what is is now) to 'host locally', it
uses
the 'non-configured' state for all of the hostnames for the next 4 hours, which
means that OL.org would essentially be down for those for hours (potentially),
as well as all of the sub-sites of OL.org.
At this point, that cost is lower than getting screwed over continuously because
I'm a terrible sysadmin, so we should do our best to make this change as soon
as possible while inconveniencing as few as possible.
Again, my apologies: I really thought we had migrated all of this stuff out of
my
personal accounts, but apparently not.
And sorry as well for not documenting this until now; I got a frantic IM to fix
it,
and did what I could in the meantime, but just now got the chance to write it
up.
-- Chris
On Aug 30, 2011, at 1:46 PM, ext Alex Mandel wrote:
> Seems to be working again(on it's own), for now.
> Looking at the everydns webpage it's appears that some action will be
> necessary on our part though. There's a notice about them switching all
> their services over to a different host (dyndns which own everydns) and
> that it requires us to migrate the account.
> http://www.everydns.com/
>
> The deadline is in 13 hours (before the start of Aug 31st). Did anyone
> get a notice about this?
>
> So urgency has dropped some, but we should come up with a plan in the
> event that Chris isn't reachable in the next 12 hours.
>
> Thanks,
> Alex
>
>
> On 08/30/2011 10:20 AM, Howard Butler wrote:
>> Alex,
>>
>> It appears that openlayers.{com|net|org} is available to manage via OSGeo's
>> pairNIC account, but I am hesitant to just flip the DNS management over to
>> pairNIC. I'm sure there are a bunch of aliases and this DNS doesn't allow
>> transfers to allow us to see what all is there.
>>
>> Chris?
>>
>> Howard
>>
>> On Aug 30, 2011, at 11:33 AM, Alex Mandel wrote:
>>
>>> Something seems to be amiss with openlayers.org DNS right now.
>>>
>>> It appears to be pointing to:
>>> 216.146.38.113
>>> lock-everydns.dyndns.com
>>>
>>> As noted on IRC
>>> message from http://www.everydns.net: "You must migrate your
>>> EveryDNS.net account to our rock-solid DynDNS.com system before August
>>> 31, 2011, or your site will no longer be accessible!"
>>>
>>> Reported as early as 7:30 PST (so 2 hours ago).
>>>
>>> Not sure who has the keys or where to go to fix this - or even what the
>>> problem is.
>>>
>>> Real location should be projects.osgeo.osuosl.org 140.211.15.75 afaik
>>>
>>> Thanks,
>>> Alex
>>> _______________________________________________
>>> Sac mailing list
>>> s...@lists.osgeo.org
>>> http://lists.osgeo.org/mailman/listinfo/sac
>>
>
> _______________________________________________
> Sac mailing list
> s...@lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/sac
_______________________________________________
Dev mailing list
d...@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/openlayers-dev
