This sounds like a great plan, except a word of caution on what we need to do for a few “other things”. We can of course move first and do some of those other things later too. So to be more concrete:
Migrate hackage, www, and lambda, yes! Migrate community — well yes, but… migrate all the logins too? If not, there’s really no point. On the plus side, all the software community needs is really user and project directories, and git and darcs, really. Kill trac.h.o — absolutely! However, we really do need to take that last poll and give warning if anyone actually still is using it. Kill rt.h.o in favor of phab — absolutely! But it still is used as part of the community.haskell.org process, and tickets are still created on it to manage those accounts. So we can always redirect [email protected] emails to create phab tickets, etc, but we must make sure that we set all the bits up to do so. www -> new-www. Yep, we still need more hands on deck to finish the site and polish it. I put out a call at the HIW for help with this, but in general it will take some collective effort to get it the “second 90%” there. Also, don’t forget: planet.haskell.org — served off community using Venus I believe. Also don’t forget that many www.haskell.org subdirectories continue to correspond to plain old websites and not wiki pages. /ghc /platform /hoogle /cabal /haddock for example. Cheers, Gershom On September 8, 2014 at 1:43:29 AM, Austin Seipp ([email protected]) wrote: > Hi *, > > As I'm sure some of you are aware, Haskell.org recently had > reliability problems with its virtualization host, an EX40 machine > located on Hetzner. > > We dealt with it (for the most part), but a part of this was actually > migrating servers to a new provider to alleviate I/O load on the host > system. > > As I mentioned last year, we've gotten support from Rackspace for > Haskell.org. In particular, this weekend we moved ghc.haskell.org to > Rackspace. This comes with nice amounts of dedicated bandwith and more > resources that should last a long time. > > I plan on migrating Hackage to a new dedicated server this week. Me > and John hope to migrate www to Rackspace by the end of this week, > hopefully. These are the 3 servers located on the EX40 server, Rock. > > Furthermore, we have an older server, the one running > lambda.haskell.org and community.haskell.org, running with Hetzner > too. I think we can migrate these quickly as well. > > This means that Rock, the EX40 server, will now be effectively unused. > And we will move the old servers too. So both of these can be shut > down, and we don't have to pay for them! That's a lot of money saved! > > We'll also be doing some other things: > > - We'll be splitting the mailing system to a separate server, mail.haskell.org > > - We'll be moving the DNS nameservers, as I said in my last email. > - I'll probably clean up the DNS entries, and we'll be able to add > AAAA records for all the servers now. > > - We'll be shutting down several services, like > - trac.haskell.org (unmaintained, spammed, mostly unused). See the > discussion last month about this. > - rt.haskell.org (we're using and suggest > https://phabricator.haskell.org for new issues) > > - The wiki will move to a new server, wiki.haskell.org, and will > still serve as the homepage for a little bit. > > - Then we'll deploy http://new-www.haskell.org as www.haskell.org, > and redirect things to wiki.haskell.org where necessary. > > - We'll move the downloads for the Haskell Platform to a new server > so Mark can upload them there. > > So, what does everyone think about this? Is there anything I'm > forgetting? We can hopefully get this done in the next few weeks if > the admin team just pitches in a few hours here and there. > > -- > Regards, > Austin - PGP: 4096R/0x91384671 > _______________________________________________ haskell-infrastructure mailing list [email protected] http://community.galois.com/mailman/listinfo/haskell-infrastructure
