welcome: please sign in

The following 1433 words could not be found in the dictionary of 7 words (including 7 LocalSpellingWords) and are highlighted below:
24h   2x   2xserver   32bit   4pm   4x   7200rpm   A66   A7   able   about   above   Abu   abu   access   accessible   act   Action   action   actually   adam   Adam   adapter   add   adding   Admin   admin   admining   administration   administrative   administrators   admins   advantage   advice   affect   after   After   afterward   again   agree   Agreed   agreed   agreeing   agrees   ah   alias   all   allow   allowed   along   already   also   always   am   amended   amongst   amount   an   and   And   announcement   announcing   another   answer   answered   answering   answers   anti   any   Any   anymore   Anyone   anyone   Anything   anything   anyway   appointed   appointing   Appraisal   appreciate   appropriate   architecture   Architecture   Are   are   aren   around   arranged   arrive   as   As   ask   asked   asking   asp   aspects   associated   assume   assumed   assuming   assumption   Asus   at   Athlon   attendance   attractive   authentication   author   automated   automatic   availability   available   average   avoid   away   back   backup   bad   bah   balancer   balancers   bandwidth   based   basic   be   because   become   been   before   behind   being   belongs   besides   best   better   Between   between   big   billion   bit   board   book   bootup   bought   bound   bounds   box   bpt   brains   branch   breaks   Brian   bring   broke   budget   build   building   built   But   but   buy   buying   buys   by   bylaws   Bynum   C2100   C311   C313   C811   ca   cabinet   cable   cache   California   came   can   Can   Canada   candidate   Cape   card   case   cat   catch   category   centralized   certain   certainly   chance   changing   charge   Chat   chatter   cheaper   check   Chlipala   choice   choices   choose   choosing   chose   Cisco   clean   Clinton   closed   closest   co   Cod   collaboration   collection   college   colo   colocate   Colocation   com   combined   combining   come   comes   coming   commercial   commercials   commit   commitment   complete   complicate   complicated   compute   Computing   concerned   concerns   concludes   conclusion   conclusions   concrete   condensed   confident   configuration   configurations   configure   configured   confused   connect   connection   Connection   consequences   consider   consideration   considered   considering   console   contained   content   continue   contract   contracts   conversation   conversion   Cool   cooling   coop   Coop   corresponded   cost   could   couldn   count   counterbalancing   course   cover   crappy   create   created   creating   cred   criteria   curious   current   Currently   custom   cyclades   damage   darnit   data   database   date   Davor   day   days   deactivate   deal   Debug   Debugger   decide   decided   Decided   deciding   decision   Decisions   decisions   declaration   declare   declaring   dedicated   definite   definitely   degrades   delete   deliberation   dell   Dell   Depends   designed   detail   determined   did   Did   didn   dies   difference   different   discuss   discussed   discussion   discussions   disk   distance   distracting   distributed   dmisinetworks   do   Do   docelic   Does   does   doesn   doing   dollars   domtool   don   donations   dont   double   down   downed   drill   drops   dsl   dual   dumb   dump   dunno   duration   Duron   dynamic   E16833118021   E16833122111   each   earlier   easier   east   easy   edict   effectively   effort   eh   either   Electric   elements   eliminate   else   email   emergency   employer   encountered   encouraged   end   ends   enough   ensure   entire   environment   equip   Erlang   error   Escalade   especially   estimate   estimating   etc   Evaluation   even   event   Eventually   Everyone   everyone   everything   exactly   except   existing   expansion   expect   expense   expenses   experience   experiences   explains   explicit   expressed   extent   extra   extreme   extremely   F12   factor   failed   failure   failures   fan   far   feasible   feature   feel   figure   figured   file   files   fileserver   Filtering   finalize   find   fine   firewall   first   fixed   fixedtime   fo   folks   follow   following   for   Force   forgot   forking   forms   forward   four   Frank   free   Fremont   friend   friends   from   front   full   fully   fundamental   fur   future   fyodor   Gates   geek   generic   geographically   get   gets   getting   gigabit   Gigabyte   give   given   gives   go   goals   goes   going   good   got   group   groups   grow   guaranteed   guarantees   guess   guessing   guys   had   half   Hallgren   hand   handle   handling   handy   hanging   happen   happened   happens   hard   Hardware   hardware   has   Hat   have   haven   Having   having   hcoop   He   he   head   healthy   heared   hearted   heartedly   heatsink   hello   help   here   Here   Hey   Hi   hi   high   himself   his   History   hit   home   homework   hope   hopefully   Hopefully   Hopper   host   hosting   hotshot   hour   hours   house   houses   how   How   However   html   http   https   hundreds   Hurricane   Hz   id   idea   identified   if   If   ignorant   Im   imagine   implied   imply   important   in   In   Incidentally   inclined   include   increase   inexpensive   information   informed   infrastructure   inheritance   initial   inquiries   inside   install   installation   instance   instead   Intel   intended   intention   interconnect   interest   interested   interesting   interfere   interim   Internet   interruption   into   investing   investment   involve   involved   iriefrank   is   Is   isn   isntall   issue   issues   it   It   item   Item   Items   items   Its   its   Jag   jch5   job   John   join   joined   jpi   July   jump   Jun   just   Just   Justin   keep   keeping   Kennedy   Kerberos   kerberos   key   kickstart   killed   kind   Kingston   know   knowledgeable   known   L2   lamer   laptop   last   late   Later   later   lay   ldap   lead   least   leave   leaves   Leaving   left   Leitgeb   leitgebj   less   Let   let   level   life   lifetime   like   Like   limit   limited   lines   link   Link   linux   lisr   list   listed   lists   literate   little   ll   load   loan   loans   local   located   location   locations   locked   log   Log   logical   login   logins   lol   long   look   looking   Looks   lost   lot   lots   Lots   love   low   lower   lowest   lowly   luck   lurking   M266   machine   machines   made   magically   mail   mailed   mailing   mailserver   main   mainframe   maintainer   maintaining   maintenance   majority   make   makes   making   manage   managed   management   manufacturer   many   matter   may   Maybe   maybe   mc   me   mean   meant   meet   meeting   meetings   member   members   membership   mental   mentioned   meta   metaperl   Michael   Micro   might   Mike   min   mind   minute   minutes   mirrored   missed   Mmmm   mo   mode   models   modemcable135   modemcable169   modern   modifying   moment   money   month   montly   more   More   most   Most   mostly   motherboard   mountable   move   moving   Mr   much   multiple   must   mwolson   My   my   N266   N82   nah   naked   name   names   Nathan   natural   nature   ncie   near   nebulous   need   needs   negligible   neighbor   net   network   networks   never   new   newegg   newer   next   Next   nic   nice   nics   nitpicky   no   No   nominate   nomination   non   not   Not   note   Nothing   nothing   notify   novice   now   Now   ntk   number   object   objection   objections   obtain   obvious   Ocelic   of   off   office   official   officially   often   oh   Oh   ok   old   Olson   on   On   onboard   once   one   One   ongoing   only   oops   op   open   operate   operation   opinion   opinions   Optikal   optikal   option   options   or   Or   order   ordinary   Other   other   Our   our   ourselves   out   outgrown   outrageous   outweigh   over   Over   overbuy   own   p1   page   paid   pain   part   particular   parts   passes   password   past   pasted   path   pay   paying   payment   payments   peer   pending   Penguin   people   per   performance   Perhaps   perhaps   period   permitting   person   personnaly   pgmr   phase   physical   pick   pictures   piece   pin   pings   place   places   plan   planned   planning   Plans   plans   played   please   plug   plugged   plus   pocket   point   pointless   points   Policies   poll   pondering   pool   popular   portal   Portal   porting   ports   position   possibilities   possible   post   potential   pre   predictable   pref   prefer   Preferences   preloaded   presence   present   pretty   previous   price   prices   primary   private   Probably   probably   problem   problems   procedures   process   processors   Product   production   products   profile   programs   prohibitive   prompt   proposed   protocol   protocols   provide   provider   providers   Providers   provides   prt   public   purchase   purchasing   purpose   push   quality   quebec   question   questioned   questions   quick   quit   quite   quote   quotes   rack   rackmount   raise   rare   rated   rates   rather   re   read   Read   Ready   ready   real   really   reason   reasonable   reasons   receive   recommended   reconvene   recorded   recover   Red   reduce   redundancy   redundant   referred   referring   reflect   reilly   reiterate   related   relation   relative   reliability   reliable   relilable   reload   relocating   rely   remember   remote   Remote   remotely   rent   replace   replacement   report   request   requests   require   required   res   research   Researching   reset   response   responsibilities   restart   results   retain   return   returns   ride   right   Right   riser   risk   robust   role   room   root   rooted   roster   round   run   running   said   sake   same   Saturday   save   saver   say   saying   scale   scarcity   scary   scenes   scheduled   schema   scratch   scratches   screw   scrotum   sec   second   security   see   seem   seems   seen   self   sell   send   sense   sensible   sent   separate   Serial   serial   serious   seriously   server   servers   service   services   set   sets   setting   settled   setup   sheet   shell   shifted   ship   shipment   shipped   shipping   shit   short   should   Shouldn   shouldn   show   shrug   sigh   similar   Since   single   site   sites   sitting   situation   size   sliding   slightly   slim   slot   slots   slows   small   smart   Smerdyakov   smoothly   so   So   social   software   solid   some   Some   somehow   someone   something   somewhat   somewhere   sort   sounds   Sounds   space   spare   Spare   spec   specifications   specs   spectator   speed   spend   spending   spent   split   stable   stage   standard   star   start   stated   static   still   structure   stuff   subject   subscribe   subscribed   subscribing   subscriptions   suffices   sufficient   suggest   suggested   suggesting   suggestions   suggests   Summary   summary   Super   supply   Support   support   supported   supporting   suppose   sure   Sure   surprised   surrounding   switch   switched   switches   switching   sys   sysadmin   sysadminesque   sysadmins   sysasmin   System   system   systems   systens   take   taken   taking   talk   talking   target   tarsin   task   tasks   team   techie   technical   Technically   tell   Templeton   temporarily   tenative   tend   term   terms   testing   than   thanks   that   That   the   The   their   them   then   theoretically   There   there   therefore   these   they   They   thing   things   think   thinking   third   this   those   though   thought   thoughts   three   threshold   through   throughout   throw   Thunderbird   time   Time   timeanddate   timed   times   to   today   together   too   top   topic   topics   Toshiba   total   traffic   transfer   transitioning   tricky   trigger   trivial   trump   try   trying   turn   two   ulimits   unable   unaffordable   undeniable   understand   underused   Unfortunately   unknown   unless   unneeded   unstable   until   up   update   upgrade   upload   ups   uptime   us   use   used   useful   user   users   using   Using   usual   usually   usuals   vacation   value   ve   vendor   verizon   very   Very   via   vid   videotron   Vidia   visible   voider   volunteer   volunteered   Volunteers   volunteers   vote   voted   vs   wait   want   wants   Ware   warranties   warranty   was   wasn   waste   watching   way   ways   We   we   web   week   well   Well   were   what   What   whatever   when   where   Which   which   while   who   Whoever   whole   whom   whose   Whose   why   Why   wiki   wild   will   willing   winner   with   With   within   without   wlafyte   won   wonder   wondering   work   working   works   world   worldclock   worse   Worst   worth   would   Would   wouldn   writing   wrong   www   yeah   year   years   yep   Yep   yes   Yes   yet   yield   yielding   You   you   young   your   yourself   zone  

Clear message
Edit

IrcMeetings / 20060624

1. Summary

We discussed plans related to setting up our new, more robust hosting set-up.

Unfortunately, board member NathanKennedy didn't show up, so we couldn't declare it an official board meeting.

2. Decisions

3. Action Items

4. IRC Log

AdamChlipala created this log with X-Chat. 

Jun 24 11:20:47 <Smerdyakov>    OK, the appointed time is here.
Jun 24 11:21:01 <Smerdyakov>    How about we start with an official announcement of presence from everyone who's here and watching?
Jun 24 11:21:03 <Smerdyakov>    Hi!
Jun 24 11:21:15 <iriefrank>     no its not healthy
Jun 24 11:21:35 <Optikal_>      I am here.
Jun 24 11:21:41 <iriefrank>     Frank Bynum here
Jun 24 11:21:46 <leitgebj>      Hi all.
Jun 24 11:21:47 <jch5>  this is John Hallgren from Cape Cod MA
Jun 24 11:21:48 <Smerdyakov>    And deactivate off-topic conversation for the duration. :)
Jun 24 11:22:20 *       Smerdyakov pings docelic.
Jun 24 11:22:30 <bpt>   i am here
Jun 24 11:22:51 >docelic_<      Ready to start?
Jun 24 11:22:59 <docelic_>      Yep we can start
Jun 24 11:23:20 <Smerdyakov>    OK.  I think the first thing is to decide officially who is in charge of being in charge of things. ;)
Jun 24 11:23:21 <docelic_>      (Davor Ocelic here)
Jun 24 11:23:44 <bpt>   <- Brian Templeton
Jun 24 11:23:44 <Smerdyakov>    ntk isn't here, so this doesn't count as a board meeting per our bylaws, but docelic and I can still make official decisions by majority vote.
Jun 24 11:24:12 <Smerdyakov>    We have our admin volunteers listed here: http://wiki.hcoop.net/wiki/AdminVolunteers
Jun 24 11:24:44 <Smerdyakov>    ntk volunteered himself half-heartedly in case we didn't find enough folks.  What do we think?  Is 3 sysadmins a good number, or do we want 4?
Jun 24 11:25:52 <jch5>  from a lowly user POV, if 4 is possible, I'm appreciate it...to cover all time slots and backup/vacation
Jun 24 11:25:53 <docelic_>      Do we want to make a complete decision now, or a decision of "3 now, with one slot open" is good enough?
Jun 24 11:25:54 <Smerdyakov>    We're making do now with, effectively, 2, where one (me) doesn't really want to be doing day-to-day sysadmin stuff.
Jun 24 11:26:25 <leitgebj>      I think that if it was a half-hearted self-nomination, we are OK with 3 until someone comes along who is interested.  ntk is doing a lot with the co-op without taking on more responsibilities, unless he really wants them.  I am also confident that someone will come along in short time who is willing to do the job.
Jun 24 11:26:29 <Smerdyakov>    jch5, the counterbalancing issue is that more people with root access degrades security.
Jun 24 11:26:53 <iriefrank>     three is fine with me, i dont think we need an official declaration of how many sysasmin "slots" there are either
Jun 24 11:27:06 <docelic_>      I'm a little surprised we couldn't come up with 4. There's been a number of people who expressed their interest over time :(
Jun 24 11:27:07 <Smerdyakov>    Incidentally, I'm also suggesting that I retain root access as the person in charge of software infrastructure... AKA maintaining domtool and related stuff.
Jun 24 11:27:25 <Smerdyakov>    So that would be 4 with docelic, leitgebj, mwolson, and me.
Jun 24 11:27:46 <leitgebj>      Smerdyakov, I think that sounds like a solid plan.
Jun 24 11:27:54 <docelic_>      Sounds good.
Jun 24 11:27:55 <Smerdyakov>    iriefrank, there are administrative things like whom to make sure is present at meetings about sysadmin stuff. :)
Jun 24 11:27:58 <jch5>  10-4
Jun 24 11:28:34 <Smerdyakov>    OK, so it sounds like for now we'll assume the roster of sysadminesque people I just stated.
Jun 24 11:29:19 <Smerdyakov>    Now I think we should turn to choosing our next colo provider: http://wiki.hcoop.net/wiki/ColocationProvidersEvaluation
Jun 24 11:29:41 <Smerdyakov>    HE seems to have the most attractive deal at present.  Any objections to taking them as our intended provider going forward?
Jun 24 11:30:09 <leitgebj>      I also think that he.net is the best option.
Jun 24 11:30:30 <leitgebj>      docelic, how do you feel?
Jun 24 11:31:45 <docelic_>      I wasn't much into data collection for this round
Jun 24 11:31:52 <docelic_>      I've read all there was on the subject earlier today
Jun 24 11:31:59 <Smerdyakov>    docelic, you can see on their page that they have the lowest price. ;)
Jun 24 11:32:17 <docelic_>      and I heared of HE.net before. We considered them in the previous round too, or I heared of them from some other place ?
Jun 24 11:32:52 <Smerdyakov>    I don't know.
Jun 24 11:33:48 <Smerdyakov>    OK... I don't see a lot of deliberation.  docelic, are you pondering this now, or are you just yielding to our conclusion? :)
Jun 24 11:34:16 <docelic_>      I was just doing inquiries, but I agree with going on with HE
Jun 24 11:34:33 <docelic_>      My point was that, if it wasn't in our previous round, then I heared good things about HE from other places as well
Jun 24 11:35:08 <Smerdyakov>    OK.  That leaves us to figure out exactly what we want to host there.
Jun 24 11:35:36 <Smerdyakov>    Whose natural home is http://wiki.hcoop.net/wiki/SystemArchitecturePlans
Jun 24 11:35:52 <docelic_>      Are we looking to move abu and fyodor there ?
Jun 24 11:36:02 *       voider (n=Is@modemcable169.254-80-70.mc.videotron.ca) has joined #hcoop
Jun 24 11:36:16 <Smerdyakov>    docelic, we don't own fyodor, so I think that's a "no" there.
Jun 24 11:36:26 <Smerdyakov>    docelic, Abu is something we haven't decided on yet.
Jun 24 11:36:29 <docelic_>      Well I mean in a software sense
Jun 24 11:36:35 <voider>        hi guys
Jun 24 11:36:36 <Smerdyakov>    docelic, eh?
Jun 24 11:36:40 <docelic_>      hello voider 
Jun 24 11:36:40 <voider>        i am late?
Jun 24 11:36:56 <voider>        *am i*
Jun 24 11:37:06 <docelic_>      voider: not much, we agreed on admin structure (leitgebj, mwolson, adam and me), and to go with HE.net for next provider
Jun 24 11:37:06 <Smerdyakov>    voider, not so late.  We decided on who the sysadmins would be, decided to go with HE as our next colo provider, and now we discuss system architecture to colocate.
Jun 24 11:37:26 <docelic_>      Smerdyakov: I didn't mean "moving" in sense of relocating hardware, but services
Jun 24 11:37:29 <voider>        ok good
Jun 24 11:37:49 <metaperl>      what about Mike_L as admin?
Jun 24 11:37:50 <Smerdyakov>    docelic, it has been my assumption the whole time that all of our primary services would move to the new servers.
Jun 24 11:37:58 <Smerdyakov>    metaperl, he didn't add himself to the candidate list.
Jun 24 11:38:09 <voider>        i'll mostly act as a spectator, I don't know enough about the way you operate hcoop yet
Jun 24 11:38:15 <Smerdyakov>    metaperl, and I think he doesn't have much interest in the level of time commitment that is appropriate.
Jun 24 11:38:15 <docelic_>      metaperl: just like Clinton didn't 
Jun 24 11:38:27 <docelic_>      Smerdyakov: ok then
Jun 24 11:39:05 <Smerdyakov>    The SAP wiki page lists three servers: file server, member accessible server (the scary zone!), locked-down server (where we keep as much as we can).
Jun 24 11:39:35 <Smerdyakov>    I think of all of these as limited to "Internet hosting" stuff only.
Jun 24 11:39:54 <Smerdyakov>    We could also consider (space permitting) to move Abu to the new colo place and make that a generic "shell server" for whatever.
Jun 24 11:40:06 <docelic_>      ok
Jun 24 11:40:07 <Smerdyakov>    We could even buy a new server for that, but this seems a good use for something we have around anyway.
Jun 24 11:40:31 <Smerdyakov>    But I consider that something to do after the main stuff is up and running smoothly.
Jun 24 11:41:14 <Smerdyakov>    Does anyone object to the very high-level proposed configuration for the main 3 servers?
Jun 24 11:41:28 <Smerdyakov>    Or have any suggestions at that level before we drill down..
Jun 24 11:42:32 <docelic_>      I think it's OK. One thing that wasn't mentioned, and it should have been as it's the same logical level as AFS fileserver, is centralized login DB
Jun 24 11:43:03 *       mwolson (i=mwolson@jpi-wlafyte-213-192.dmisinetworks.net) has joined #hcoop
Jun 24 11:43:06 <docelic_>      Very probably using LDAP
Jun 24 11:43:15 <docelic_>      mwolson: short summary: 
Jun 24 11:43:19 <Smerdyakov>    Yes.  I've always associated that with AFS in my CMU days.  I think it was based on Kerberos; that is, I know Kerberos was used, but I'm not sure if there was something else more fundamental.
Jun 24 11:44:01 <mwolson>       oh ... darnit
Jun 24 11:44:04 <docelic_>      mwolson: admins: leitgebj, mwolson, adam, docelic. Next provider: HE.net. System structure OK as proposed on http://wiki.hcoop.net/wiki/SystemArchitecturePlans
Jun 24 11:44:09 <leitgebj>      Right, I was thinking of Kerberos as well, even though I haven't used it before.  I have used ldap a bit.
Jun 24 11:44:42 <docelic_>      Well, AFS has something like kerberos built-in for authentication, but you can plug it into existing kerberos setup if you have one
Jun 24 11:44:54 <Smerdyakov>    I know Kerberos and AFS also work well with geographically distributed networks... so it could scale well to future set-ups where we branch out into multiple main locations.
Jun 24 11:45:15 <docelic_>      So AFS would imply using kerberos, and somehow by nature of things, kerberos would imply LDAP
Jun 24 11:45:29 <Smerdyakov>    Maybe.  I never used LDAP in my experiences in college.
Jun 24 11:45:45 <Smerdyakov>    That is, I never did something where it was obvious that the LDAP protocol was involved.
Jun 24 11:45:55 <Smerdyakov>    Anything could have happened behind the scenes.
Jun 24 11:46:21 <docelic_>      I've read the O'reilly book on LDAP and played with it last month, in a more serious way than before
Jun 24 11:46:42 <mwolson>       i've set up LDAP before at my LUG
Jun 24 11:47:54 <mwolson>       i'd be willing to do so here, if we decide on it
Jun 24 11:48:19 <Smerdyakov>    I think we should pick a main person in charge of the purchase and set-up of the new systems.
Jun 24 11:48:36 <Smerdyakov>    I anti-nominate me because I do a lot of stuff already, AND I don't know much about this stuff. ;)
Jun 24 11:48:53 <docelic_>      I volunteer for the software setup, as usual
Jun 24 11:49:16 <Smerdyakov>    What about hardware elements?
Jun 24 11:50:02 <Smerdyakov>    *a pin drops*
Jun 24 11:50:08 <leitgebj>      I am in the process of buying a house right now, and that may complicate things.  
Jun 24 11:50:48 <Smerdyakov>    Well, I know I'll screw something up if _I'm_ in charge of it!
Jun 24 11:50:48 <leitgebj>      I am willing to help, but I imagine that what will happen is that we will have the servers shipped to a physical location, plugged into a network, e.g., dsl connection.  then they are configured and sent to the colo provider.
Jun 24 11:51:14 <Smerdyakov>    leitgebj, there are serious issues of just _which_hardware_ to buy, and I think you expressed opinions there in the past.
Jun 24 11:51:17 *       Optikal___ (n=optikal@pool-70-18-225-250.res.east.verizon.net) has joined #hcoop
Jun 24 11:51:37 <leitgebj>      Right, that's another stage of the planning that seems to have not been settled fully.
Jun 24 11:51:44 <leitgebj>      Maybe we should start there.  
Jun 24 11:51:55 <leitgebj>      How much money do we have, or have access to?
Jun 24 11:51:58 <Smerdyakov>    That's what I was talking about: appointing someone in charge fo that stage.
Jun 24 11:52:22 <leitgebj>      I volunteer to lead that part of the process.
Jun 24 11:52:51 <Smerdyakov>    Here we can see how much we theoretically will commit to spending per month: https://members.hcoop.net/portal/poll?id=8
Jun 24 11:52:57 <jch5>  i'd just ask to not overbuy and spend extra $$$ unneeded
Jun 24 11:53:18 <voider>        I have a question, dunno if it has been discussed before: how will all this cost? How will it affect the montly payment?
Jun 24 11:53:31 <voider>        ah
Jun 24 11:53:32 <docelic_>      voider: look at the URL Smerdyakov pasted
Jun 24 11:53:36 <Smerdyakov>    I get $317 dollars considering the lower bounds of people who voted.
Jun 24 11:53:58 <Smerdyakov>    voider, we will use sliding scale payments to make sure there is no serious increase for people who don't feel HCoop is worth investing more.
Jun 24 11:54:21 <voider>        shit, I don't have my password here
Jun 24 11:54:29 <voider>        and I forgot it :/
Jun 24 11:54:38 <voider>        I'll check that link when I return back to home
Jun 24 11:55:01 <jch5>  in my case, it's not that it's not worth it...just that I have a very limited budget for my site
Jun 24 11:55:11 <Smerdyakov>    I was somewhat surprised by the scarcity of high-end answers to that poll. :\
Jun 24 11:55:24 <Smerdyakov>    Lots of people not willing to spend as much as they do for cable TV. :P
Jun 24 11:55:56 <docelic_>      We need to figure out new ways to increase hcoop service value
Jun 24 11:56:16 <voider>        personnaly I'm not in a situation that allow me to spend much for my hosting
Jun 24 11:56:23 <Smerdyakov>    leitgebj, there are issues to consider of "do we want to take out loans?" and "do we want to have some loan-like structure arranged amongst ourselves?" and "do we expect a number of members to make one-time donations for hardware that will never be paid back?".
Jun 24 11:56:29 <mwolson>       bah, who would pay money to see mostly commercials ...
Jun 24 11:56:39 <jch5>  I've got $6 a month for all my site expenses...
Jun 24 11:56:49 <Optikal___>    $50 a month is outrageous for web hosting, especially non-commercial
Jun 24 11:57:01 <Smerdyakov>    jch5, that's what's always confused me about your situation.  What's so bad about paying out of your own pocket?
Jun 24 11:57:06 <mwolson>       (referring to cable TV)
Jun 24 11:57:14 <Smerdyakov>    jch5, I pay out of my own pocket for a number of groups whose web sites I host.
Jun 24 11:57:36 <Smerdyakov>    jch5, but that's distracting now and perhaps interesting to bring up later.. :)
Jun 24 11:57:54 <Smerdyakov>    Optikal_, $50 a month is a negligible investment for the average young IT hotshot. :P
Jun 24 11:58:05 <Optikal___>    I don't think of money in relative terms
Jun 24 11:58:13 <jch5>  it's designed to be self supporting by users.. i provide the brains...well most 
Jun 24 11:58:51 <Smerdyakov>    leitgebj, I hope that answered your question. ;)
Jun 24 11:59:28 <voider>        someone can tell the big lines of the above pasted link please?
Jun 24 11:59:29 <leitgebj>      OK, that is fine.  There is something that I've been thinking of in relation to the system architecture, though.  It seems that we are just over the point where one system suffices.  Can we consider buying two solid servers to host at he.net?  That should be more than sufficient for quite a while, if we are smart about building them.
Jun 24 11:59:30 <Optikal___>    and this isn't exactly an investment, all the returns go to other hcoop members.
Jun 24 11:59:38 <voider>        you're talking about 50/month ?
Jun 24 11:59:55 *       voider scratches his scrotum
Jun 24 12:00:04 <Smerdyakov>    leitgebj, that might be sensible.  Would you think of combining "file server" and "locked-down server"?
Jun 24 12:00:31 <leitgebj>      Perhaps.  I guess I didn't get that far yet! ;)
Jun 24 12:00:34 <Smerdyakov>    voider, the most popular answer was that people are willing to pay $6-$10/mo. in the short term while we build membership and pay off loans.
Jun 24 12:00:53 <Smerdyakov>    leitgebj, because there are many nice consequences of having a server where only admins can log in.
Jun 24 12:01:19 <voider>        ok
Jun 24 12:01:19 <leitgebj>      Smerdyakov, certainly, I agree that we should have one machine with only admin logins.
Jun 24 12:01:50 <mwolson>       that machine had better be doing other things as well though :^)
Jun 24 12:02:04 <Smerdyakov>    leitgebj, well... it seems only natural that the file server should be on that machine instead of the one where users go wild running their programs. :)
Jun 24 12:02:16 <leitgebj>      Smerdyakov, I agree.
Jun 24 12:02:39 <Smerdyakov>    As long as we keep a clean upgrade path to dedicated fileserver mode..
Jun 24 12:03:15 <docelic_>      Technically we could group together 1) and 3). (fileserver, and other services used through protocols and not shell)
Jun 24 12:03:23 <Smerdyakov>    Anyone else's thoughts on leitgebj's idea?
Jun 24 12:03:48 <docelic_>      It would reduce bandwidth too. Like, no need for mail server to lay data on a fileserver that's remote
Jun 24 12:03:56 <Smerdyakov>    leitgebj, also, I wouldn't really say that we've outgrown a single machine... rather that a certain level of redundancy is important for reliability.
Jun 24 12:04:08 <Smerdyakov>    docelic, but it would be local bandwidth only. *shrug*
Jun 24 12:04:40 <Optikal___>    Are most the reliability problems at the machine level rather than the network level?
Jun 24 12:04:45 *       voider has quit (Read error: 104 (Connection reset by peer))
Jun 24 12:04:57 <docelic_>      Smerdyakov: yeah.. still. Well the price would be most visible factor in this condensed setup. And admining would be a little easier as well
Jun 24 12:05:10 <docelic_>      Why have 2 machines with admin-logins only when there can be one
Jun 24 12:05:24 <docelic_>      I kind of like the 2-in-1 idea actually, in this phase at least
Jun 24 12:05:25 <mwolson>       docelic_: yeah, it would definitely be best to have fileserver and mailserver on the same machine; i know NFS at least does not like writing to many small files
Jun 24 12:05:35 <Smerdyakov>    docelic, compute-bound and disk-bound servers have very different criteria for good hardware specs... that's one reason.
Jun 24 12:06:27 <docelic_>      Smerdyakov: right. However, I estimate buying one is still better, as you can save on prt of specs that would be underused if there was 2 separate systens
Jun 24 12:06:34 <Smerdyakov>    But I agree that price concerns trump that sort of thing today, as long as we can meet a certain performance threshold to be determined.
Jun 24 12:06:39 <docelic_>      part* and systems*
Jun 24 12:07:19 *       voider (n=Is@modemcable169.254-80-70.mc.videotron.ca) has joined #hcoop
Jun 24 12:07:20 <docelic_>      With modern AMD or Intel processors, we can do anything
Jun 24 12:07:44 <Smerdyakov>    docelic, for instance, file server might be more reliable with a complicated RAID mode that slows down ordinary operation.
Jun 24 12:07:44 *       Optikal_ has quit (Read error: 110 (Connection timed out))
Jun 24 12:08:03 *       TheDebugger (i=TheDebug@modemcable135.111-81-70.mc.videotron.ca) has joined #hcoop
Jun 24 12:08:12 *       TheDebugger (i=TheDebug@modemcable135.111-81-70.mc.videotron.ca) has left #hcoop
Jun 24 12:08:55 <voider>        oops, lost connection
Jun 24 12:09:10 <Smerdyakov>    OK... so have we magically shifted to agreeing on a two-server main set-up?
Jun 24 12:09:20 <docelic_>      Smerdyakov: you are right. But it comes down to estimating what's good enough, and as far as Im concerned, 1 system setup could work in this phase.
Jun 24 12:09:33 <docelic_>      With a physical room to grow at HE, we can always split later
Jun 24 12:10:01 <Smerdyakov>    It would be nice to have an automated system for switching static web sites between servers in an emergency.
Jun 24 12:10:27 <docelic_>      Between which servers ?
Jun 24 12:10:30 <leitgebj>      Smerdyakov, I was thinking about that, too... actually would be pretty easy to make it work with dynamic sites as well, in the future.
Jun 24 12:10:41 <Smerdyakov>    The big pain I've encountered with fyodor is ulimits and related stuff.
Jun 24 12:10:58 <Smerdyakov>    They tend to interfere with system services, at least when a novice like me sets things up
Jun 24 12:11:06 <Smerdyakov>    Which is why I think two servers is worth the effort.
Jun 24 12:11:15 <docelic_>      Smerdyakov: the ulimits thing is working pretty well now that it settled down , doesn't it ?
Jun 24 12:11:39 <Smerdyakov>    docelic, maybe you figured out the right way to restart services so they aren't killed for forking, but I'm still not sure. :)
Jun 24 12:13:20 <leitgebj>      I think that there are lots of reasons to have two systems.  In an extreme case (hardware on one machine dies), we could also recover in less time by switching services to the other box.  Currently, we would have to buy a machine, configure it, and send it out.  That is a lot of time that the coop would be down for.
Jun 24 12:14:27 <docelic_>      Smerdyakov: there was a discussion in Debian about it, I made a mental note to read the conclusion, but IIRC I think this was some system level problem with limit inheritance that was identified and planned to be fixed 
Jun 24 12:14:53 <Smerdyakov>    Well... are we agreed on two servers, or are one or three still in consideration?
Jun 24 12:15:38 <docelic_>      I suppose the "two servers" you and leitgebj mentioned just now, were actually meant as "two admin-only servers", as in fileserver and other server separate ?
Jun 24 12:15:45 <Smerdyakov>    No.
Jun 24 12:15:49 <Smerdyakov>    Not when I said it, at least
Jun 24 12:15:51 <jch5>  and that is one reason I was wondering why Abu can't be used as emergency backup for current sys...just curious.. 
Jun 24 12:16:03 <docelic_>      "<leitgebj> I think that there are lots of reasons to have two systems. "
Jun 24 12:16:07 <Smerdyakov>    jch5, we haven't taken the time to set it up.
Jun 24 12:16:25 <leitgebj>      I meant two systems, total.  One for admin, and one for user access.  Nothing else, to start.
Jun 24 12:16:28 <Smerdyakov>    jch5, and Abu has some crappy set-up aspects to it... plus unaffordable bandwidth rates at its location.
Jun 24 12:17:29 <Smerdyakov>    Using fyodor as an automatic back-up would be feasible once the new stuff is going.
Jun 24 12:17:55 <Smerdyakov>    Since we have enough "free bandwidth" to transfer a complete dump of users' web site content in an emergency without going broke.
Jun 24 12:18:02 <jch5>  Smerdyakov: that idea works for me.. 
Jun 24 12:18:23 <docelic_>      leitgebj: ah, that's assumed. We were thinking about 2 servers vs. 3 servers choice
Jun 24 12:20:15 <Smerdyakov>    _OK_, so it seems that we agree on 2 main servers total.
Jun 24 12:20:48 <docelic_>      ok
Jun 24 12:20:51 <jch5>  is there any situation where we might need a server just for time of conversion but not ongoing?
Jun 24 12:21:01 <docelic_>      no
Jun 24 12:21:18 <leitgebj>      I don't think so... that is what fyodor and abu are for! ;)
Jun 24 12:21:47 <jch5>  i meant in terms of what is located at HE
Jun 24 12:22:40 <jch5>  remember, i'm just an slightly literate user who used to be a mainframe pgmr
Jun 24 12:23:34 <Smerdyakov>    jch5, I think the answer is no, but I don't quite understand the question... let's move on, OK? :)
Jun 24 12:25:06 <Smerdyakov>    I am officially declaring that we have decided on 2 main servers: one for things that require member log-in and one for things that don't.
Jun 24 12:25:08 <jch5>  fine...(once my employer had to rent a second CPU just for period of conversion from old to new)
Jun 24 12:25:30 <Smerdyakov>    That isn't meant as an undeniable edict, but just to get things going... make an explicit objection if you want to question the assumption.
Jun 24 12:25:46 <Smerdyakov>    Now, what hardware should we buy and where?
Jun 24 12:26:07 <Smerdyakov>    unknown_lamer questioned the value of a hardware firewall on the wiki.  I'm inclined to agree that we can at least avoid one to start.
Jun 24 12:26:32 <leitgebj>      I agree.
Jun 24 12:26:34 <Smerdyakov>    So what does that leave us to buy that wouldn't be contained inside one of the servers?
Jun 24 12:26:56 <Smerdyakov>    Serial console seems like a potential life-saver and worth any reasonable one-time cost.
Jun 24 12:27:15 <docelic_>      You mean a serial console we can connect to remotely ?
Jun 24 12:27:19 <Smerdyakov>    Yes.
Jun 24 12:27:27 <docelic_>      That's a winner, for sure
Jun 24 12:27:29 <leitgebj>      With only two machines, there would also be no reason to a switch.  Most modern servers come with two nics, and we can set up a network between them for the LAN.
Jun 24 12:27:48 <Smerdyakov>    leitgebj, what if we move Abu in as a generic shell server?
Jun 24 12:28:00 <leitgebj>      Can one of you point me to a manufacturer of a serial console?  I've never used one.
Jun 24 12:28:14 <docelic_>      leitgebj: cyclades.com (long-time linux friends)
Jun 24 12:28:19 <Smerdyakov>    leitgebj, no.  I'm ignorant.  tarsin recommended them to me first, I think.
Jun 24 12:28:30 <leitgebj>      Smerdyakov, do we have a wiki page with specs on abu?
Jun 24 12:28:44 <Smerdyakov>    leitgebj, I don't think so.
Jun 24 12:29:18 <Smerdyakov>    Here's something from the old wiki on it:
Jun 24 12:29:19 <Smerdyakov>        *
Jun 24 12:29:19 <Smerdyakov>          AMD Athlon Thunderbird 1.33GHz 266MHz FSB 256KB cache
Jun 24 12:29:19 <Smerdyakov>        *
Jun 24 12:29:19 <Smerdyakov>          CoolJag JAC311C low profile 1U cooling fan and heatsink
Jun 24 12:29:19 <Smerdyakov>        *
Jun 24 12:29:19 <Smerdyakov>          2x 256MB Kingston PC2100 CL2.5 DDR RAM
Jun 24 12:29:19 <Smerdyakov>        *
Jun 24 12:29:19 <Smerdyakov>          Asus A7N266-VM nVidia nForce 220 266MHz FSB motherboard (onboard vid + LAN)
Jun 24 12:29:19 <Smerdyakov>        *
Jun 24 12:29:19 <Smerdyakov>          2x IBM 40GB 120GXP IDE 7200rpm 2MB cache 3YR warranty hard disk, mirrored with RAID-1
Jun 24 12:29:19 <Smerdyakov>        *
Jun 24 12:29:19 <Smerdyakov>          SuperMicro SC811-IDE 1U 200W rackmount case
Jun 24 12:29:19 <Smerdyakov>        *
Jun 24 12:29:19 <Smerdyakov>          3Ware Escalade 3W-6410 PCI IDE RAID adapter (4x UDMA66 EIDE ports)
Jun 24 12:29:19 <Smerdyakov>        *
Jun 24 12:29:19 <Smerdyakov>          32bit 33MHz PCI riser card
Jun 24 12:29:19 <Smerdyakov>        *
Jun 24 12:29:19 <Smerdyakov>          12X IDE slim CDROM (from a Toshiba laptop)
Jun 24 12:29:48 <Smerdyakov>    Later amended with these:
Jun 24 12:29:49 <Smerdyakov>        *
Jun 24 12:29:49 <Smerdyakov>          AMD Duron 900MHz 192KB cache 200MHz FSB
Jun 24 12:29:49 <Smerdyakov>        *
Jun 24 12:29:49 <Smerdyakov>          CoolJag JAC313C low profile fan and heatsink
Jun 24 12:29:49 <Smerdyakov>        *
Jun 24 12:29:49 <Smerdyakov>          Gigabyte GA-7VKML VIA KM266 motherboard (onboard vid + LAN)
Jun 24 12:30:45 <leitgebj>      As long as it doesn't push us over our space limit, and it doesn't become an important single point of failure, it shouldn't matter.  1U should be fine hanging out at HE until we have a need for the space.  Would we want to send it somewhere to reload the OS before shipping it to HE?
Jun 24 12:31:14 <Smerdyakov>    We asked for the initial size quote assuming three servers, so I imagine it would be fine.
Jun 24 12:31:18 <docelic_>      Depends on HE and setup
Jun 24 12:31:32 <docelic_>      I would love to be able to do basic setup without hardware at my place
Jun 24 12:31:42 <Smerdyakov>    It would probably be cheaper to have it mailed to some member to set up first.
Jun 24 12:31:45 <docelic_>      If we would have serial console, I could install remotely
Jun 24 12:31:47 <leitgebj>      He gives us 7U with the plan that we were looking at.
Jun 24 12:31:53 <docelic_>      Smerdyakov: ^--
Jun 24 12:32:14 <Smerdyakov>    docelic, but you're pretty far from the current location... and extra distance is extra chance of damage.
Jun 24 12:32:19 <leitgebj>      I would probably be able to have it shipped to my place and plug it in to a high-speed connection... as long as it isn't in that period when I am moving houses.
Jun 24 12:32:37 <Smerdyakov>    mwolson is closest, I think. :)
Jun 24 12:32:38 <docelic_>      Smerdyakov: I said I could do it *without* it coming over to me
Jun 24 12:33:12 <Smerdyakov>    There probably is very little that needs doing, with the serial console.
Jun 24 12:33:22 <Smerdyakov>    Just get the OS going and hardware working..
Jun 24 12:33:25 *       Optikal___ has quit (Read error: 110 (Connection timed out))
Jun 24 12:33:33 <docelic_>      Smerdyakov: yes, as I said above
Jun 24 12:33:39 <leitgebj>      I would have no problem doing that, or mwolson, if he wants...
Jun 24 12:33:40 <docelic_>      I can even do a remote isntall with the console
Jun 24 12:33:51 <mwolson>       i'd be willing to receive a shipment and do initial installation stuff
Jun 24 12:34:15 <Smerdyakov>    mwolson, and it would also involve shipping it out again afterward, of course. :)
Jun 24 12:34:20 <mwolson>       i could just bring it with me to work and use the space there :^)
Jun 24 12:34:24 <voider>        where will be located the physical servers?
Jun 24 12:34:37 <mwolson>       Smerdyakov: right
Jun 24 12:35:13 <Smerdyakov>    voider, the quote we got seems to be for Fremont, CA... a 50 minute BART ride away from me.
Jun 24 12:35:22 <mwolson>       we usually go with Debian testing, right?
Jun 24 12:35:45 <Smerdyakov>    mwolson, I think that's a good idea for the users-can-log-in server; stable sounds good for the admins-only.
Jun 24 12:35:55 <mwolson>       ok
Jun 24 12:36:03 <Smerdyakov>    And even unstable for Abu if we use it as a generic shell server.
Jun 24 12:36:17 <voider>        CA?
Jun 24 12:36:21 <Smerdyakov>    California, USA
Jun 24 12:36:21 <voider>        US ?
Jun 24 12:36:26 <voider>        ok
Jun 24 12:37:19 <voider>        Im from quebec, CA
Jun 24 12:37:23 <voider>        as in Canada
Jun 24 12:37:36 <Smerdyakov>    Let's see... the reason we got into this discussion was because we were considering if we need to obtain a switch.
Jun 24 12:37:46 <Smerdyakov>    Let's get back to that... try to finalize the list of new hardware.
Jun 24 12:38:33 <leitgebj>      Well, with three machines the dual-nic's won't be sufficient anymore.  I need to look at the He.net spec sheet on the colo page of our wiki to see if they provide one.
Jun 24 12:38:37 <Smerdyakov>    It sounds like, at most, we want 2 servers, serial console, and a switch.
Jun 24 12:39:38 <mwolson>       it'd probably be best to have a switch so that we don't get double traffic on one of the machines (and for the sake of future expansion)
Jun 24 12:40:19 <jch5>  and might we name one of them "Hopper"? (My suggested name that came in second place, i think)
Jun 24 12:40:27 <mwolson>       also, if the machine that passes the connection on gets rooted or goes down, well, there goes our redundancy plan
Jun 24 12:40:27 <Smerdyakov>    jch5, anything's possible. ;)
Jun 24 12:40:49 <Erlang>        I guess the names should be voted later right?
Jun 24 12:40:49 <leitgebj>      I'm guessing that HE won't provide a switch... we should look around at available models.
Jun 24 12:40:58 <Smerdyakov>    We should probably also have a full supply of replacement parts in the cabinet, which also suggests buying servers with similar hardware.
Jun 24 12:41:01 <Smerdyakov>    Erlang, yes
Jun 24 12:41:07 <Erlang>        good.
Jun 24 12:41:24 *       Erlang goes back to lurking mode.
Jun 24 12:41:32 <leitgebj>      Smerdyakov, definitely.  Support contracts also often come in handy.
Jun 24 12:42:15 <Smerdyakov>    So the list seems to be: 2 servers, serial console, switch, replacement parts to cover the 2 servers (assuming failures are rare, so just one of each kind of part (?))
Jun 24 12:42:15 <leitgebj>      It is nice to know that someone will ship you a part in a known amount of time when something breaks... and something always does.
Jun 24 12:42:50 <leitgebj>      I don't know if we need replacement parts to cover the two servers... I don't think we want to buy a whole third server just for parts.  I would rather spend the money on a support contract.
Jun 24 12:43:13 <Smerdyakov>    leitgebj, that would mean that, to keep high uptime, we'd really need to be sure one server can take over for a downed neighbor.
Jun 24 12:43:36 <leitgebj>      We may be able to get a support contract that guarantees that the part will arrive in 24h if something goes down.
Jun 24 12:43:41 <jch5>  i know names come later..(just thought a tenative name might make discussions of which machine being referred to easier)
Jun 24 12:43:51 <Smerdyakov>    With parts on hand, it can be fixed in 10 minutes...
Jun 24 12:44:39 <leitgebj>      Smerdyakov, it will be tricky with only two servers.  Eventually, we should have a pool of servers behind a load balancer, then redundant database back-ends, etc.  But I think that this is cost-prohibitive at the moment.
Jun 24 12:45:06 <jch5>  but does cost of spare parts on hand outweigh the advantage of getting better/newer parts to replace a failed?
Jun 24 12:45:32 <Smerdyakov>    jch5, they would only be meant to be used temporarily and would probably have worse specifications than the usuals.
Jun 24 12:46:42 <jch5>  fine...just asking the dumb questions that might prompt a smart thought by admins :) 
Jun 24 12:47:44 <Smerdyakov>    leitgebj, so deciding by yourself you would choose to try to get a 24-hour part replacement contract?
Jun 24 12:48:25 <leitgebj>      yeah, or something similar.  I think that we should move to eliminate single points of failure in the near future, but having an entire extra system in parts sitting around doesn't make sense to me.
Jun 24 12:48:46 <Smerdyakov>    OK.  Some colo provider I corresponded with recommended it.
Jun 24 12:49:44 <Smerdyakov>    In any case, it's a feature that's easy to add later with no service interruption! :D
Jun 24 12:49:48 <jch5>  if we had 10 or 15 servers, having one as spare parts makes more sense....but one for two...nah
Jun 24 12:50:32 <mwolson>       jch5: yeah, having pre-bought spare parts for us seems pointless
Jun 24 12:50:40 <leitgebj>      Spare parts are a good idea... but I think that I would rather take a risk without the parts now to save money, and plan for the near future where we can have load-balancers, etc., in order to have no down time in the event of failures.
Jun 24 12:50:58 <Smerdyakov>    OK, then it seems our list of hardware to buy is 2xserver, serial console, and switch.
Jun 24 12:51:20 <mwolson>       sounds good to me
Jun 24 12:51:39 <leitgebj>      I move to at least seriously think about adding in the cost of hardware support contracts to the cost required for our configuration.
Jun 24 12:51:55 <Smerdyakov>    That's OK with me.
Jun 24 12:53:10 <Smerdyakov>    Time to move on to "where do we get this stuff?"?
Jun 24 12:53:17 <docelic_>      sounds good to me
Jun 24 12:53:20 <leitgebj>      Sure.
Jun 24 12:54:18 <mwolson>       at work newegg.com seems to be popular
Jun 24 12:55:19 <Smerdyakov>    PenguinComputing seems ncie for complete machines with combined warranties, but they only sell machines preloaded with RedHat or SUSE. :(
Jun 24 12:55:34 <docelic_>      does that matter? 
Jun 24 12:55:43 <docelic_>      Over a console I can run install from scratch
Jun 24 12:56:01 <Smerdyakov>    docelic, seems a waste... and they must charge extra to some extent for the installation.
Jun 24 12:56:50 <leitgebj>      Smerdyakov, they probably have some kickstart schema setup where they only hit F12 on bootup to load the OS.  Probably just a standard part of their hardware testing anyway.
Jun 24 12:57:17 <mwolson>       (newegg in particular might be good for the switch)
Jun 24 12:57:30 <Smerdyakov>    mwolson, ah, but not for the servers, right?  (No on-site support contracts)
Jun 24 12:57:38 <mwolson>       Smerdyakov: right
Jun 24 12:58:01 <leitgebj>      I have seen hundreds of dell machines work well in production, and they have solid support contracts.  They are also switching to AMD in the near future.  That said, I wouldn't mind if we switched to someone like Penguin, it's just that I don't have experience with them.
Jun 24 12:58:02 <Smerdyakov>    I wonder what happens if the switch breaks, though. :o
Jun 24 12:58:31 <mwolson>       Smerdyakov: it's a very inexpensive item ...
Jun 24 12:58:40 <Smerdyakov>    mwolson, OK.
Jun 24 12:58:47 <mwolson>       manufacturer probably provides a warranty as well
Jun 24 12:59:02 <leitgebj>      I'd like to look at Cisco switches from a vendor with solid support.  mwolson, I'm not so sure this is going to be a trivial expense ;)
Jun 24 12:59:25 <mwolson>       leitgebj: why? we only have 2-3 things to plug into it
Jun 24 13:00:03 <leitgebj>      If this interconnect goes down, all of our services go down.
Jun 24 13:00:19 <leitgebj>      It is a single point of failure, and therefore we want it to be extremely relilable and well-supported.
Jun 24 13:02:06 <Smerdyakov>    Having servers from Penguin Computing vs. Dell buys geek cred points, if nothing else. :)
Jun 24 13:02:27 <docelic_>      ++
Jun 24 13:02:30 <leitgebj>      Hey, if the support is there, I'm all for it.
Jun 24 13:02:54 <Smerdyakov>    They sell next-day on-site service for 3 years for $119.
Jun 24 13:03:00 <mwolson>       leitgebj: http://www.newegg.com/Product/Product.asp?Item=N82E16833122111
Jun 24 13:03:01 <mwolson>       ^^ 5-star rated $59.99 gigabit switch
Jun 24 13:03:31 <mwolson>       s/59/56/
Jun 24 13:03:56 <leitgebj>      mwolson, now what about one that's rack-mountable :)
Jun 24 13:04:42 <leitgebj>      I still think that money on a switch is well-spent... that seems like something that is better for an office LAN than a production server environment.  Shouldn't we get a managed switch?
Jun 24 13:05:19 <Smerdyakov>    I yield to any informed opinion and just reiterate that I'm willing to spend money up-front to ensure quality. :)
Jun 24 13:07:04 <mwolson>       *sigh* http://www.newegg.com/Product/Product.asp?Item=N82E16833118021
Jun 24 13:07:04 <mwolson>       ^^ $249.99 rack-mountable switch
Jun 24 13:07:12 <docelic_>      I can;t say much on switches
Jun 24 13:07:21 <jch5>  newegg was recommended to me by a very techie friend...i've had good luck with them also..
Jun 24 13:07:24 <Smerdyakov>    Do we want to leave these issues as "homework" and move on to something else?
Jun 24 13:07:29 <leitgebj>      I still need to do research on this as well.  Can we start a wiki page with some possible products and have a quick meeting, or email list discussion on this later?
Jun 24 13:07:31 <docelic_>      good idea
Jun 24 13:07:43 <mwolson>       can't we just throw the switch on top of a machine?  does it really need to be rack-mountable?
Jun 24 13:07:48 <docelic_>      it's too nitpicky for the management level we're trying to keep
Jun 24 13:08:06 <leitgebj>      Right.  OK with everyone if I start a wiki page on it?
Jun 24 13:08:15 <Smerdyakov>    It's always OK to start a wiki page on anything.
Jun 24 13:08:27 <Smerdyakov>    Worst that happens is that it's naked pictures of your cat and I have to delete it. ;)
Jun 24 13:08:38 <leitgebj>      What's wrong with my cat??? :)
Jun 24 13:09:04 <docelic_>      too little fur for a cat
Jun 24 13:09:17 <leitgebj>      docelic, lol
Jun 24 13:09:26 <Smerdyakov>    OK, I think we have our first action item for the meeting!  Whoever is interested is going to research possibilities for the four items we want to buy and post information on the wiki page that leitgebj is creating.
Jun 24 13:10:07 <mwolson>       ok
Jun 24 13:10:44 <jch5>  and the budget for these items will be how much for each piece of equip?
Jun 24 13:10:59 <Smerdyakov>    A billion dollars
Jun 24 13:11:01 <Smerdyakov>    Next question?
Jun 24 13:11:29 <jch5>  ok..Mr Gates...whatever
Jun 24 13:11:30 <Smerdyakov>    I think we'll figure out the budget after we see the options.
Jun 24 13:12:12 *       voider has quit ("Leaving")
Jun 24 13:14:01 <Smerdyakov>    Other things to discuss include administrative structure.
Jun 24 13:14:18 <Smerdyakov>    Is there anything else besides that that we should discuss by the end of the meeting?
Jun 24 13:15:03 <leitgebj>      We should set a date to reconvene at, with concrete goals.
Jun 24 13:15:29 <Smerdyakov>    OK.  That seems to be a meta-point that belongs at the end.
Jun 24 13:16:07 <mwolson>       let's see ... so i'm an admin now?
Jun 24 13:18:09 <Smerdyakov>    mwolson, sort of
Jun 24 13:18:19 <Smerdyakov>    It sounds like no one has any other topics to raise, so let's move to that.
Jun 24 13:18:48 <Smerdyakov>    Since a lot of our configuration will be changing, it doesn't seem to make particular sense for either leitgebj or mwolson to start as admins now.
Jun 24 13:19:46 <Smerdyakov>    The first task for the new admin team is this hardware and software planning and set-up, as I see it.
Jun 24 13:20:16 <Smerdyakov>    After that, we move into maintenance mode, handling issues of predictable time availability, handling support requests, etc..
Jun 24 13:20:53 <Smerdyakov>    One thing I want to do is create a new hcoop-sysadmin mailing list that is open to all members.
Jun 24 13:21:17 <Smerdyakov>    The idea would be that any sysadmin stuff is discussed there.  There's no reason not to get advice from knowledgeable members just because it doesn't make sense to give them all root access.
Jun 24 13:22:27 <mwolson>       can we make portal requests trigger an automated email to that new list, for those that prefer email to web forms?
Jun 24 13:22:38 <mwolson>       (thinking of metaperl in particular)
Jun 24 13:22:58 <docelic_>      Smerdyakov: isn't that admins@ alias now ?
Jun 24 13:23:01 <Smerdyakov>    mwolson, portal requests already e-mail everyone subscribed to the associated category.
Jun 24 13:23:05 <docelic_>      we could just turn it into a ML
Jun 24 13:23:14 <mwolson>       ah, didn't know that
Jun 24 13:23:25 <Smerdyakov>    mwolson, you can already subscribe to any that interest you.
Jun 24 13:24:07 <docelic_>      I also think Smerdyakov  and me can handle issues of porting current infrastructure to new servers, while new admins (mwolson, leitgebj) can be given tasks that don't rely on our current setup
Jun 24 13:24:12 <Smerdyakov>    docelic, yes, but the key difference is that anyone can subscribe.
Jun 24 13:24:29 <mwolson>       if we make hcoop-sysadmin a catch-all mailing list that gets all portal requests, though, it might be a good thing
Jun 24 13:24:33 <docelic_>      Smerdyakov: sure, Im just saying, move admins@ to be a ML and notify admins that it's now a public lisr
Jun 24 13:24:36 <docelic_>      list*
Jun 24 13:24:54 <docelic_>      Smerdyakov: or not, we should have an alias for admins that is private
Jun 24 13:24:56 <Smerdyakov>    mwolson, I don't know about that.  Filtering based on subject is quite useful.
Jun 24 13:25:20 <mwolson>       (unless we want to separate support from administration, which is somewhat nebulous)
Jun 24 13:25:23 <Smerdyakov>    mwolson, I'm thinking of the mailing list as being for discussion, like that surrounding our current planning.
Jun 24 13:25:51 <Smerdyakov>    mwolson, I never intended to suggest that the sysadmin list would be the target for e-mailed support requests.
Jun 24 13:25:56 <mwolson>       ah, i see
Jun 24 13:26:16 <Smerdyakov>    The portal has supported collaboration from all members in handling support issues from the start, even if you didn't read the part of the wiki page that explains that. ;)
Jun 24 13:28:13 <Smerdyakov>    So I think I'll create that list after the meeting ends, along with modifying the portal to let people manage their subscriptions to it.
Jun 24 13:28:44 *       docelic has quit (Remote closed the connection)
Jun 24 13:28:50 *       docelic_ is now known as docelic
Jun 24 13:29:06 <Smerdyakov>    And we can switch all potential hcoop-discuss chatter on technical planning issues to hcoop-sysadmin.
Jun 24 13:29:21 <leitgebj>      That seems to make sense.
Jun 24 13:30:28 <Smerdyakov>    On AdminPolicies, I talk about guaranteed within-24-hours e-mail response time from anyone in an official position.
Jun 24 13:30:44 <Smerdyakov>    I think e-mail to hcoop-sysadmin sounds like a good way to notify of times when you're away and unable to provide that.
Jun 24 13:31:05 <docelic>       yep
Jun 24 13:31:25 <leitgebj>      Sounds good.
Jun 24 13:31:54 <Smerdyakov>    There are a number of other suggestions for procedures that I've made on AdminPolicies, but it seems like we can/should wait for those until we are switched over to the new set-up.
Jun 24 13:33:09 <Smerdyakov>    Mmmm.... anything else before we jump to the conclusion?
Jun 24 13:33:41 <docelic>       not here
Jun 24 13:34:28 <leitgebj>      not here either...
Jun 24 13:34:51 <Smerdyakov>    OK, then when should we meet next?  If we're just doing research on hardware in the interim, one week seems like long enough to me.
Jun 24 13:35:08 <docelic>       it's ok
Jun 24 13:35:10 <leitgebj>      Agreed.
Jun 24 13:35:29 <Smerdyakov>    How about same time next week?
Jun 24 13:35:34 <docelic>       yep
Jun 24 13:35:35 <jch5>  user agrees also..
Jun 24 13:35:40 <docelic>       (I thought that was implied)
Jun 24 13:35:41 <leitgebj>      mwolson, is one week OK for you?  you had some opinions on hardware...
Jun 24 13:36:09 <mwolson>       one week is fine
Jun 24 13:36:25 <Smerdyakov>    mwolson, and you can show up on time next week? ;)
Jun 24 13:36:40 <mwolson>       will try; got 4pm instead of 14:00 in my head for some reason
Jun 24 13:37:09 <Smerdyakov>    Did you follow the link to the time-all-over-the-world page?
Jun 24 13:37:16 <mwolson>       yes ...
Jun 24 13:37:19 <Smerdyakov>    OK.
Jun 24 13:37:45 <Smerdyakov>    I will send out an e-mail announcing the main results of the meeting.  Let's see if I've got everything that I should include:
Jun 24 13:37:51 <Smerdyakov>    Link to log of this meeting on the wiki
Jun 24 13:38:06 <Smerdyakov>    Decided on new admins for after the hosting switch
Jun 24 13:38:10 <Smerdyakov>    Decided on HE
Jun 24 13:38:21 <Smerdyakov>    Decided on what hardware we want
Jun 24 13:38:34 <Smerdyakov>    Researching hardware options/prices now, to be recorded on a page that leitgebj should tell me about :)
Jun 24 13:38:48 <Smerdyakov>    Next meeting in a week, hopefully with ntk in attendance so that it's a real board meeting
Jun 24 13:39:00 <leitgebj>      Sounds good, I will do that in a minute.
Jun 24 13:39:09 <Smerdyakov>    Oh, and add new hcoop-sysadmin list in there.
Jun 24 13:39:39 <leitgebj>      I will update the wiki page to reflect our conclusions here, except for the IRC meeting log that Smerdyakov will upload, OK?
Jun 24 13:39:42 <Smerdyakov>    I think that's the whole list.
Jun 24 13:39:44 <Smerdyakov>    Sure.
Jun 24 13:39:48 <Smerdyakov>    Anything I missed, anyone?
Jun 24 13:40:03 <jch5>  thanks for answering my questions!
Jun 24 13:40:06 <leitgebj>      Looks good to me.
Jun 24 13:40:58 <Smerdyakov>    OK, then I guess that concludes the meeting!

IrcMeetings/20060624 (last edited 2009-09-09 20:13:44 by ClintonEbadi)