<div dir="ltr">Notes from sysadmin meeting 20080812<br><br>Initial Scope<br>How to manage<br>How to recruit and retain volunteers<br>How to insure good communication and accountability<br><span class="__mozilla-findbar-search" style="padding: 0pt; background-color: yellow; color: black; display: inline; font-size: inherit;">How to eval</span>uate the project<br>
How many people are enough?<br><br><b>Initial Scope<br></b><br>the discussion thus far has been around using the infrastructure-gang for trac/git, wiki admin, and possibly rt admin<br><br>
<b>How to manage</b><br><br>sysadmin will be "benevolent dictator", everyone else will be everyone else<br><br>
<gregdek> 1. Keep a list of tasks and go through them every week/meeting interval. At least, the highest priority items.<br><Ian_Daniher> gregdek: +1<br><gregdek> 2. Make sure that every task has (a) a clear description, (b) a clear owner, and (c) a deadline of some kind.<br>
<gregdek> 3. The "leader" basically walks everyone thru the agenda.<br><gregdek> 4. The gathered peers publicly embarrass those who continually drop tasks. :)<br><br><b>How to recruit and retain volunteers<br>
<br></b><gregdek> I think you're dealing with a different class of volunteer in sysadminland.<br><gregdek> You're not going to need to do a whole lot of training.<br><gregdek> What you need to do is figure out what responsibilities you're willing to hand over...<br>
<isforinsects> ideally the sysadmin team will train you.<br><gregdek> ...figure out what your trust relationship is...<br><gregdek> ...and then get the hell out of the way. :)<br><b><br>How to insure good communication and accountability<br>
<br></b><hhardy> Quis custodiet ipsos custodes?<br><isforinsects> on the wiki/rt/db cleaning note, a wiki-team might be in the works<br>* gregdek doesn't speak Latin. :)<br><hhardy> "who will watch the watchmen"<br>
<Ian_Daniher> hhardy: tu<br><gregdek> The watchmen will watch each other, and you, hhardy, will be accountable.<br><b><br>
<span class="__mozilla-findbar-search" style="padding: 0pt; background-color: yellow; color: black; display: inline; font-size: inherit;">How to eval</span>uate the project<br><br></b><hhardy> I want to keep an eye on things and expand on what works and triage what not works<br>
<cjl> evaluate -> see weekly meetings<br>
<gregdek> Yep.<br><gregdek> You might want to consider some metrics...<br><gregdek> ...but most things are pretty pass/fail.<br><hhardy> we will have a sense from meetings if formalisms are needed they will porbably evolve<br>
<cjl> also declare successes and admit failures to OLPC mgmt.<br><gregdek> "Did we get monitoring set up for these 5 hosts? No? Why not?"<br><CBIgenho> is this "project" goal driven?<br>
<gregdek> Very much so.<br><gregdek> The goal:<br><hhardy> s/porbably/probably/<br><gregdek> To ensure 100% uptime of all systems without hhardy having to do a thing. ;)<br><gregdek> With additional goals coming along all the time.<br>
<b><br>
How many people are enough?<br></b><br>Gregdek says 6 are enough for the initial scope, but if more we will expand over time<b><br></b><br><b><br>transcript follows:</b><br><br><div class="gmail_quote"><div dir="ltr"><div class="gmail_quote">
<div dir="ltr">* gregdek hullos.<br><hhardy> gregdek: hello<br><gregdek> Who do we have for the meeting?<br><CanoeBerry> Hola, I'll try to join most of yr mtg Henry, thx!<br><hhardy> welcome to the OLPC sysadmin meeting<br>
<hhardy> today has a special topic: organizing the OLPC Infrastructure Group, aka olpc infrastructure-gang<br><hhardy> this meeting is open to all<br><hhardy> however, please don't jump the gun and announce anything discussed here as definitely decided<br>
<CanoeBerry> reubencaron: hi!<br><reubencaron> hello!<br><hhardy> agenda follows:<br><cjl> hello<br><hhardy> Today's meeting will be devoted to discussing the proposed OLPC infrastructure volunteer group (up to 50 minutes):<br>
<hhardy> Initial Scope<br><hhardy> How to manage<br><hhardy> How to recruit and retain volunteers<br><hhardy> How to insure good communication and accountability<br><hhardy> <span class="__mozilla-findbar-search" style="padding: 0pt; background-color: yellow; color: black; display: inline; font-size: inherit;">How to eval</span>uate the project<br>
<hhardy> How many people are enough?<br><hhardy> end of meeting (up to 10 minutes):<br><hhardy> update on weka, owl and swan<br><hhardy> new business<br><hhardy> I am user hhardy, aka Henry Edward hardy, olpc sysadmin<br>
<hhardy> who else is here?<br><CanoeBerry> reubencaron: plz private-message me<br><hhardy> anyone here who is new to OLPC?<br><coderanger_> coderanger == noah kantrowitz == olpc expat and hypothetical keeper of the Trac<br>
<cjb> the only person who doesn't look obviously familiar is mark__ <br><hhardy> Mark is a new support-ganger yes?<br><mark__> Me .... I am Mark Bauer ... I teach EE at the University of Nebraska and a G1G1 guy<br>
* Ian_Daniher has quit (Remote host closed the connection)<br>* gregdek from RH/Fedora<br><mark__> I just joined the support gang<br><hhardy> welcome Mark__<br><isforinsects> welcome mark<br><hhardy> Initial scope:<br>
<CanoeBerry> Hi Mark, thx for the phone call in July, good to see you.<br><hhardy> here is a list of some functional areas of IT<br><hhardy> trac/git<br><hhardy> provision activities-related accounts for git access<br>
<hhardy> mailing list setup and maintenance<br><hhardy> build system<br><hhardy> websites<br><hhardy> wikis<br><hhardy> mail and shell account provisioning<br><hhardy> rt ticket system<br>
<hhardy> the discussion thus far has been around using the infrastructure-gang for trac/git, wiki admin, and possibly rt admin<br><hhardy> then expanding to other areas<br><CanoeBerry> RT Admin for sure! But more on that later.<br>
<hhardy> discussion?<br><cjl> a fair amount of front-end wiki work gets handled by sysops already<br><CanoeBerry> (Old News: I'm recruiting Adric & JGay to be RT Admins & toolsmiths for our Support Gang)<br>
<isforinsects> I'm strongly for a team that can do this.<br><hhardy> there is already a wiki admin application<br><hhardy> we can also use wiki support from the sysadmin side for upgrades, bugfixes, etc<br>
<isforinsects> I'm confident that we have qualified people in the community that would like to get involved.<br><gregdek> So then. What are the impediments to giving root access responsibly to these potential admins?<br>
<isforinsects> Topic: "Initial Scope"<br><m_stone> initial scope and some agreement on how and where coordination should take place and be recorded.<br>* Ian_Daniher (~<a href="mailto:it@69.61.230.246" target="_blank">it@69.61.230.246</a>) has joined #olpc-admin<br>
<isforinsects> Do we want regular meetings to be IRC and/or phone?<br><isforinsects> I am partial to IRC<br><gregdek> irc ftw. Non-english speakers are more comfortable, meetings are self-documenting.<br>
<hhardy> IRC seems the best choice for day to day admin<br><cjl> also teamWiki Team: space seems like a good place to organize some more permanent things (unless it seems necessary to have a further differentiated space), but teamwiki in any event<br>
<hhardy> disagree?<br><isforinsects> members of the [I-G] should be competent in IRC at the very least.<br>* _sj_ (~<a href="mailto:sj_@wireless-19-210.media.mit.edu" target="_blank">sj_@wireless-19-210.media.mit.edu</a>) has joined #olpc-admin<br>
<cjl> +1 IRC<br><isforinsects> I see nothing wrong with a few introductory meetings over the phone for new members. A big part of the S-G community and conversation takes place there.<br><Ian_Daniher> +1 IRC<br>
<_sj_> send invites to the dev lists of our infra pieces<br><hhardy> ok we can use this channel until further notice if no objection<br><Ian_Daniher> phones have worked well for non-public discussions<br>
<_sj_> we've been talking about this for rt for a while anyway<br><CanoeBerry> Relevant: We need better IRC training for the Support Volunteers. Talk to me if you can help.<br><hhardy> we could use ventrilo rather than phones<br>
<_sj_> hhardy: can you unlock the channel topic here?<br><Ian_Daniher> hhardy: why not use freenode like support gang?<br><isforinsects> freenode is a limited number of total channels that one can have open there.<br>
<isforinsects> #olpc-devel is here, as are a few other channels<br><hhardy> Ian Daniher: no reason this channel was inherited from neuralis<br><Ian_Daniher> hhardy: ah, alright<br>* cjb hereby announces that he doesn't have time for any phone meetings. They require too much attention.<br>
<gregdek> Yep. IRC is good because it's async.<br><gregdek> So we've all agreed, IRC meetings, yes?<br><gregdek> Weekly? This channel? This time? :)<br><Ian_Daniher> Very true, it's also unlikely very sensative info will be discussed during meetings<br>
<hhardy> phone works for support-gang to help build community but I think we have a consensus on IRC, lets move on<br><isforinsects> I would like to include ffm's proxy vote for thursday or friday<br><hhardy> how about sunday at 2pm EDT for weekly meeting time for the i-g?<br>
* joef (<a href="mailto:joe@dhcp-47-117.media.mit.edu" target="_blank">joe@dhcp-47-117.media.mit.edu</a>) has joined #olpc-admin<br><gregdek> I prefer weekdays myself.<br><isforinsects> Yeah, employees at *other* companies could participate at work were it during the week.<br>
<hhardy> weekdays during day not works for people who have school or work, and nights are tough for people w/ kids<br><gregdek> Depends on the class of volunteer.<br><gregdek> Remember, we're talking sysadmins. :)<br>
<hhardy> so anytime after midnight? lol<br><gregdek> Heh.<br><gregdek> Maybe ask for consensus on list.<br><gregdek> Or just decide arbitrarily and change if a fistfight ensues. ;)<br><hhardy> <a href="http://lists.laptop.org/listinfo/olpc-sysadmin" target="_blank">http://lists.laptop.org/listinfo/olpc-sysadmin</a><br>
<CanoeBerry> Current time works for me..<br><hhardy> good place to mention the mailing list for the i-g<br><Ian_Daniher> hhardy: your sunday 2pm time would work marvelously, as it's before s-g meetings<br>
<hhardy> Ian Daniher: my thinking as well<br><gregdek> Do you see a high overlap between s-g and i-g?<br><hhardy> there has been significant interest from that direction<br><hhardy> and they have a model which works for OLPC<br>
<isforinsects> gregdek, I see a higher overlap between the domains of the s-g and the testing-gang<br><gregdek> I might drive as many people to the list as possible, and then ask on-list.<br><hhardy> ok y'all please get on list I'm going to say tentative Sunday 2pm subject to discussion on list<br>
* isforinsects is joining now<br>* gregdek joins.<br>* cjl too<br><hhardy> next topic: how to manage the i-g<br><gregdek> WITH AN IRON FIST.<br><gregdek> :)<br><Ian_Daniher> seriously though, in #s-g, holt lays down the law<br>
<isforinsects> RT tickets or trac?<br><CanoeBerry> Sunday late afternoon / dinnertime works much better than 2pm for me. if it must be Sunday.<br><hhardy> lol I will try to be like Jon Postel was as much as possible<br>
<gregdek> CanoeBerry: slug it out on-list. :)<br>* _sj_ has quit (Ping timeout: 480 seconds)<br><gregdek> My take?<br><gregdek> People mostly manage themselves.<br><gregdek> The only "management" involved in the groups I run for Fedora:<br>
* isforinsects agreed<br><gregdek> 1. Keep a list of tasks and go through them every week/meeting interval. At least, the highest priority items.<br><Ian_Daniher> gregdek: +1<br><gregdek> 2. Make sure that every task has (a) a clear description, (b) a clear owner, and (c) a deadline of some kind.<br>
<gregdek> 3. The "leader" basically walks everyone thru the agenda.<br><gregdek> 4. The gathered peers publicly embarrass those who continually drop tasks. :)<br><gregdek> That's it.<br>* Culseg (<a href="mailto:1893d58d@67.207.141.120" target="_blank">1893d58d@67.207.141.120</a>) has joined #olpc-admin<br>
* isforinsects points and laughs<br><isforinsects> (not at Culseg)<br><hhardy> I suggest rt provides us a good vehicle for 1) and 2)<br><Ian_Daniher> there needs to be a central organization, based around the decrees of our benevolent dictator, and as this is infrastructure where there's mandatory low downtime, there needs to be some fairly serious organization with regards to "who's doing what?"<br>
<hhardy> using the existing sysadmin queue<br><Ian_Daniher> fully utilized RT would be perfect<br><Ian_Daniher> hhardy: can custom fields be added to RT?<br><hhardy> yes<br><gregdek> Fedora uses Trac, I think. But it's either/or.<br>
<Ian_Daniher> RT has worked marvelously so far, for s-g, and is already established<br><hhardy> both are established and either would work<br><gregdek> Make a call.<br><gregdek> Dictator.<br><hhardy> rt<br>
<gregdek> :)<br>* cjl genuflects to hhardy as uber-admin<br><hhardy> the email integration is useful<br><hhardy> lol<br><hhardy> I am very humble in this company<br><hhardy> <-- genuflects<br>
<hhardy> with Minsky pottering around how can I not? :)<br><coderanger_> hhardy: For reference, Trac has email integration too<br><hhardy> do we need any formal structures other than dictator and team?<br>
<coderanger_> lookup email2trac<br><Ian_Daniher> hhardy: I don't believe so<br><Ian_Daniher> that's pretty much the way it is in s-g<br><gregdek> hhardy: When they are required, you'll know. :)<br>
<cjl> hhardy: RT (within existing sysadmin queue?)<br><hhardy> rt:sysadmin queue<br><Ian_Daniher> hhardy: would subdivisions of the sysadmin queue be needed?<br><hhardy> support gang already has access as do "volunteers"<br>
<isforinsects> There are already a backlog of low- and medium- priority tickets in RT<br><Ian_Daniher> if so, is it possible to add them?<br><hhardy> we might set some categories of tickets to correspond with functional working groups<br>
<Ian_Daniher> can we have a central list of who has access to what servers and services?<br><hhardy> we can also add more queues if there is due cause<br>* _sj_ (~<a href="mailto:sj_@wireless-19-210.media.mit.edu" target="_blank">sj_@wireless-19-210.media.mit.edu</a>) has joined #olpc-admin<br>
<gregdek> Ian_Daniher: +1.<br><hhardy> and its a plus to be able to share tickets with s-g<br><cjl> I like idea of sysadmin queue within RT (only needs some definition of who doesn't get fingers chopped off :-)<br>
* CBIgenho (<a href="mailto:41db8208@67.207.141.120" target="_blank">41db8208@67.207.141.120</a>) has joined #olpc-admin<br><gregdek> Share tickets liberally, but make simplified views for ease of management.<br><hhardy> Ian Daniher: there is some docuemntation on internal wiki, this should be put public as much as possible<br>
<CanoeBerry> Welcome Caryl!<br><hhardy> that is a good wiki-task<br><Ian_Daniher> I agree<br><hhardy> <-- waves caryl<br><CBIgenho> Sorry I'm late...I was hopelessly lost!<br><Culseg> has protocol for mentoring RT tickets been discussed?...ie liberal use of 'Comments"<br>
<Ian_Daniher> It would help to know that this person has access to activation server, this other individual to t.l.o, etc<br><hhardy> culseg: say on<br><_sj_> hiya caryl<br><hhardy> I tend to use respond a lot, s-g uses comment more so they end user doesnt see everything<br>
<Culseg> add those with more exp to be allowed to make comments freely in RT tickets until persons are up to speed on issues<br><cjl> Culseg focus has been more directly on sysadmin style tasks, less on SG enhancing suggestions like that<br>
<hhardy> pretty much anyone with read access to sysadmin queue now has powers to open, close, comment etc now<br><CanoeBerry> Comments are badly broken (counterintuitive: fails to email the owner, and worse) but were working on it.<br>
<hhardy> CanoeBerry: use respond if you want to go to requestor<br><CanoeBerry> Comments are so badly broken they require you know the actual email addresses, beyond just RT usernames. But again, we're working on it (RT14244 :)<br>
<gregdek> hhardy: still counterintuitive.<br><hhardy> I want to at least know the real email address of anyone who is into the sysamdin queue<br><hhardy> gregdek: better idea?<br><Ian_Daniher> hhardy: know from memory or from a spreadsheet?<br>
* kimquirk (~<a href="mailto:kimquirk@wireless-42.media.mit.edu" target="_blank">kimquirk@wireless-42.media.mit.edu</a>) has joined #olpc-admin<br><hhardy> Ian_Daniher whichever<br><gregdek> hhardy: Fix RT as planned. :)<br>
<CanoeBerry> Agreed.<br><cjl> RT improvements very much a project for infrastructure-gang to contribute to on behalf of support-gang<br><Ian_Daniher> it's fairly easy to add a commenter if you've memorized their address, but when you have to look it up, it becomes quite a pain<br>
<gregdek> If you have two input mechanisms that people are accustomed to using, make sure that (a) both work, or (b) one is forbidden. Confusion is bad.<br><hhardy> kimquirk: welcome we are at how to manage<br>
<_sj_> this group in particular can note counterintuitive bits and spec out how to fix them<br><kimquirk> sorry i'm late<br><hhardy> kimquirk: they have already plotted how to take over the universe :)<br>
<kimquirk> cool<br><CanoeBerry> Beyond <a href="http://rt.laptop.org/Ticket/Display.html?id=14244" target="_blank">http://rt.laptop.org/Ticket/Display.html?id=14244</a> , my job of cleaning up "Comment" functionality is here: <a href="http://rt.laptop.org/Ticket/Display.html?id=14352" target="_blank">http://rt.laptop.org/Ticket/Display.html?id=14352</a><br>
<CBIgenho> Have you already discussed recruitment, training, and retention?<br><cjl> currently agreement to use sysadmin queue in RT under benevolent dictatorship of hhardy.<br><hhardy> so far we have a organizational structure of dictator + everyone else<br>
<CanoeBerry> Caryl: we delegated that all to you!<br><CBIgenho> Yikes! We need to all work on that<br><kimquirk> oh... if someone is taking over RT mods -- they can have all the RT tickets I own on that subject<br>
<CanoeBerry> I'm on 'em :)<br><hhardy> we will use rt, this IRC chan and will discuss proposed meeting time on the mailing list <a href="mailto:olpc-sysadmin@lists.laptop.org" target="_blank">olpc-sysadmin@lists.laptop.org</a><br>
<CanoeBerry> Henry is forcing us to give up our wkds.<br><CanoeBerry> ;)<br><hhardy> CBIgenho: canoeberry is kidding with you :)<br>* hhardy has changed the topic to: How to recruit and retain volunteers<br>
<CanoeBerry> How are we doing towards wrapping up 5pm?<br><hhardy> I am a bit daunted at the prospect of putting out and announcement and having a slashdot hoarde arrive<br><CBIgenho> Did everyone see the brainstorm results on this topic?<br>
<gregdek> ...<br><hhardy> we will wrap at 17:00 and carry on over the list, this chan and Sunday<br><cjl> CBIgenho: some of those are a bit more SG "classic", to some extent the infrastructure gang (current topic) is more about the "SG for the SG" and sysadmin tasks that hhardy can use help with.<br>
<hhardy> CBIgenho: please feel free to summarize<br><gregdek> May I?<br><hhardy> gregdek: go<br><CBIgenho> one important suggestion was that we have several trained s-gers to help Adam assist in the training<br>
<CanoeBerry> Great! Further Results of Caryl's Brainstorming: <a href="http://lists.laptop.org/pipermail/library/2008-August/000694.html" target="_blank">http://lists.laptop.org/pipermail/library/2008-August/000694.html</a><br>
<gregdek> I think you're dealing with a different class of volunteer in sysadminland.<br><gregdek> You're not going to need to do a whole lot of training.<br><gregdek> What you need to do is figure out what responsibilities you're willing to hand over...<br>
<isforinsects> ideally the sysadmin team will train you.<br><gregdek> ...figure out what your trust relationship is...<br><gregdek> ...and then get the hell out of the way. :)<br><isforinsects> :)<br>
<gregdek> If you get 6 good recruits, you're basically set with the systems you have now.<br><isforinsects> Do we want to migrate the third private wiki to a separate server?<br><isforinsects> And let it die a horrible, unmaintained death? :)<br>
<cjl> Some of infrastructure task is sysadminning, some is training trainers as Caryl mentions.<br><hhardy> kimquirk: opinion of moving internal wiki away from where volunteers can see things?<br><kimquirk> how is it available to volunteers?<br>
<hhardy> if it was only 6 trusted people maybe not a problem, or maybe we should have some nda like agreement<br><hhardy> if they are wiki admins with root access on pedal they could see into internal potentially<br>
<CBIgenho> nda is good<br><kimquirk> i thought the internal wiki was not to be used except for internal contact infroamtion<br><gregdek> I would be wary of mixing hardcore sysadmin tasks with more general "how to keep wiki/rt/db clean" tasks.<br>
<hhardy> so initially they will live in userland<br><gregdek> You may be talking about two sets of problems.<br><kimquirk> I would suggest the internal wiki be moved to 'internal infrastructure'<br>
<gregdek> +1<br><gregdek> hhardy is stuck with that stuff. ;)<br><hhardy> works for hhardy<br>* hhardy has changed the topic to: How to insure good communication and accountability<br><isforinsects> +1<br>
<hhardy> Quis custodiet ipsos custodes?<br><isforinsects> on the wiki/rt/db cleaning note, a wiki-team might be in the works<br>* gregdek doesn't speak Latin. :)<br><hhardy> "who will watch the watchmen"<br>
<Ian_Daniher> hhardy: tu<br><gregdek> The watchmen will watch each other, and you, hhardy, will be accountable.<br><hhardy> or, "who will clean up after the janitors" :)<br><gregdek> The leader reaps the rewards when things go well, and takes the bullet when things go badly. ;)<br>
<Ian_Daniher> hhardy: tu semper custodiet custodes<br><gregdek> But if you do things right, accountability is tracked in RT -- "who has this task? Ask RT" -- and communication is mailing lists + weekly meeting.<br>
<gregdek> The weekly meeting, in my world, is *incredibly important*.<br><hhardy> gregdek: +1<br><gregdek> It's the opportunity to praise people who are doing well, and call out people who are slacking off.<br>
<Ian_Daniher> gregdek: +1<br><gregdek> Oh, and every meeting ends with a common understand of actions.<br><gregdek> s/understand/understanding/<br>* hhardy has changed the topic to: <span class="__mozilla-findbar-search" style="padding: 0pt; background-color: yellow; color: black; display: inline; font-size: inherit;">How to eval</span>uate the project<br>
<isforinsects> gregdek, %2e1<br><CBIgenho> did the desired results happen<br><hhardy> I want to keep an eye on things and expand on what works and triage what not works<br><cjl> evaluate -> see weekly meetings<br>
<gregdek> Yep.<br><gregdek> You might want to consider some metrics...<br><gregdek> ...but most things are pretty pass/fail.<br><hhardy> we will have a sense from meetings if formalisms are needed they will porbably evolve<br>
<cjl> also declare successes and admit failures to OLPC mgmt.<br><gregdek> "Did we get monitoring set up for these 5 hosts? No? Why not?"<br><CBIgenho> is this "project" goal driven?<br>
<gregdek> Very much so.<br><gregdek> The goal:<br><hhardy> s/porbably/probably/<br><gregdek> To ensure 100% uptime of all systems without hhardy having to do a thing. ;)<br><gregdek> With additional goals coming along all the time.<br>
<hhardy> to keep the ticket queue at a reasonable level<br><CBIgenho> see "volunteer" queue<br><hhardy> to expand the capabilities of the OLPC movement<br><gregdek> BINGO.<br><gregdek> Of course, that's not so obviously actionable. ;)<br>
<gregdek> There will be new tactical goals all the time.<br><gregdek> The purpose of the meeting is to drive those, and evaluate periodically whether they actually matter or not.<br>* hhardy has changed the topic to: How many people are enough?<br>
<cjl> general tasking to generate sub-project efforts in pursuit of technical improvements to OLPC "capacity" (ability to channel vol effort)<br><gregdek> cjl: Well put.<br><CanoeBerry> henry: anything else from your stated agenda for this mtg we need to cover quick? or later?<br>
<CBIgenho> what type of people/skills are you looking for?<br><gregdek> CBIgenho: sysadmins at first, I would think.<br><CBIgenho> we see all kinds on the Volunteer queue<br><gregdek> Narrow focus, strong results.<br>
<hhardy> CanoeBerry nothing too pressing that cannot be deferred<br><gregdek> So long as we recap action items, and owners, and deadlines. ;)<br><hhardy> yes this isn't a "we will train you" thing till we reach some critical mass<br>
<CBIgenho> we can alert you when we find good prospects<br>* hhardy has changed the topic to: final comments<br><cjl> CBIgenho: They may need to have been around long enough to be trusted "semi-insiders", not sure picking them out of the vol queue will necessarily be necessary <br>
<CBIgenho> how about I send a note when one that seems to be a possiblility?<br><hhardy> I am prepared to accept gregdek's recommendation of people in the existing fedora infrastructure project<br><hhardy> CBIgenho: good<br>
<isforinsects> I think that the need to be known *somewhere* and have someone vouch for them maybe<br><gregdek> Yeah.<br><isforinsects> but we let people into the S-G with no history<br><cjl> It is no so much the skills that are in short supply as minimizing the risks that OLPC incurs by sharing access at a deep level.<br>
<gregdek> Different risk profiles.<br><isforinsects> Some without even giving their name<br><cjl> isforinsects: not the same as giving root<br><gregdek> The risk profile of screwing up infrastructure is *very* high.<br>
<isforinsects> How about credit card numbers?<br><gregdek> Yeah, that's high too. :)<br><cjl> isforinsects: I give my credit card to a gas station attendent regularly.<br><CBIgenho> first-born child?<br>
<hhardy> how about accidentally deleting a few TB of data?<br><CanoeBerry> Webster says we are all going off to war (infrastructure = the permanent installations required for military purposes!)<br><gregdek> Job #1 for your infrastructure team: GET BACKUPS BULLETPROOF.<br>
<isforinsects> Of course this is a war!<br><CanoeBerry> Peace Out!<br><hhardy> webster, schmebster :)<br><gregdek> srsly. One of the most important things is to remove risk from the equation.<br>
<hhardy> gregdek: +1<br>
<gregdek> Backups are item #1.<br><gregdek> hhardy: this was covered in Mike McGrath's note to you, I think.<br><gregdek> Should I invite him to the next meeting, btw?<br><hhardy> gregdek: yes<br>
<CanoeBerry> Background on Mike McGrath anybody?<br><gregdek> Mike McGrath is the head of the Fedora Infrastructure team, for those who don't know.<br><gregdek> CanoeBerry: jinx. :)<br><gregdek> Used to run Linux for Orbitz. Now runs Fedora infrastructure for us. ;)<br>
* hhardy has changed the topic to: next meeting sunday at 2pm EDT, same Bat-channal<br><cjl> folks from projects like Fedora are to be highly valued, building bridges to other efforts and leveraging shared best practices is very impt.<br>
<gregdek> Fedora folks are psyched and ready to do so. :)<br><CanoeBerry> 2pm Sunday is INSANE, everything else above I agree with :)<br><cjl> I want to sucker wome wikipedians into helping with countervandal bots. . <br>
<hhardy> Kim is rounding up peoples for activities meeting shutting down this for now<br><gregdek> We will work out meeting onlist. ;)<br><gregdek> ACTION ITEMS?<br><hhardy> THANKS ALL!!<br><gregdek> ACTION ITEMS???<br>
<hhardy> oh action items<br><hhardy> lol<br><gregdek> :)<br><cjl> Join mailing list<br><hhardy> camp out on this chan<br><gregdek> My only job is to nag you for action items every week. ;)<br>
<hhardy> backups<br><gregdek> Can we make backups a focus of next week's meeting?<br><CanoeBerry> Action Item: Gang Colors Must be Chosen (and shirts/bandanas distributed pronto..)<br><cjl> to be fair, first meeting was about scoping (not setting actions)./<br>
<hhardy> send a more general announcement to christoph, et al<br><gregdek> Or Sunday's meeting, or whenever? :)<br>* hhardy has changed the topic to: ACTION ITEMS<br><CanoeBerry> SJ/is4: any action on shirt?<br>
<cjl> thanks all<br><hhardy> gregdek: yes see if you have someone with TSM exp pls we can do TSM to MIT just need the client installed and configured<br><hhardy> TYVM all<br><cjb> (TSM is "Tivoli Storage Manager")<br>
<isforinsects> CanoeBerry, no news from SJ<br><isforinsects> I may take over :P<br><isforinsects> cjl, working on a bot today :)<br>* reubencaron is now known as register<br>* register (~<a href="mailto:reuben@crank.laptop.org" target="_blank">reuben@crank.laptop.org</a>) has left #olpc-admin<div>
<div></div><div class="Wj3C7c"><div>
<div></div><div><br>
<br></div></div></div></div></div></div></div></div><br></div>