David's "obvious option" seems like a great way to start: > One obvious option is to also move site generation to us.metamath.org on Linode. > Then the site can be re-created each day, & then moved over to its public display, > on the same machine.
Yes indeed. I would say just perform the process that way a couple of times, then look at any extra charges that occur. Judging from Linode's price list, I would expect maybe a slight charge for the CPU time used by build steps, but that's about all. If the builds need more resources, e.g. more RAM or SSD, then the monthly rate might go up. But in such a case you will see that either the build will get very slow or you will get a complaint that the system has run out of storage (SSD). You would probably want to "move" the site by renaming rather than copying, hopefully just renaming one top-level directory. A couple of renames can be done in a tiny fraction of a second, and with an absolute minimum number of intermediate states. A copy on the other hand has almost limitless numbers of intermediate states that can occur if it is interrupted, runs out of space, or whatever, and can take a quite noticeable amount of time. Any of this is liable to leave the public site in some inconsistent and/or undesirable state. -Cris On Fri, Mar 18, 2022 at 9:16 PM Jim Kingdon <[email protected]> wrote: > Assuming we stick with the current linode set up, running us2 on the > same linode is perhaps simplest. Whether that changes the cost, as far > as I could tell from glancing at https://www.linode.com/pricing/ , > depends on how close we are to the limits for RAM, storage, etc. Perhaps > linode has an admin page which shows where we stand, and/or logging into > us2 and observing the job run can help estimate? > > I'm potentially in for some sort of monthly donation. Any ideas on > whether something like https://opencollective.com/ is overkill for > managing the money? At first glance it would appear to make it easier to > do things like make it possible for all the donors to see what is being > spent on what and the like, but I've never set one up so I don't know > whether it is a lot of hassle (compared with something more informal, > which I suppose is the alternative). > > As for dynamic versus static, I suppose it is to some extent a tradeoff > between RAM/CPU on the dynamic side, versus storage on the static side, > but even if I was sure how those tradeoffs might work out (I'm not), > seems like we should not expect a change of that magnitude to happen > quickly (just in terms of the software development involved - whether > that is entirely new code to generate the pages, or even just a way to > add enough duck tape to make the existing code to function dynamically, > assuming that's even feasible). > > Another approach - which I suppose has at least some of the same issues > in terms of writing new code or at least scripts, investigation, etc, > would be pushing static HTML to Github Pages ( > https://docs.github.com/en/pages for those not familiar). I suppose this > would involve taking the job which now runs on us2 and having it push > the pages to a git repository which is for the purpose of holding the > generated HTML. I know our generated HTML is pretty big (am I > remembering 2Gbyte correctly and if so has it changed since then?) and > perhaps there are limitations of github pages (or similar services) > which mean this wouldn't work, and I think it falls in the same bucket > as a dynamic site, in terms of being a change in approach. > > On 3/18/22 17:02, David A. Wheeler wrote: > > Currently the public metamath web pages are generated on > us2.metamath.org, > > a computer in Norm's home, and then transferred to us.metamath.org > (hosted by Linode) > > where most people see the results. > > > > Susan Cass, Norm's partner, is currently paying for Linode hosting > > and providing a home for us2.metamath.org. However, she'd like someone > else to > > pay for Linode hosting, and wants to transfer the work elsewhere > > (so she doesn't need to be responsible for it). > > She'd like that done preferably by mid-June 2022 or so. I think that's > perfectly reasonable. > > > > I'll find out the cost of the Linode server, and I'm willing to help > fund it. > > I may not be able to pay for it all, depend on its costs. > > If others are willing to contribute financially please let me/us know. > > > > In addition, does anyone have suggestions for how to replace > us2.metamath.org > > to generate the website? I think it should be a cloud service so we > aren't dependent on someone's home. > > One obvious option is to also move site generation to us.metamath.org > on Linode. > > Then the site can be re-created each day, & then moved over to its > public display, > > on the same machine. I don't know if that would significantly increase > our Linode bill; > > does someone else know? If there are other reasonable options, please > reply! > > > > There are many other ways to distribute information, but I think it's > nice to have > > simple static HTML files, so I don't think we need to rewrite it all use > dynamic generation. > > In any case, I think it'd be safer to switch computers first, and then > consider massive > > changes to approach as a separate step. > > > > --- David A. Wheeler > > > > -- > You received this message because you are subscribed to the Google Groups > "Metamath" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To view this discussion on the web visit > https://groups.google.com/d/msgid/metamath/61f986f0-aab4-5ca2-35e2-cac3fa09f987%40panix.com > . > -- You received this message because you are subscribed to the Google Groups "Metamath" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/metamath/CAOoe%3DWKhv6K7bTOJRtL00QQVRFWJhTOHhGHs8yS-qApbm9Cp-A%40mail.gmail.com.
