On Thursday 04 June 2009 19:01:47 Matthew Toseland wrote:
> On Thursday 04 June 2009 15:51:31 Matthew Toseland wrote:
> > Essentially the problems are:
> >
> > 1. Hosting the static web site and files. This includes web hosting,
> > hosting big files, fixed redirects over SSL, etc. It may involve moving
> > from https://checksums to somewhere else, but it will still have to include
> > SSL.
> >
> > This can be solved fairly easily with Google Apps and Google Web Apps and
> > Google Release System, or can be solved with a very cheap paid hosting
> > service.
> >
> > 2. Hosting mailing lists.
> >
> > Most likely this will be solved by using berlios. Hopefully they provide
> > searchable archives, but arguably that isn't our problem as many third
> > parties do this already.
> >
> > 3. The bug tracker.
> >
> > This is a big one. MANTIS has some serious disadvantages - it is a pain to
> > maintain, it is written in php and therefore has security issues. And other
> > bug trackers might very well be better at helping developers to get their
> > job done. Mantis can be hosted (by sourceforge or by godaddy), but it is
> > not clear whether it will be possible to import existing data. Or it can be
> > manually maintained on any paid-for php+database web hosting service, which
> > we don't want to do if we can avoid it.
> >
> > IMHO it is important that we keep the existing data. On the other hand I
> > might support keeping it in some static form and using a new bug tracker,
> > if a new bug tracker was demonstrably better.
> >
> > What bug tracker should we use? Trac? Jira? Lighthouse? Something else?
> >
> > 4. The wikis.
> >
> > We have a French wiki using MediaWiki and an English wiki using Wikka. It
> > is highly unlikely we will find anywhere, free or otherwise, that hosts
> > managed Wikka, so we will probably have to convert it to MediaWiki - or
> > start again. Again, IMHO it is important to keep existing content as not
> > doing so results in significant new work having to be done.
> >
> > We need to decide on #3 and #4, IMHO these are the real blockers. #1 can be
> > solved, but I don't think it makes sense to try to solve it until we have
> > decided what we are going to do about #3 and #4 because it will influence
> > the architecture of the solution to #1.
> >
> > I would also like to point out that we've spent a lot of time on this
> > already!
> >
> Another hosting suggestion (this is interesting because it solves 1, 2, and
> 4; it was recommended by several people on #freenet):
>
> Dreamhost VPS. This is shared hosting in that we don't have root and they
> keep all the stuff up to date, but with a vserver, so it has more resources.
> It includes Mailman, as well as MediaWiki and obviously static web / php /
> mysql.
>
> One caveat is it is "unlimited". That means they may throttle you or pull the
> plug. Zero3 talked to them, and apparently we'd be fine with a VPS, on the
> basis that we use <200GB/mo mostly but approx 800GB in a slashdot month. IMHO
> 100,000 downloads is well over what we've had recently for a release...
>
> Costs would be a little higher than most deals we have considered so far, but
> the basic package is free for 501(c)(3)'s as long as we can provide our
> 501(c)(3) determination letter. Hopefully they would give us the basic shared
> hosting cost for free and only charge the rest.
>
> $9.95/mo (1 year contract, cancel within the first 97 days) for basic
> hosting (this part may be free because we're a nonprofit)
> $15.00/mo (Virtual Private Server with 150MB of RAM)
> $3.95/mo (Unique IP Address) (Zero3 may be able to give us a promotional code
> avoiding this, although it probably can't be combined with non-profit hosting)
>
> So $28.80/mo, $18.85/mo if they let us off the basic hosting cost due to
> being a nonprofit. Assuming 150MB of RAM is enough. It would be overkill for
> static web hosting and downloads, whether it would be enough for the wiki and
> the bug tracker as well is uncertain.
>
They have confirmed they will let us off the basic hosting cost, so $18.95/mo.
If resource usage is a problem, we could bump it up (on the fly) to 1GB when we
expect a slashdot, and pull it back down before the month ends. During that
month the cost would be about what we pay for emu now - or rather, our half of
what is paid for emu (ian pays half). The rest of the time it is 6 times less
than we pay now. And there is much less administrative overhead.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 835 bytes
Desc: This is a digitally signed message part.
URL:
<https://emu.freenetproject.org/pipermail/devl/attachments/20090604/7ca2a4f0/attachment.pgp>
