18:00 < Smerdyakov> Time's up!
18:00 < ntk> alright, meeting-time.
18:00 < jts> As of a minute ago!
18:00 < Smerdyakov> OK, I'm officially declaring the start of the meeting.
18:00 < ntk> can we get a hcoop login/real name roll-call, since there's a lot of idlers here?
18:00 < Smerdyakov> Would everyone watching please write "Here"?
18:00 < ntk> I'll start, ntk/Nathan Kennedy here
18:00 < jts> Here
18:01 < Smerdyakov> OK, and say what ntk asked for instead. :)
18:01 < jts> aka nion/John Settino
18:01 < Smerdyakov> adamc/Adam Chlipala
18:01 < leitgebj_> leitgebj/Justin Leitgeb here
18:01 < vol> shaun/Shaun Empie here
18:01 < ntk> and ... ?
18:01 < Smerdyakov> Missing docelic..?
18:02 < jts> He was here somewhere
18:02 < ntk> he just joined ~5 minutes ago
18:02 < leitgebj_> what about mwolson?
18:03 < Smerdyakov> leitgebj, I don't know, but he's not a board member, so his presence doesn't have an official significance.
18:03 < leitgebj_> ok
18:03 < ntk> well, i think we can assume that docelic is at least here, hopefully he will get back to the computer and register his presence on the nonce
18:03 < ntk> i move to have the last meeting declared official.  it was announced properly and a majority of the board was present
18:04 < Smerdyakov> ntk, is that what the bylaws say?  I forgot. :)
18:04 -!- leitgebj [n=leitgebj@pool-70-18-123-125.buff.east.verizon.net] has quit [Read error: 110 (Connection timed out)]
18:04 -!- leitgebj_ is now known as leitgebj
18:04 < ntk> yes
18:04 < jts> I would second that but I wasn't at the meeting. :p
18:04 < ntk> i don't object to anything that was done last time
18:04 < Smerdyakov> ntk, actually... I don't see any word either way in the bylaws.  Can you point to a place?
18:05 < jts> Hey since we're at a meeting, I have an idea for the portal
18:05  * unknown_lamer is here
18:05 < Smerdyakov> jts, how about you wait until the end?  I think we have some bigger-picture stuff to cover.
18:05 < jts> Oh.
18:05 < jts> It was a small idea.
18:05 < jts> But sure. When _is_ the end?
18:06 < Smerdyakov> jts, I'll tell you.
18:06 < jts> lol
18:06 < jts> ok
18:06 < Smerdyakov> Well, we can figure out what last week's meeting is later.
18:06 < ntk> it's not in there explicitly
18:06 < Smerdyakov> I think we should move to deciding what hardware it is we want to host.
18:06 < ntk> yes
18:06 < Smerdyakov> leitgebj has proposed another setup with only a single server.
18:06 < ntk> firstly i think it is a very bad idea to lease half a rack and put one server in it
18:07 < Smerdyakov> I think it's important to understand what benefit this would be bringing us over our current set-up.
18:07 < ntk> i think we'll end up overpaying for yet another single-server setup for an indeterminate amount of months.  we already have abulafia and a serious offer of a good server donation
18:07 < ntk> and we intend to buy another one
18:07 < ntk> at leaswt
18:07 < Smerdyakov> As far as I can think, the only benefits are a constant factor more disk space and faster disk system.
18:07 < ntk> which is not why we want to move hosts
18:07 < Smerdyakov> Righto.
18:08 < leitgebj> Smerdyakov, I would also suggest that he.net has better and more reliable bandwidth.
18:08 < Smerdyakov> leitgebj, yes, but we are talking about hardware now.
18:08 < leitgebj> ok, let's stick to hardware
18:08 < Smerdyakov> leitgebj, those reasons actually have much less effect on the average member than some other reasons we want to move.
18:09 < Smerdyakov> As ntk alluded to, we've already had one big move that required lots of administrative hassle in the interim.
18:09 < Smerdyakov> I've been thinking of this move as one that will be sufficient for a few years.
18:10 < Smerdyakov> (The last one was a year ago.)
18:10 < jts> Has it really been that long Adam?
18:10 < Smerdyakov> That means that _scalability_ is the main design goal.
18:10 < Smerdyakov> Any later changes that involve serious reshuffling of filesystem organization, etc., to, e.g., accommodate a network file system invalidate the whole point.
18:11 < Smerdyakov> jts, yes.
18:11 < vol> so are we decided on 2 servers and a switch?
18:11 < ntk> ok
18:11 < unknown_lamer> At the very least we need to have web serving and file servering split so that adding disk space is as simple as grabbing another machine and creating an afs master volume on it.
18:11 < Smerdyakov> So it seems to me that, if we're going to expand, we should do it with multiple servers.
18:12 < ntk> we have an offer for the donation of a gently used, capable server from leitgebj
18:12 < Smerdyakov> unknown_lamer, another desirable split is server where all members can log in and a server where only admins can log in.... and it sure would be nice to have a web server for static content separate from the members-can-log-in machine... which would indicate at least 3 servers, if we follow both suggestions.
18:12 < ntk> and we still have abulafia, which could make up the third server in a new configuration with 1 switch, 1 badass new server, leitgebj's server, and abu
18:13 -!- gnu` [n=gnu@61.246.60.123] has joined #hcoop
18:13 < unknown_lamer> Smerdyakov: I agree that three would be nice, but I think that two is the absolute minimum
18:13 -!- gnu` is now known as aadis
18:13 < unknown_lamer> Smerdyakov: + abulafia makes a fine general login server
18:13 < jts> ntk: As far as "badass server" goes, were we thinking of building it ourselves or just buying one "out of the box"
18:13 < jts> ?
18:13 < leitgebj> Smerdyakov, I agree that this would be ideal, but I have yet to see a hosting option that would make this practical for our current utilization, especially if we want to be located on a strong ISP.  From what I have seen, any setup on a strong ISP with good support that would allow us multiple servers is going to put us at least at $700/month.  Do you think that we can support this?
18:13 < vol> so the only thing we have to purchase is 1 server i nthat case
18:13 < ntk> jts: that is part of a discussion that i have been secondary to.
18:14 < jts> unknown_lamer: That and we can actually use IRC and such on abu :)
18:14 < ntk> i would address it to the channel at large
18:14 < ntk> have we ruled out peer1 as our next step up?
18:14  * leitgebj apologizes for bringing discussion back to hosting, but sees it as important.
18:14 < Smerdyakov> I think that we aren't thinking of any possible "server classifications" beyond: file server, locked-down server with services not extended by members with dynamic content, members-can-log-in server for dynamic content hosting, and general shell server.
18:14 < ntk> understand that i could install and provide basic remote hands for peer1
18:14 < leitgebj> Peer 1 would be great, but again, looking at $750 I believe to start out.
18:14 < Smerdyakov> Some of these categories can be merged.
18:14 < leitgebj> That is per month.
18:15 < ntk> we have 66 paying members now, correct?
18:15 < Smerdyakov> leitgebj, if we thought of splitting the costs evenly among members, my share would be quite reasonable to me.
18:16 < jts> If I do the math that's around $11 a month
18:16 < ntk> yes
18:16 < Smerdyakov> Less than Netflix unlimited plan ;-)
18:16 < jts> Which is still not bad
18:16 < ntk> we would have to introduce different hosting plans
18:16 < unknown_lamer> I think we have enough people to pay extra shares in the short term too (e.g. I'm able now to pay $20ish a month if needed)
18:16 < ntk> i would say have a basic, introductory plan for flat $5 with maybe half a gig of storage and about as much bandwidth
18:16 < jts> ntk: At some point we're going to have to start offering dedicated "plans" or such
18:17 < ntk> and then the regular deal for like $15 or $20 or whatever
18:17 < Smerdyakov> jts, I hope not.  More like a cost per gigabyte of disk space, and other very concrete, pass-on-the-cost schemes.
18:17 < leitgebj> Since we are on the topic of member rates, I would like to include discussion of how we can make a fund for server upgrades and maintenance part of this.  It seems that this is an important part of being a sustainable organization.
18:17 < iriefrank> as long as the basic plan doesn't take away important services: shell access, mail, unlimited domains, etc.
18:17 < jts> Smerdyakov: That's kind of what I meant. You know, you pay $x and get x of x kind of thing.
18:17 < Smerdyakov> leitgebj, our current official guidelines mandate a $10 deposit from each member.
18:17 < Smerdyakov> leitgebj, that's $660 available in a pinch now.
18:18 < unknown_lamer> iriefrank: those services cost nearly nothing
18:18 < unknown_lamer> iriefrank: thus it makes no sense to charge extra for them
18:18 < iriefrank> yeah i know, i just dont want to get silly in determining levels
18:18 < unknown_lamer> disk space has a real cost
18:18 < Smerdyakov> jts, but I'm thinking of it in terms of "each month measure Bob's resource usage to determine his bill," not "Bob chooses 'a plan' that determines what he can use"
18:18 < leitgebj> Smerdyakov, what about when we have to use the deposits?  Then are funds are dry.  I think that it should be part of monthly payments.
18:18 < ntk> i would like to mention accounting briefly--according to 501(c)(12) we can charge extra and keep retained earnings as long as we deposit them in an account and record each member's pro-rata interest in that account.  then we can spend out of that account for any hcoop expenditures.
18:19 < Smerdyakov> unknown_lamer, I think the main cost will come from an off-site back-up service, which we should add.
18:19 < ntk> so i think flat rate plans are not a bad idea
18:19 < jts> Smerdyakov: So you mean something like "well, x gb of bandwidth is going to cost $x" and then divide and round and such?
18:19 < ntk> deposits aren't going to cut it
18:19 < Smerdyakov> jts, yes
18:19 < leitgebj> I agree with ntk, and fully support the notion of a flat-rate plan of some sort.
18:20 < Smerdyakov> leitgebj, what I mean is that we plan monthly dues to cover expected costs, and the deposits provide a quick slush fund in case we underestimate, to be replenished by increased dues thereafter.
18:20 < jts> Even though Smerdyakov's idea sounds good I think flat rate plans would probably work out better.
18:20 < enthalpyX> Has anyone done any calculations as to the growth rate of hcoop?
18:20 < Smerdyakov> jts, can you make an effort to use punctuation properly? :P
18:20 < enthalpyX> how many new members are we gaining/mo?
18:20 < Smerdyakov> enthalpyX, feels like about 2.
18:21 < jts> How am I not using punctuation properly
18:21 < jts> ?*
18:21 < ntk> I was thinking that we could use some sort of membership drive
18:21 < jts> THAT was a typo.
18:21 < Smerdyakov> jts> Even though Smerdyakov's idea sounds good_,_ I think flat rate plans would probably work out better.
18:21  * docelic here
18:21 < Smerdyakov> ntk, AFTER we have a scalable architecture going!
18:21  * jts waves
18:21 < ntk> ok
18:22 < enthalpyX> ntk: that's a good idea. because if it's only 2/mo, it would take quite a while for the 750/mo figure to get down to "cheap as free"
18:22 < Smerdyakov> ntk, if you disagree, then I'd like to hear what offsets the pain of moving extra members manually when we _do_ switch.
18:22 < ntk> but if we did a serious member drive and outreach, we could probably double our size within a couple of months
18:22 < leitgebj> ntk, I agree.  I was also thinking that we could offer hosting to other cooperative organizations and promote this service.  Of course, this would require a hosting provider with stable bandwidth.
18:22 < ntk> don't forget, that once we get a real rack with power we can do collocation
18:22 < ntk> that would reduce costs significantly
18:23 < jts> On the issue of hosting... I'm assuming we're looking to colocate with a dedicated 1/3/5/10/?/mbps instead of a bandwidth package
18:23 < Smerdyakov> ntk, this came up before, and we don't seem yet to have anyone willing to commit to colo with us.
18:23 < vol> yea i like the idea of doing colo
18:23 < jts> Am I correct?
18:23 < Smerdyakov> jts, yes.
18:23 < ntk> we haven't really tried to find anyone either though
18:23 < ntk> once we actually have a rack it will be easier to sell
18:23 < ntk> especially if we are underselling all the commercial providers
18:23 < leitgebj> ntk, it could be a sticky point, too, if someone exceeds their bandwidth quota, and therefore ours as well.  This doesn't rule it out of course, but I think we would have to choose our tennants carefully.
18:23 < Smerdyakov> ntk, I asked Josh Tauberer... he said no, at least until his current contract ran out.
18:24 < leitgebj> ntk, or have an agreement that you wrote up. ;)
18:24 < jts> Smerdyakov: There is a local datacenter I was thinking about using. At the point where we decide what exactly we're going to colocate, I can ask for a quote.
18:24 < ntk> leitgebj: we'll do that the same way, just bill them at cost for any excess bandwidth they use.  that's not a big problem.  our managed switch will take care of it
18:24 < jts> Meaning I was thinking about using it for my own business.
18:24 < Smerdyakov> Somehow Josh has a sweet $50/mo. deal including a rented server.  That's probably hard to beat.
18:25 < jts> I have a $65/mon leased dedicated server from sagonet.com
18:25 < leitgebj> jts, during our search for colocation providers we gravitated away from using small providers.  We wanted something that would allow us to scale and have stable bandwidth.  Of course, member opinions could have changed since then.
18:25 < unknown_lamer> My dad's colo has solid hardware and connectivity, but they won't have more rack space for a few months :-9
18:25 < Smerdyakov> leitgebj, we have a 100% "getting bitten" rate with small providers so far.  I'm all in favor of running with the big dogs.
18:25  * ntk notes that clinton has a conflict of interest :-)
18:26 < unknown_lamer> Nah, I hate my dad :-)
18:26 < jts> Smerdyakov: I'm pretty sure this place has a lot of space available. My partner visited them recently, and said he saw empty racks. I think he mentioned they are a new(er) company.
18:26 < jts> LOL
18:26 < Smerdyakov> jts, our past problems have nothing to do with "not having space available," so I don't know why you mentioned it.
18:27 < jts> Smerdyakov: You mentioned scalability. Space = allows for more servers/racks/etc.
18:27 < docelic> I read all the backlog, and I don't see where this discussion is going?
18:27 < Smerdyakov> leitgebj, it may be interesting to mention that your own vote in the poll on prices indicates that you're not willing to pay an even share of the $750/mo..  Is that accurate?
18:27 < ntk> alright, have we decided on 2 new servers (with possibly one donated) + abu for a new rack?
18:28 < jts> ntk: Sounds good to me.
18:28 < leitgebj> No, if we are using a flat rate structure I will pay more.
18:28 < docelic> ntk: "with one donated" means three, or 1 of the 2 is donated ?
18:28 < Smerdyakov> leitgebj, huh?  Any rational reason? :)
18:28 < ntk> docelic: the latter
18:28 < ntk> one of two
18:28 < vol> docelic: 1 donated, 1 new, and abu
18:28 < leitgebj> Smerdyakov, I try not to let reason dominate anything that I do unnecessarily.
18:29 < Smerdyakov> ntk, what role do you have in mind for it?
18:29 < jts> brb
18:29 -!- jts is now known as jts|poop
18:29 < Smerdyakov> ntk, I'm not sure it's suitable anymore for anything but a misc. shell server.
18:29 < leitgebj> But I think that if we have a concrete costs, and a solid pricing structure, I will pay what is necessary to sustain the organization.
18:29 < docelic> lol jts, this is an official meeting. 
18:29 < Smerdyakov> (Where we don't mind it's going down.  We won't find someone willing to sell us a support contract for it.)
18:29 < docelic> Ok, that sounds good to me, as agreed previously, just amended with the fact we would get one from donation. Two servers, a swtich, + Abu. Great deal
18:29 < ntk> ok desk man here
18:29 < ntk> brb
18:30 < Smerdyakov> Does anyone else have a view on what Abulafia's role is here?
18:30 < Smerdyakov> It would be nice if we can get a filesystem appliance to serve as one of the "servers".... that would be a very clear point of scalability.
18:30 < docelic> Smerdyakov: someone said a general shell server, I think this fits it perfectly.
18:30 < Smerdyakov> So plus Abu, that leaves just one slot.
18:30 < Smerdyakov> Which can't be both a server that any member can log into and a server for admins only.
18:31 < leitgebj> Smerdyakov, what are we talking about with one slot?  Peer 1?
18:31 < Smerdyakov> leitgebj, if we agree on exactly three servers
18:31 < vol> Smerdyakov: so you want to go with 3 servers plus abu as a shell box?
18:31 < Smerdyakov> vol, no.  I think that is too much for our current needs.
18:32 < unknown_lamer> Smerdyakov: I don't think an fs appliance would be good; they are essentially expensive and uncustomizable normal servers
18:32 < docelic> Yes, I tend to agree with unknown_lamer 
18:32 < vol> well, i think instead of a 7u setup we can find a provider that will have 1/2 rack configs available
18:32 < unknown_lamer> Smerdyakov: giving us another point of failure ("we found a bug" vendor: "tough shit you only bought one and we only care about our big customer")
18:32 < unknown_lamer> for colo:
18:33 < unknown_lamer> http://www.globaltap.com may be worth emailing
18:33 < Smerdyakov> unknown_lamer, fine, but then at least a dedicated file server.
18:33 < unknown_lamer> Smerdyakov: yes
18:33 < leitgebj> vol, have you seen the Peer 1 package?  It is 10 U.
18:33 < Smerdyakov> unknown_lamer, how did you determine that globaltap is in the same league as Peer1 and HE?
18:33 < unknown_lamer> (globaltap is run by a guy from the local lug + another guy from dc sage)
18:33 < docelic> Smerdyakov: not to forget, we had a serial console in mind last week as well.
18:33 < unknown_lamer> Smerdyakov: my roommate'
18:33 < unknown_lamer> s current client has stuffed racked there
18:34 < unknown_lamer> Smerdyakov: they aren't a *big* vendor but they aren't going anywhere anytime soon and they have good connectivity
18:34 < Smerdyakov> unknown_lamer, what about bandwidth?
18:34 < Smerdyakov> Globaltap Advantage Colocation Special: Free 1U of space with 10mbps/month commitment for 6 months.
18:34 < unknown_lamer> Smerdyakov: not sure, I have to email them about it (they blanked my mind until my roommate just walked by and my brain was in colo mode)
18:34 < Smerdyakov> The magic "free" colocation offer says: Almost unlimited bandwidth
18:35 < Smerdyakov> So where does the money come in??
18:35 < unknown_lamer> Corporate customrs
18:35 < Smerdyakov> unknown_lamer, OK... that sounds like something fork off and join back later.
18:36 < Smerdyakov> We need to figure out what we would ideally host in a usual sized rack.
18:36 < unknown_lamer> I think the free 1U is an extra 1U of space if you pay for bandwidth (which I imagine is a ... costly if you want 10Mbps)
18:36 -!- ntk [n=ntk@pdpc/supporter/student/ntk] has quit []
18:36 < unknown_lamer> I'm emailing him now
18:36  * Smerdyakov throws a bone at ntk's ghost.
18:36 < unknown_lamer> (not their usual support line but the owner)
18:36 < unknown_lamer> advantage of being in his extended social circle
18:37 < vol> i think 1 member accessible server, 1 admin server, nas box, shell box, switch, console box
18:37 < vol> is what we're looking at right now
18:37 < docelic> Only to have him reply with "Contact my staff" :)
18:37 < Smerdyakov> vol, nas?
18:37 < vol> network area storage
18:37 < Smerdyakov> vol, so 4 of those are things we'd think of as "servers"?
18:37 < docelic> Yes, that's basically what we discussed previously. 
18:38 < docelic> With a possible merge of admin only and fileserver
18:38 < vol> Smerdyakov: yes
18:38 -!- ntk [n=ntk@pdpc/supporter/student/ntk] has joined #hcoop
18:38 < ntk> dammit
18:38 < ntk> ok
18:38 < Smerdyakov> ntk, I threw a bone at your ghost.
18:38 < vol> we could fit all of that in 10u
18:38 < ntk> i am back
18:38 < ntk> thank tcp/ip that ntk_logger was still here
18:39 < Smerdyakov> ntk, you suggested a 3-server config that includes Abu.  What role did you have in mind for Abu, seeing that I don't think we can achieve high reliability for it?
18:40 < leitgebj> vol, remember that the base Peer 1 package was only 10 amps I believe.  We have to check on this if we're considering using their base package.
18:40 < ntk> i think it would make a good login node, not actually serving anything besides shells
18:40 < ntk> we don't want web or mail on it
18:40 < Smerdyakov> ntk, OK, so then what are the other 2 servers?
18:41 < ntk> i thought this was discussed earlier in a 2-server configuration?
18:41 < Smerdyakov> ntk, you are thinking of shared filesystem and most services on the same machine?
18:41 < ntk> i guess one running the dynamic hosting crap, the other one doing files, mail, static hosting, etc
18:41 < ntk> with the latter being a strictly admin-login only machine, and being the badass server
18:41 < Smerdyakov> I have this funny, unjustified feeling that the most important change for scalability is keeping a dedicated fileserver.
18:42 < ntk> it is
18:42 < enthalpyX> will it be tricky to separate dynamic hosting from static hosting?
18:42 < ntk> i think it will be a separate subdomain is all
18:42 < Smerdyakov> ntk, a fileserver that is also a web server is not dedicated.
18:42 < ntk> that is true
18:42 < Smerdyakov> enthalpyX, I don't think so.  Just changes to domtool.
18:42 < ntk> but having a fileserver at all, is a big change
18:42 < ntk> we don't need a dedicated fileserver for 3 servers
18:43 < ntk> especially if one is a staid shell-login server
18:43 < leitgebj> Smerdyakov, we should be able to mount new volumes rather easily later on, right?  vol's idea of a NAS is what I was thinking about.  I don't think it's reasonable to consider buying all of the space we will need right now.  If we use the PowerEdge, the 146 GB RAID 10 drive should be enough until we can afford and install a NAS later to supplement our available space.
18:43 < docelic> In fact, what would users of the "shell server" Abu do? They'd log in, and ? What would be typical usage? Irc sessions? Compiling? Network probes? Location from which you'd run wget? 
18:44 < ntk> docelic: everything that most users currently do on fyodor
18:44 < Smerdyakov> docelic, ask metaperl... number one abuser. ;)
18:44 < ntk> except they would be on a network filesystem
18:44 < ntk> so all that crap would be moved off of the production hosting servers
18:44 < Smerdyakov> ntk, that's not what I have in mind.  Things related to Internet hosting should be on a reliable machine.
18:44 < Smerdyakov> ntk, I still believe that most HCoop usage is truly related to hosting.
18:44 < docelic> In that case obviously it can't be low-reliability that Abu is 
18:44 < ntk> people could still log into our "dynamic" server and do the exact same things (or in case Abu dies), but by default they would be on abu
18:45 < ntk> or whatever the login server is
18:45 < unknown_lamer> I really think that avoiding any prepacked file servers and just using afs would be a good idea
18:45 < unknown_lamer> lower cost and more flexible
18:45 < Smerdyakov> ntk, I would prefer to use technical means to _prevent_ people from doing general shell server things on anything but the shell server.
18:45 < ntk> docelic: why not?  it's not important, if the login server fails it is the least important one
18:45 < ntk> well that would be fine too
18:45 < docelic> ntk: if it provides shell access to your files, and it fails, you can't access your files.
18:45 < Smerdyakov> ntk, if people can't administer their dynamic web sites, that's just about the most serious problem possible here.
18:45 < ntk> i think it is less of a mayday situation if people can't log in than if mail or web goes down
18:45 < leitgebj> unknown_lamer, I don't understand.  using a NAS box, or something like iSCSI, or a coraid box doesn't eliminate AFS or AFS.
18:46 < docelic> unknown_lamer: yeah, we assume AFS 
18:46 < jts|poop> ntk: Agreed. Email and web are more important to me than being able to login to my shell.
18:46 < unknown_lamer> I don't trust commercial buzzwords
18:46 < unknown_lamer> NAS is one of them
18:46 < jts|poop> erm
18:46 < ntk> it is for everyone
18:46 -!- jts|poop is now known as jts
18:46 < unknown_lamer> jts: and I would hope jabber ;-)
18:46 < Smerdyakov> jts, any web site with dynamic content MUST run on a separate server... and it seems to be that server that ntk is suggesting be Abu.
18:46 < ntk> not being able to log into hcoop for a few hours is an annoyance.  pages down, mail bouncing or inaccessible, is horrible
18:46 < ntk> Smerdyakov: no
18:47 < Smerdyakov> ntk, it is not my conception that someone who uses HCoop purely for hosting purposes should ever have to log into a server without a support contract.
18:47 < ntk> i am suggesting: 1. server to do fileserving, mail, static web, NO USERS 2. server to do dyamic web and user daemons and everything else 3. abulafia, our default login node
18:47 < Smerdyakov> ntk, and I'm suggesting that people log-in to 2. most of the time.
18:48 < ntk> smerdyakov: i agree, but it is obviously less of a disaster.  mail bouncing is horrible
18:48 < ntk> smerdyakov: then that would make abu basically what it is now
18:48 < ntk> a useless lump of fluff not doing its job
18:48 < leitgebj> Well, if we have the space and power, we can throw abu online.  it won't hurt.  otherwise, drop it.  Also, mail bouncing is entirely avoidable if we have backup mx's set up appropriately.  It is an easy fix.
18:49 < Smerdyakov> ntk, no.  People like metaperl don't want to run their own local UNIX machines and will persist to use HCoop for that purpose.
18:49 < jts> brb
18:49 < ntk> metaperl could be using abu right now, and he's not, correct?
18:49 < Smerdyakov> ntk, reading newsgroups, IRC'ing, etc..
18:49 < ntk> and we may get multiple metaperls in the future also
18:49 < Smerdyakov> ntk, I asked him not to.
18:49 < Smerdyakov> ntk, remember, bad bandwidth rates.
18:49 < vol> unknown_lamer: when i say nas i dont mean a commercial prepackaged box, i just mean its purpose will be network storage
18:49 < ntk> i think our default shell machine should be dedicated shell, and i think abu can  handle that
18:50 < ntk> in any event, abu will be more economical on our own rack than trolling around xiolink
18:50 < Smerdyakov> OK, let's be sure we're clear on what the cost is having a dedicated file server vs. merging some of that functionality.
18:50 < unknown_lamer> Smerdyakov: if hcoop could support it I would honestly keep screen with an emacs in it because hcoop is guaranteed to be more available than my desktop
18:50 < Smerdyakov> If the extra power doesn't bring us over the 10A limit at Peer1 then.... the cost is nothing beyond one-time costs for hardware, right?
18:51 < Smerdyakov> (And perhaps administrative complexity)
18:51 < leitgebj> Smerdyakov, exactly.
18:51 < ntk> don't they also have setup fees themselves?
18:51  * ntk consults wiki
18:51 < unknown_lamer> email has been popped off to globaltap
18:51 < leitgebj> I believe it was a $400 rack setup fee for Peer 1.  It's definitely on the wiki.
18:52 < ntk> yes
18:52 < ntk> $400
18:52 < unknown_lamer> seeing how much space / cost / power / if we could buy more power / bandwidth included / how much extra costs / how much rack space they have available for any potential expansions up to a full rack for us
18:52 < Smerdyakov> So the fee does not depend on amount of equipment.
18:52 < ntk> 3 servers + 1 switch would NOT be > 10 amps
18:52 < unknown_lamer> ntk: not if they were xeons ;-)
18:52 < ntk> smerdyakov: no.  and i could go install extra or maitenance it for free
18:52 < ntk> otherwise if i was unavailable they have remote hands
18:53 < unknown_lamer> globaltap is also a train ride away for me
18:53 < ntk> ixnay on the eonsxay
18:53 < unknown_lamer> Yeah, intel sucks
18:53 < unknown_lamer> if only compaq hadn't killed digital we could be using alpha servers :-)
18:53 < ntk> obviously my installation abilities should be a secondary & getting-off-the-ground consideration only
18:53 < Smerdyakov> I'd like to move that we stop discussing colo providers at the moment, if globaltap is really a serious contender.  We need more information to make a good choice.
18:54 < ntk> especially considering it was not part of our initial round of solicitatione
18:54 < ntk> they are rather late in the game
18:54 < Smerdyakov> ...and continue discussion here assuming Peer1, just to have a sense of costs of different choices.
18:54 < leitgebj> I would like to proceed, pretending we are going to use Peer 1 for a while.  unknown_lamer can check on globaltap on his own and present a summary on the wiki for us to consider later.
18:54 < unknown_lamer> Smerdyakov: we'll probably have that monday, and I'll post to the wiki (I would imagine we won't be making the colo switch before then :-) )
18:54 < Smerdyakov> unknown_lamer, good.  Now don't mention it again, or I'll eat your eyes.
18:54 < docelic> Yeah, I'd like to move on too.
18:55 < jts> LOL
18:55  * unknown_lamer guards his eyes
18:55 < Smerdyakov> leitgebj, so you lean towards buying a machine designed as a file server and using your donation as the main admin-only server?
18:55 < ntk> so what's the next order of business now, hardware vendors for our new server? do we have an agenda?
18:55 < ntk> ah
18:56 < unknown_lamer> I recommend penguin computing
18:56 < leitgebj> unknown_lamer, can you stay on a topic for a minute?
18:56 < Smerdyakov> No, that is not the next order of business, and let's please hold off on talking about it yet..
18:56 < Smerdyakov> We have not yet decided on what hardware (in a general sense) to colocate.
18:56 < unknown_lamer> ok
18:56 < Smerdyakov> I want to be able to say a configuration and verify that no one disagrees. :)
18:57 < docelic> 3 machines, a switch, and a console ?
18:57 < Smerdyakov> docelic, there are also suggestions of 4 machines + same.
18:57 < leitgebj> Smerdyakov, I don't have a strong opinion on this matter.  I think that there are lots of options, but the poweredge should be able to serve as the preliminary fileserver for everything for a while without breaking a sweat.  This gives us time to save for a really nice NAS box, or whatever it might be, later.
18:58 < Smerdyakov> leitgebj, and you favor a 4 server set-up?
18:58 < ntk> i think we should definitely start with 3 servers
18:58 < unknown_lamer> I like the 3 machines idea; 4 servers is a bit overkill for the coop now I think
18:58 < leitgebj> I think three is fine to start.
18:58 < ntk> we need to control our investment costs at this point
18:58 < ntk> yes
18:58 < vol> sounds like everyone agrees on 3 server
18:58 < ntk> it should be possible to set things up to utilize a 4th server well when the time comes
18:59 < leitgebj> I think that we can roll by with two servers for a while, if one is the poweredge.  It's not ideal, but nothing ever will be.  It's just about sustained, incremental improvement, and I think that all of our designs we're talking about now meet that goal.
18:59 < Smerdyakov> The main issue here is that a significant proportion of members use their accounts for purposes unrelated to hosting.
18:59 < Smerdyakov> And I _really_ want to keep that stuff isolated from everything else.
18:59 < ntk> and i think that should go on Abu
18:59 < ntk> as long as it shall live
18:59 < vol> ntk: i second that
18:59 < Smerdyakov> ntk, it can't do that when it's at Xiolink.
19:00 < Smerdyakov> ntk, so you are assuming that we move it to Peer1.
19:00 < unknown_lamer> Abu should probably have its disks replaced
19:00 < ntk> then let's contact Xiolink yesterday and tell them to fedex it to me
19:00 < ntk> yes
19:00 < ntk> i am 
19:00 < Smerdyakov> Which means merging fileserver and non-dynamic-content services.
19:00 < ntk> i can definitely put new disks in Abu.  i am not an imbecile
19:00 < Smerdyakov> Does anyone disagree with that as the best next move?
19:00 < leitgebj> Smerdyakov, if that is an issue, how about two "real" servers, the poweredge and a front-end web server, and abulafia for "unrelated" purposes.  Abu could also be the shell box that users pushes to the front-end web.  It still wouldn't be a mission-critical situation in this case.  It also gives us time to ship abu and do whatever repairs are necessary.
19:00 < vol> Smerdyakov: i agree
19:01 < ntk> leitgebj: how is that different from what i said earlier?
19:01 < Smerdyakov> leitgebj, it's probably much more convenient to debug web services on the same machine were they run.
19:01 < unknown_lamer> Smerdyakov: I think it would be best to have a dedicated static server, but if the coop cannot afford a dedicated static server then merging the afs master and the static web server would work
19:01 < Smerdyakov> unknown_lamer, I'm pretty sure we can afford it, which is why it's unclear why people are so quickly dismissing the idea.
19:02 < Smerdyakov> unknown_lamer, I thought we determined the _repeating_ cost is zero from that choice.
19:02 < unknown_lamer> Yes
19:02 < ntk> Smerdyakov: you are proposing a 4-server option?
19:02 < leitgebj> Smerdyakov, maybe a bit.  But we could have a development web server on abu.  Also, the log files would be available from abu.  I know of lots of places that have this kind of separated development/production environment and it works well.  I think that users specifically shouldn't have shell access to the production web server any more than necessary.
19:02 < Smerdyakov> ntk, I am questioning why we would pick the 3-server if 4-server has no additional long-term costs.
19:02  * unknown_lamer misrea what the 3 server option was ,oops
19:02 < ntk> Smerdyakov: because it is easy to migrate to 4 servers
19:02 < ntk> and because we have limited cash pool
19:02 < docelic> I second ntk
19:03 < leitgebj> third
19:03 < docelic> and setting 4 servers at once is a lot of work
19:03 < Smerdyakov> ntk, so if I donated the extra server, you'd say the same thing?
19:03 < ntk> even if you want to float us a large interest-free loan, then it would still have to be paid, raising member costs
19:03 < leitgebj> When the time comes, our next move should be to have multiple web servers behind a load balancer to reduce down-time.
19:03 < Smerdyakov> ntk, nah, I said _donate_.
19:03 < ntk> no, if you donated a second server than as a boardmember i would say 4 servers for sure
19:03 < ntk> hell, donate a whole rack :-)
19:03 < ntk> the IRS will seriously question our business model at that point
19:04 < Smerdyakov> OK, so the only reason not to go with 4 servers is the cost of the extra server..
19:04 < Smerdyakov> (The one-time cost, I mean)
19:04 < ntk> yes, the initial costs over the next 4-6 months or so
19:04 < vol> Smerdyakov: thats how i see it
19:05 < Smerdyakov> OK, so perhaps we go with 3-server for now with plans to go to 4 as soon as the cost per member is low enough?
19:05 < Smerdyakov> (And make sure that the space, power, etc., we go with from the start can accommodate this expansion.)
19:05 < ntk> yes, let's consider that done and proceed. a 4th server could easily be configured as either dedicated file or dedicated static web
19:06 < ntk> the 10U peer1 could easily accomodate 4 such servers & switch
19:06 < Smerdyakov> Any objections to this?
19:06 < leitgebj> Sounds good.  But I would like to really try to make the next move to a load-balanced web server configuration which would allow us to take one entire web server off-line, while the rest of our servers stay up.
19:06 < ntk> speak now or forever hold your peace
19:06 < leitgebj> That can be discussed later, however.
19:06 -!- iriefrank [n=frank@ppp-71-135-250-63.dsl.irvnca.pacbell.net] has left #hcoop []
19:07 < Smerdyakov> OK, it sounds like we have made a Decision.
19:07 < docelic> Yes.
19:07 < Smerdyakov> And for the rest of the meeting, we focus only on this starting configuration.
19:07 < ntk> yes x3
19:08 < Smerdyakov> So now we turn to purchase issues for the dynamic-content server?
19:08 < ntk> why not?  Vendor bidding time./
19:08 < Smerdyakov> (Since it seems we agree to leitgebj's donation as fileserver and Abu as goof-off server.)
19:08 < vol> i second the penguin computing unknown_lamer suggested
19:08 < leitgebj> I liked unkown_lamer's suggestion of penguin computing as well.
19:09 < Smerdyakov> We shouldn't make any decisions on vendors before contacting them about discounts for our uniquely cool organization. :)
19:09 < ntk> yes
19:09 < ntk> i think we should have another meeting schedule in maybe 2 weeks to finalize purchasing
19:09 < ntk> for hardware
19:09 < docelic> No, but we can select prospective ones to offer them to offer us their package :)
19:09 < leitgebj> Right, and also we should talk to them on the phone.  Dell, at least is extremely negotiable, for whatever that's worth.
19:09 < unknown_lamer> IIRC penguin is friendly towards free software and other such related things
19:09 < unknown_lamer> as in "here are servers for less money because you are cool"
19:10 < Smerdyakov> Out of curiosity, since I'm a hardware anti-expert, how much less awesome is leitgebj's donation than the server we would otherwise choose to purchase today for the same purpose?
19:10 < ntk> clinton: that is nice, but also what hardware to they offer for what price?
19:10 < ntk> have you researched this?
19:10 < unknown_lamer> ntk: yes, and so to leitgebj (wiki wiki)
19:10 < unknown_lamer> at least standard prices
19:10 < ntk> Smerdyakov: i would be leaning towards a 2-server configuration w/o a donation
19:10 < docelic> Smerdyakov: I dont think it's any less awesome, maybe even on the contrary
19:10 < unknown_lamer> They cost almost half as much as the equiv dell machines...
19:10 < docelic> It's big, and disk rich
19:11 < ntk> disk-rich is good and means that it can be our fileserver, freeing us to buy a machine that is cooler in nondisk waysw
19:11 < Smerdyakov> docelic, OK, if you the expert wants to declare that, then that's all I need to read. :)
19:11 < docelic> I'd better see it's processing power and a lot of ram used
19:12 < docelic> since disks are not new
19:12 < Smerdyakov> leitgebj, did you get any info on costs for Dell to support that machine?
19:12 < ntk> switching gears for a moment, would we want to buy new disks for leitgebj's machine?  it has many but they are not huge
19:13 < Smerdyakov> If that somehow ends up being above costs for a new machine (that would be weird!), then we might reconsider.
19:13 < ntk> 2nd hand scsi disks are cheap and they last a long time, but that would defeat the purpose
19:13 < leitgebj> Not yet.  I will call on Monday.  Currently we know that most of it is under warranty on one-day support for a couple of years or so.  I will check on the bronze-level support package and email the list with details.
19:13 < docelic> I;m not so sure second-hand SCSIs are cheap
19:13 < leitgebj> I am fairly sure that the Dell bronze support package is relatively low, and sufficient for our needs.
19:14 < leitgebj> s/low/cheap
19:14 < ntk> Smerdyakov: is that not true of any vendor?  support ain't cheap
19:14 < ntk> if we're not supporting ourselves
19:14 < docelic> Smerdyakov: could you give me an insight on what else's left to discuss today? 
19:14 < leitgebj> Ebay has 146 GB scsi disks for around $200/each.
19:15 < Smerdyakov> ntk, if the machine is old and they've stopped doing something they need to do to be prepared to support them...
19:15 < docelic> Yes. I was thinking about >200 Gb diska, and $200 isn't cheap for a used disk
19:15 < Smerdyakov> docelic, buying choices for the one new machine we've identified... I don't know, anything else?
19:15 < docelic> Smerdyakov: buying, as in specs, or vendor ?
19:15 < Smerdyakov> docelic, both
19:15 < ntk> what do you mean by hardware "support"?  all i can think of is fixing something that breaks, which is a warranty....
19:16 < Smerdyakov> ntk, guarantees of doing that within 24 hours, on site.
19:16 < docelic> Smerdyakov: I think we should give a week for this to sort out on the Wiki. We can't decide on any of that now, or today. It's too ad-hoc.
19:16 < leitgebj> ntk, dell has a scaling support system which gives you access to on-site support, and also to more knowledgeable engineers on the phone when you have a problem.
19:16 < Smerdyakov> docelic, that's a concrete action within that discussion topic.:P
19:16 < ntk> why the hell would we want to do that? we already can get remote hands from peer1 with much better than 24 hour support, and most likely i or another member can  be onsite within 24 hours anyway
19:16 < ntk> that just sounds like overpriced insurance, which would cost an arm and a leg from any vendor
19:17 < leitgebj> I propose that we find out prices before declaring anything overpriced.  I know companies that find the bronze package worthwhile, even considering the fact that they have physical access to two cages.
19:17 < ntk> ok
19:17 < unknown_lamer> leitgebj: why pay more for dell hardware *and then* buy support when we could just buy hardware that wasn't going to break in the first place *for less money*
19:17 < jts> leitgebj: What SCSI, 68 or 80?
19:17 < docelic> Folks, I would leave now. Do we agree Smerdyakov  would create 2 wiki pages, one for potential vendors, and one for a definition of what we want in those 2 new machines (hardware-wise)?
19:17 < Smerdyakov> unknown_lamer, don't be ridiculous.  Anything can break.
19:18 < unknown_lamer> dells are more likely to break
19:18 < leitgebj> jts, good question... 80?  I'd have to check the docs...
19:18 < docelic> I tend to agree with unknown_lamer 
19:18 < unknown_lamer> when your claim to fame is building cheap things
19:18 < unknown_lamer> "we sell cheap shit!"
19:18 < ntk> well keep in mind that we'll have a locked cage, so for it to be meaningful we would have to give dell copies of the keys somehow.  it makes little sense to have dell come in 24 hours and still have to have a member onsite, when that member could have just bought the hardware and replaced it himself cheaper anyway
19:18 < Smerdyakov> OK, we need to organize this conversation if we're going to draw to a close.
19:18 < docelic> But I don't think it matters in our case.. it'll last long enough
19:18 < jts> leitgebj: That's the # of pins on the cable. 68 pin or 80 pin
19:18 < ntk> yes
19:18 < Smerdyakov> How about everyone has to ask me permission to speak until the meeting is over?
19:18 < ntk> okay, +m
19:18  * docelic agrees
19:19 < ntk> mr. president, preside!
19:19 < jts> lol
19:19  * unknown_lamer goes to make lunch anyway
19:19 < Smerdyakov> It seems the issues remaining for the meeting involve choosing the hardware specs of the 1/3 new server and the vendor.
19:19 < Smerdyakov> Does anyone propose doing anything besides opening these up for wiki suggestions?
19:20 < Smerdyakov> We should decide on who will contact particular vendors for discount info.
19:20 < ntk_logger> no.
19:20 < docelic> I volunteer to make a wiki page. List the two machines (hardware specs), and suggest uses. Then what's left is see what would be new machine be used for, and come up with an appropriate setup.
19:20 < Smerdyakov> Should we contact vendors first, or wait until we know exactly what on their product lists we're interested in?
19:21 < docelic> I think we should come up with a plan for our ideal machine, then see how that fits their predefined packages
19:21 < jts> Adam, I'm going for lunch, I will email my idea over to you.
19:22 < Smerdyakov> docelic, we already have NewSystemHardware page... perhaps just refactor that?
19:22 < docelic> Smerdyakov:  Ok
19:22 < ntk> ok
19:22 < ntk> let's do that.
19:22 < Smerdyakov> OK, it sounds like we will continue the process of accumulating ideas on NewSystemHardware.
19:22 < Smerdyakov> At some point in the next week, we should appoint people to contact interesting vendors for discount info, once we know what of theirs we're interested in.
19:23 < ntk> ok
19:23 < Smerdyakov> Now I'm going to list what I think are _all_ of the action items coming out of this meeting.
19:23 < ntk> can we do that without an official meeting?
19:23 < ntk> it would be nice
19:23 < docelic> Probably through a wiki, yes. 
19:23 < ntk> go ahead
19:23 < docelic> I think we have a list of people who said they'd help
19:23 < Smerdyakov> 1. docelic will update NewSystemHardware to reflect our current plans, including two already available servers.
19:23 < docelic> they put their names on the Wiki.
19:24 < Smerdyakov> 2. I will announce the results on hcoop-announce and ask people to update NewSystemHardware with any opinions.
19:24 < Smerdyakov> 3. About a week from today, we will have our list of candidates, and we will, over mailing lists, appoint people to contact the vendors for pricing info.
19:25 -!- Optikal_ [n=optikal@pool-70-17-114-141.res.east.verizon.net] has joined #hcoop
19:25 < Smerdyakov> 4. leitgebj will let us know about costs for Dell support contract options.
19:25 < jts> Going now. Adam, you have mail. 
19:25 < Smerdyakov> 5. unknown_lamer will let us know about costs for this new maverick colo company.
19:25 < jts> Have a good morning/afternoon/day/evening/night all. 
19:25 < leitgebj> bye jts
19:25 < Smerdyakov> Anything I've missed?  Anything anyone is expected to d?
19:25 < Smerdyakov> do
19:25 < ntk> i object to 5.
19:25 < jts> Smerdyakov: Did you want me to check into the local colo near me or no?
19:25 < Smerdyakov> jts, information is always useful.
19:25 < docelic> Smerdyakov: I agree on all. It's what we've agreed on, looks ok.
19:26 < ntk> i think we should limit our options to he.net or peer1
19:26 < ntk> if this is going to happen soon.
19:26 < jts> bbl
19:26 < docelic> ntk: it doesnt matter at the moment, let's let unknown_lamer come up with something
19:26 < Smerdyakov> And we don't want to schedule the next meeting now?
19:26 < ntk> it does matter, because i would like abulafia taken offline ASAP
19:26 < ntk> and it either needs to be shipped to CA or me, depending on what host we are going with
19:26 < ntk> but i guess that can wait for the next meeting
19:27 < docelic> Smerdyakov: probably 2 weeks from now? 1 week to appoint people, and one week to mail them and discuss replies?
19:27 < Smerdyakov> Any objections to next meeting 2 weeks from now?
19:27 < ntk> i agree.  2 weeks it is.
19:27  * docelic agrees on +2 weeks, same time.
19:27 < Smerdyakov> Then:
19:27 < Smerdyakov> 6. The next IRC meeting will be exactly 2 weeks from the start of this one.
19:27 < Smerdyakov> Actually, let me think about my schedule.
19:28 < Smerdyakov> I'm on vacation starting that day.. no go.
19:28 < ntk> grrr.
19:28 < Smerdyakov> You could meet without me.  *shrug*
19:28 < ntk> let's not schedule it for a time when we know one of the boardmembers can't be present
19:28 < docelic> We could. But I suppose we wouldn't need a meeting then, ntk and me.
19:28 < Smerdyakov> Or we could leave next meeting indeterminate.
19:28 < ntk> how about friday night then?
19:29 < ntk> 13 days?
19:29 < docelic> Why not discuss things as new things happen, over the next two weeks
19:29 < Smerdyakov> ntk, conference submission deadline. :)
19:29 < docelic> So we don't have to concentrate everything in a single day ?
19:29 < ntk> 15 days?
19:29 < Smerdyakov> 7/[15-20] is the vacation I'm talking about.
19:29 < docelic> I vote for leaving next meeting undetermined then.
19:30 < Smerdyakov> So do I. :)
19:30 < ntk> it would be best to schedule it now since we're all here, otherwise definitely at least 1 week prior.
19:30 < ntk> alright
19:30 < Smerdyakov> OK.  Meeting over?
19:30 < docelic> Meeting over.
19:30 < leitgebj> Have we made any concrete decisions about the membership pricing structure?
19:30 < ntk> we will meet in an indeterminate amount of time.
19:30 < ntk> are we done ?
19:30 < ntk> i move to adjourn
19:30 < Smerdyakov> leitgebj, no.  Should we wait until we know costs of hardware?
19:30 < ntk> no, other than to have a sliding scale or multiple plans
19:30 -!- docelic [i=docelic@ri02-016.dialin.iskon.hr] has quit ["Bye"]
19:31 < ntk> i do think that we should hold off
19:31 < leitgebj> Sure.
19:31 < Smerdyakov> OK, we're done.