(meta discussion note: If there are no objections before March 1, I think it would be a good idea to move this discussion to the DistributedEditing page. – DavidCary).

Alas, the server hosting the WikiIndex ( http://wikiindex.org/ ) is having problems today. Oddly, “search” seems to still be working, giving me a list of matching wiki and a tantalizing excerpt of information about each one. But actually clicking on any search result sends me to http://wikiindex.org/Internal_error.html . I hope it is just a temporary glitch.

What can I do about not just today’s glitch, but wiki glitches in general?

  • shorter term: what can I do to get the WikiIndex back online?
  • short term: promote the WikiNode network, since it could (in theory) replace the WikiIndex, and yet it survives any one server failing.
  • medium term: I’m looking for people to help me build a fault-tolerant DistributedEditing wiki ( WikiFeatures:FailSafeWiki ). What’s the Wiki:SimplestThing I could do to get a fault-tolerant wiki online?
  • longer term: how do I attract a critical mass of people to the fault-tolerant wiki?

Hi David, looks to be a temporary glitch, but raises many great questions and ideas - as you have outlined. One that I also thought of is where to have contact information, off the wiki. Hmmmm…. WikiIndex, are you here or are you a NearLink?

I’m glad Wikiindex is back online. For any wiki other than wikiindex, perhaps putting contact information in its wikiindex entry would be a good idea. Or would http://Aboutus.org/ be a better place?

    • David: I’ve been thinking for quite some time that it would be good to have a second Oddmuse WikiHive like the one that Alex provides for the ‘bliki’s.
      • SamRose has implemented one for his Social… initiatives.
      • I would like one that I can …
        • open up for use by other business associates, without worrying about “leeching” bandwidth and other consumable resources (disk space, processor time, etc…)
      • It believe it should be possible to “mirror” various blikis, provding a more effective ‘fault-tolerant’ environment since any single system failure should be automatically able to catch up, whenever it somes back on line.
      • Sam has asked me to make some notes about what my requirements for such a system wold be, since I have stated several times that I am very interested in funding it, but currently lack some of the the skills I suspect are needed to install and maintain it.
    • I hope you see this interjection as a well intended expression of interest, and not as an intrusion.
    • HansWobbe

Good points.

  • “leeching” – an even more precious resource is human time.
  • “mirror” – When the primary system fails, being redirected to a read-only mirror would be better than the “server not found” error we have now. (Using services such as Wikipedia:Dynamic_DNS). Later I hope we implement systems that are even better than that.
  • I hope that we can get enough people interested in this project that no one gets overloaded. I hope that we can lower the barrier to contributions such that people – even ones not willing to give us a lot of time – can still help us out by contributing rack space, hardware, bandwidth, disk space, bits of code, etc.
  • It is possible that some of your requirements may conflict with some of my requirements. In that case, I would be happy to help you build your system, and for you to help build my system, ending up with 2 distributed wiki.
  • We want to spread the disk space and bandwidth load over as many people as possible, right? Is releasing the source code under some Open Source license the best way to get the distributed wiki installed on as many machines as possible?
  • What is the next step?

Some suggested next steps …

  • I’ve started to make a few notes about what I think my immediate needs are (in my bliki, fir the moment, since I do not want to add ‘clutter’ to a page like this one.
    • I’ll incorporate links or TransClusion(s) as soon as I have something to show that is worth looking at.
  • I agree that …
    • “(time) leeching” is an even greater concern than just the loss of technical resources.
    • depending on the frequency of the “mirroring”, synch issues may not even be important, especially since all wikis generally have good “DIFF and RecentChanges mechanisms that can be used to resync as needed after a failed systen is recovered.
      • It might even be relatively easy to have a failed system restart that checks its known mirrors for updates that occured while it was “off-line”.
  • I’ll try to create an inventory of the technical resources I can contribute as well as what I think I will need for my own purposes. Ideally, I like to keep my utilization below 50% of the available capacity. Hence I should have ‘spare’ resources that can be used within a ‘second-tier’ Service Level Agreement. (I like this because volatile resources (like the unused bandwidth I currently pay for) cannot be stored.)
  • Since I would be remise if I did not plan for a possible “fork”, why not just assume that a shared resource has some limited ‘term’ and define a reasonable “notice” period that allows for an orderly separation, should it be needed.
  • Personally, I believe eveything I place on the publically accessible internet is Public Domain. I prefer Open Source solutions, but also know that there are quite a few thins I prefer to pay for and I do so. Its all just a question of current values, as far as I’m concerned. In short, I really don’t care what its called, I care about what resources I need consume in order to meet my Objectives.

I have thought about this mirroring idea. it led me to another thought: that people can cluster their servers together to act like one big server (quoted from http://socialsynergyweb.net/cgi-bin/wiki/2007-02-20_Talk):

This whole thing also makes me think about how Google distributes, mirrors, and serves it's search data. Also known as the Wikipedia:Google_File_System (GFS).

I have heard of ways to take different co-located Apache webservers and combine them into a distributed serving cluster. This is, I think, what we are really talking about here. UltraMonkey may be one sultion. But I have no experience setting that up. Although, that doesn’t mean we could not experiment with it and see what we can come up with.

See also:


It would be interesting to have a conversation with Linux/UNIX admins who’ve already had experience with LinuxVirtualServer? to see what recommendations or feedback they might have.

UltraMonkey? is interesting, because it can cluster servers with any operating system to gether. This seems like one of the best ways to go about doing this, to me.

Also, for mirroring, this looked interesting: http://www.supersparrow.org/ss-0.0.0/


Define external redirect: UltraMonkey LinuxVirtualServer

EditNearLinks: FailSafeWiki WikiIndex