Want to take part in these discussions? Sign in if you have an account, or apply for one below
Vanilla 1.1.10 is a product of Lussumo. More Information: Documentation, Community Support.
I've never been happy with the fact that only Urs, Toby, and I can make proper backups of the nLab, and that owners of personal webs have no easy way to take backups of their webs. So I've written a little program which takes the database and spits the contents of the pages out into bzr repositories which are available over the web. It's easy to take a backup of such a repository, and to automatically schedule updates each day. There's one repository for each personal web, and the repositories for webs that have passwords are also password protected. What you get from this is the contents of each page (including history) together with the basic meta-data (who edited what and similar stuff). It's not quite so easy to simply reconstruct an Instiki installation from these backups as from a raw backup of the database, but this seems the easiest way to make it possible for anyone to keep a backup of whatever they are interested in.
It's quite possible that I'm the only one round here that understands what a "bzr repository" is! No matter. If someone wants help setting up a backup for their personal web, or for the nLab itself, then post a response here and I'll talk you through it.
Indeed, given that the nLab is quite large, the initial download for the nLab is also quite large and so if someone wants to start keeping backups for the nLab itself, then it's best if I send you the initial repository as a suitable archive file (zip or tar or something) than you start by downloading it from the server. Once it's set up then the increments aren't much to download each day.
In fact, I'm not going to put the link here just yet to protect against someone just going out and effectively downloading the whole nLab in one go! That way, I can keep an eye on who's doing it and ensure that this doesn't end up taking up too much bandwidth (I have a backup plan in place if it does).
But I recommend that owners of personal webs do this. If you really feel that this is outside your comfort zone, I can set up my machine to backup anyone else's personal web as well (at the moment, I just do the nLab). Again, post a response here if you'd like me to do that.
I would like to do this for my personal web. And I even know what a bzr repository is! So I’ll probably be good if you just give me the link.
How big is the main nLab download?
I would like to have some occasional backup of my web, but I don’t know what a bzr repository is :)
Regardless of what I know and do, can I trust that some backup of the entire nLab, including all the personal webs, is kept somewhere? Or is this some thing that needs me to take care of it?
@Eric: bzr is the short name for the version-control system "bazaar". The main page for this is http://bazaar.canonical.com/en/. There, you should be able to download the correct version for your operating system. That's where you need to start. (Please let me know if you want more detailed instructions.)
@Mike: The main nlab repository is 117Mb, which is why I'd rather start people off with a "seed". The next biggest (probably Urs' web) is less than 6Mb!
@Urs: that's partly what this is meant to fix: the fact that there's no automatic system for backups. The nlab produces backups, but they simply sit on the nlab server waiting for you, me, and Toby to download them. Unfortunately, to download them, you need to properly log in so you can't set it up automatically.
I see. So I need to make a backup myself as soon as possible.
I would like to suggest that anyone who thinks about doing a backup of his personal web does one of the entire nLab at the same time. The more copies there are around, the less paranoid we all have to feel.
Urs, I completely agree with that. The only thing I’m trying to avoid here is lots of people starting their backup of the nlab from scratch. I want to be sure that they get the initial repository from me since it’s rather large.
Right, Andrew, I understand. All I am saying is that if anyone goes through the trouble of producing a backup – by getting an initial repository from you ! – should please consider doing a backup of the full nLab, not just of one personal web.
Digital information is volatile like a single DNA strand is. The fact that DNA/RNA information is passed on to us through a good 4.5 billion years now and through a handful of planet-wide catastrophes is because copies were made in abundance.
Again, I agree.
It’d be useful to put some instructions on how to do the backups (maybe on nlabmeta); but as I already know exactly how to do it, I’m the worst person to write the instructions!
Well, maybe I could start by writing the basic steps at backups (nlabmeta) and then others can fill in the details.
I added some comments to backups (nlabmeta) about the commands one needs to run in each case. Andrew, please correct if I am wrong; I am not as familiar with bzr as with some other vcs.
This issue came up again here. Perhaps it is possible to also do a backup to some central git repo, i.e. github. Adeel commented that the nlab (in bzr) may be too big (1GB) for a free github account. bitbucket has a 1GB soft limit and a 2GB hard limit. However, git is known for its efficient data structures.
It may also be possible to use other git repos, i.e. one run by a university.
My impression is that git has “efficient” data structures in the sense of speed, not size.
I know nothing about it, but this link would suggest that git could make a significant difference regarding size. I have used both mercurial and git with bitbucket, and am entirely content with both, and with bitbucket.
I agree with Bas: I would have thought it possible to host a git repository on a university server. Around 1GB should, I would think, be easy to find space for on such a server.
Have you had the chance to think any more about the possibility of backups for the nForum, Adeel?
Richard: thanks for reminding me, I’ll set that up soon.
I believe that GitHub is very friendly towards academic purposes, so they might allow a somewhat larger repository. I have communicated GitHub statt before and could ask them, if wanted.
Thanks! Since the repository is at the moment still slightly under 1GB, I am trying to upload it to GitHub right now. I’m using a tool called bzr-git which allows pushing and pulling to/from git repositories. It will take some time though, as it’s been uploading all day and it’s only at 25%. Once that’s uploaded I will try to contact the GitHub staff.
Apologies for the late reply to #14, Adeel: that would be great!
Great also to hear that you are looking into uploading the nLab to GitHub!
Currently Joseph Ramsey, who, kindly and voluntarily, is physically handling the server that the Lab is running on, is backing up by hand to an external hard drive. And it seems Adeel is busy. If you want to lend a hand, maybe best to contact Joe directly.
The git repo https://github.com/ncatlab/nlab-content is no longer updated: the last commit was on April 16. Would it be possible to fix this?
I will take a look when I get the chance, hopefully over the weekend.
1 to 22 of 22