On Thu, Aug 18, 2011 at 11:46:22AM -0600, Kevin Fenzi wrote: > On Thu, 18 Aug 2011 00:17:00 +0530 > > I think we can start looking at staging now. ;) > > Here's what I would suggest (and feedback especially from the other > application developers very welcome): > > * Add it into proxy01.stg as https://admin.fedoraproject.org/ask/ > (so we can share the cookie. Or do we want to do that anymore?) > I think we decided this wasn't a huge loss and the benefits of going without it were better domain name for end users and getting to use the upstream openid auth plugin instead of maintaining our own fas auth plugin. [snip plan I generally agree with] > > I'm a bit unsure if we want the db on the ask01 instance itself, or > using a shared db backend. On the one hand thats less things in one > machine and we can reboot/restart ask01 when we might not be able to do > so to the db backend machine. But it's also another machine to back up > and manage databases on, and if we get replication working another > place we would need to replicate. > If we think we're going to load balance and/or want higher availability, then we need separate machines. The db is always critical to keeping an application up. app servers for well designed apps are much less critical if we have multiple ones. We'd need to setup db backup/replication/etc whether its on ask01 or on a new db server... so really those points really seem to ask whether we should host the db on db01/db02 or make a new db server for it. -Toshio
Attachment:
pgpW2Ac4Sz9Kl.pgp
Description: PGP signature
_______________________________________________ infrastructure mailing list infrastructure@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/infrastructure
