[MOBY-dev] BioMoby central dead...
Paul Gordon
gordonp at ucalgary.ca
Thu Jun 12 17:11:22 UTC 2008
It brings a tear (of joy) to my eye that Moby Central is becoming so
essential, we need a backup. :-)
I think a European mirror is a great idea, because just having two
servers here in Calgary would not have prevented the down-time
experienced yesterday. We'd need to change the dynamic Central lookup
policy (which goes through biomoby.org) to change from returning one
server to returning a list of servers. Pretty simple, but still has a
single point of failure, now at biomoby.org. Dynamic Central lookup is
disabled by default unless a connection to Central fails, so maybe it's
not such a big deal (unless moby.ucalgary.ca AND biomoby.org in Boston
(or wherever) fail at the same time).
The other (robust) option is to have an LSID that points to a document
listing the servers. The resolution protocol allows for redundant data
serving in theory, but we'd need to set up moby as a "proper" authority
that goes in the lsid.org DNSPTR records.
-Paul
Heiko Schoof wrote:
> Hi,
> we have implemented synchronization of moby centrals. But it needs
> some testing under production conditions. Read-only should work fine.
> So if anyone is willing and able to host an alternative
> moby-central... we are in the process of setting one up, and could
> possibly act as European mirror. But I need to discuss this with
> Andreas first, and see in what time scale this is feasible.
> Has anyone else been experimenting with the synchronization code from
> Andreas?
>
> What would then be missing is the implementation for a fall-back in
> the client libraries, I think. In effect, the ability to give a second
> moby-central, which is used if connections to the primary fail.
>
> Best, Heiko
>
> On 12.06.2008, at 09:41, sneumann wrote:
>
>> On Mi, 2008-06-11 at 19:10 +0200, Andreas Groscurth wrote:
>>> someone killed it again ;-(
>>
>> Hi,
>>
>> wouldn't that be the time to call for redundancy for moby-central ?
>> Either some read-only secondary server which keeps
>> at least the services running if the primary is unreachable,
>>
>> or even a full-blown anycast a la Root-DNS servers
>> with synchronized backend-something for the registry database ?
>>
>> Or has that already been discussed and dismissed ?
>>
>> Yours,
>> Steffen
>>
>> --
>> IPB Halle AG Massenspektrometrie & Bioinformatik
>> Dr. Steffen Neumann http://www.IPB-Halle.DE
>> Weinberg 3 http://msbi.bic-gh.de
>> 06120 Halle Tel. +49 (0) 345 5582 - 1470
>> +49 (0) 345 5582 - 0
>> sneumann(at)IPB-Halle.DE Fax. +49 (0) 345 5582 - 1409
>>
>> _______________________________________________
>> MOBY-dev mailing list
>> MOBY-dev at lists.open-bio.org
>> http://lists.open-bio.org/mailman/listinfo/moby-dev
>
> _______________________________________________
> MOBY-dev mailing list
> MOBY-dev at lists.open-bio.org
> http://lists.open-bio.org/mailman/listinfo/moby-dev
>
> !DSPAM:60005,4850f8cc10950795412468!
>
>
>
More information about the MOBY-dev
mailing list