Web   ·   Wiki   ·   Activities   ·   Blog   ·   Lists   ·   Chat   ·   Meeting   ·   Bugs   ·   Git   ·   Translate   ·   Archive   ·   People   ·   Donate

#treehouse, 2010-01-19

Index | Today     Channels | Search | Join

All times shown according to UTC.

Time Nick Message
15:59 dogi #topic agenda
16:00 http://etherpad.com/ZNBbiFY2BA
16:00 something to add?
16:00 silbe can't read it :-P
16:01 dogi Last Meeting:
16:01 Log: http://me.etin.gs/treehouse/tr[…]0100112_1546.html
16:01 Minutes: http://me.etin.gs/treehouse/tr[…]0100112_1546.html
16:01 silbe is switching to laptop, brb
16:01 dogi Agenda:
16:01 * short munin overview for treehouse/housetree servers
16:01 http://sunjammer.sugarlabs.org[…]ugarlabs.org.html
16:01 http://sunjammer.sugarlabs.org[…]ugarlabs.org.html
16:01   #action fix housetree munin ...
16:01   Who knows how?
16:02 * update on pootleVM sayamindu
16:02 * aslo and RIT dfarning http://lists.sugarlabs.org/pri[…]nuary/thread.html
16:02 * service dns/ldap on lightwaveVM bernie
16:02 * ?new VM for OLE rowe? bernie
16:02 * redhat/fedora offers VMspace is that some for sugarVNC? silbe
16:02 ok lets wait for silbe ...
16:02 hui silbe
16:03 unmadindu hello
16:03 silbe re
16:03 dogi http://me.etin.gs/treehouse/tr[…]0100119_1559.html
16:03 hi unmadindu
16:03 alsroot hi all
16:03 dogi hi alsroot
16:04 ping _bernie basti lfaraone djbclark
16:04 #topic last meeting notes
16:04 something to add there?
16:05 lfaraone waves
16:05 dogi lfaraone, ciao
16:05 who wants to start?
16:05 unmadindu?
16:05 unmadindu ok :)
16:05 We have shifted Pootle to a new VM (thanks to dogi and bernie). We also upgraded to Pootle 2.0, which is almost a rewrite of the old Pootle. The new one is based on Django, and much better than the older one, in terms of both functionality and performance.
16:06 dogi #topic updates on PootleVM
16:06 unmadindu The major hurdle was migrating the permissions, since the Pootle developers had not provided a path to do that. I had to come up with a script of my own which has seemed to work (no one has complained yet ;-).
16:06 dogi so translate.sl.o is running on pootleVM
16:06 unmadindu yes
16:07 bernie changed the DNS last weekend
16:07 A lot of help wrt testing is also being offerred by cjl (Chris Leonard)
16:07 dogi #link http://translate.sugarlabs.org/
16:07 unmadindu We also have some of our own home-brewed features in Pootle 2.0, message validation (so that translators do not input wrongly formatted strings by mistake), and showing of commit status. The source is in http://git.sugarlabs.org/projects/pootle (and I'm trying to get the features upstream as well).
16:08 _bernie here I am
16:08 dogi #link http://sunjammer.sugarlabs.org[…]ugarlabs.org.html
16:08 _bernie aren't we early?
16:08 dogi go job unmadindu
16:08 unmadindu (currently we are using PPA to do the package management, which should make updates, etc easier)
16:08 dogi hi _bernie
16:08 unmadindu Currently, I'm migrating all the Pootle helper scripts (which keep the PO files in a sane and up to date state). I should be done with that by the end of this week.
16:09 The major problem initially was memory usage, but after a bit of tweaking and thanks to some suggestions from the Pootle devs, it seems to be under control (http://sunjammer.sugarlabs.org[…].org-memory.html)
16:09 dogi unmadindu, what are this helper scripts?
16:09 unmadindu dogi: cron jobs which pull in the latest source codes, ensure that the translation template files have all the latest stuff, etc
16:10 http://git.sugarlabs.org/projects/pootle-helpers
16:10 dogi so they connect to other servers?
16:10 unmadindu yes - they pull from git.sugarlabs.org
16:10 dogi ok
16:11 silbe unmadindu: can you adapt the helper scripts to build the pot file so it doesn't have to be in git, please?
16:11 dogi silbe, ?
16:11 what is the advantage
16:12 unmadindu silbe: sure could be done. however, we'll have to do a bit more tweaking. if your package does not have a POT file, I can special case it right now to work
16:12 silbe dogi: see e.g. http://bugs.sugarlabs.org/ticket/1225 for what breaks if you put files that are modified by a build into git
16:12 unmadindu dogi: sometimes the developers update the POT file as well, causing git conflicts
16:13 silbe unmadindu: i'm talking about sugar-*
16:13 dogi ahhh
16:13 ok
16:13 unmadindu silbe: ok, I think it can be handled
16:13 silbe unmadindu: thanks!
16:13 unmadindu CCs himself to 1225
16:14 dogi silbe, would u like help unmadindu in this? ...to late :) thx
16:14 other request opininons with pootle?
16:14 _bernie unmadindu:, dogi: it looks like the machine would benefit from one additional GB of memory (judging by munin's memory chart)
16:14 dogi bingo
16:15 forgot about that
16:15 silbe unmadindu: there are some more, e.g. #1414 (already assigned to you)
16:15 dogi tell us _bernie which section u speak about?
16:16 http://sunjammer.sugarlabs.org[…]s.org-memory.html
16:16 found it
16:16 _bernie dogi: yep
16:17 dogi the question is more how much?
16:17 #action dogi/bernie give more RAM to Pootle
16:17 unmadindu I think the memory usage will increase a bit when I add more projects to Pootle (very soon I'll add Sugar 0.86 and ASLO)
16:18 dogi does anybody should/wants to have access there?
16:18 unmadindu, sure
16:18 _bernie dogi: let's try 2GB and see if it's enough
16:18 dogi +1
16:19 next topic
16:19 _bernie dogi: ole's vm, right?
16:19 dogi #topic munin and housetree
16:19 unmadindu _bernie, dogi: thanks for all the help :)
16:20 _bernie unmadindu: thank _*YOU*_ for doing all the work
16:20 silbe unmadindu: i have to second _bernie
16:20 _bernie unmadindu: the link bar at the top looks great
16:20 unmadindu :)
16:20 stole from git.sl.o ;-)
16:21 dogi :)
16:21 http://sunjammer.sugarlabs.org[…]ugarlabs.org.html
16:21 does not work
16:22 silbe _bernie: BTW, any news from prgmr (or whatever the spelling was) about bs4 munin alerts?
16:22 dogi #topic treehouse
16:23 http://sunjammer.sugarlabs.org[…]ugarlabs.org.html
16:23 looks fine?
16:23 _bernie unmadindu: lfaraone wanted to unify the link bars... I think he liked gnome's black linkbar, but I guess we should use "sugar grey"
16:23 silbe: they answered with more questions, did you see that?
16:24 dogi good job
16:24 silbe _bernie: don't think so
16:24 _bernie dogi: looking
16:24 dogi both _bernie and unmadindu
16:24 unmadindu _bernie: sure - just send me the color values, and I'll update
16:24 dogi by the way where is chrowe
16:24 _bernie silbe: Luke Crawford said: "is it pingable, etc, during that time?   when did this start?"
16:25 silbe: I answered, and he never got back to me
16:26 unmadindu: I think you got it right already
16:26 silbe _bernie: ok, i guess you don't know about pings? maybe we should set something up to ping bs4 once per minute?
16:26 unmadindu _bernie: ah ok
16:26 dogi unmadindu, http://sunjammer.sugarlabs.org[…].org-libvirt.html
16:27 _bernie silbe: I said I'd setup another munin node to ping it
16:27 silbe: maybe you could do it?
16:27 silbe: you know, it's a munin plugin "ping_<hostname>"
16:27 silbe _bernie: still busy with the thesis. i could do a q&d script, though - that'll take much less time for me.
16:28 dogi who wants to be next?
16:28 hi rgs
16:28 unmadindu dogi: yeah Pootle _is_ CPU intensive. There's a lot of string manipulation using Python. There's an alternative module written in C, but it's a bit sloppy with memory
16:28 dogi :)
16:29 #topic lightwave
16:30 ping bernie
16:30 silbe that's the DNS/LDAP one, right?
16:31 _bernie dogi: pong
16:31 dogi http://sunjammer.sugarlabs.org[…]ugarlabs.org.html
16:31 hi chrowe
16:31 chrowe dogi: hi
16:31 _bernie I checked the munin graphs for treehouse. looks great
16:31 munin is a fantastic tool to find problems
16:31 dogi chrowe,  http://me.etin.gs/treehouse/tr[…]0100119_1559.html
16:32 _bernie, +1
16:32 silbe automatic update running on lightwave? (wondering about the eth0 spikes)
16:33 dogi silbe, can u post the link there?
16:33 silbe dogi: ?
16:33 _bernie silbe: we should have automatic security updates everywhere (but I haven't checked if they work)
16:33 dogi #link http://sunjammer.sugarlabs.org[…].org-if_eth0.html
16:34 that?
16:34 lfaraone unmadindu: have we looked into replacing these expensive string operations with optimized alternatives?
16:34 silbe _bernie: I think I never figured out how to do it for Fedora, so the F* buildslaves are not auto-updating :-/
16:34 dogi _bernie, think thats the ubuntu heartbeat
16:35 unmadindu lfaraone: yeah - the alternative leaks memory. I'm trying to see if that can be fixed
16:35 _bernie silbe: there's a yum plugin iirc
16:35 silbe (F* = Fedora 11+12, not f-word)
16:35 _bernie: pointer?
16:36 dogi _bernie, how long do u think lightwave has to run stable until we will turn this to the master dns?
16:36 _bernie silbe: searching
16:36 dogi: I think we're fine... it's just that I had no time to switch over all the slaves and the domains
16:37 lfaraone unmadindu: well, I mean optimizing the existing python code.
16:37 dogi ok _bernie
16:37 unmadindu lfaraone: it has been optimized quite a lot over the last few releases, but even then it is slow and cpu expensive (especially for large files with > 1000 strings)
16:38 _bernie unmadindu: all python web applications seem to be super-heavy
16:38 unmadindu yeah :(
16:38 _bernie unmadindu: trac is giving me trouble on multiple hosts
16:39 unmadindu switched from trac to redmine recently for his personal project hosting stuff
16:39 dogi #action lightwave -> master dns? dogi/bernie
16:40 chrowe, u want to be next?
16:41 chrowe dogi: sure
16:41 What should I say?
16:41 _bernie dogi: I guess I'll have to do most of the work
16:41 dogi #topic oleVM?
16:41 silbe _bernie: how does it work currently? what machine is master, how do we transfer to slaves (AXFR+notify)?
16:41 _bernie chrowe: is ole.net hosted by our nameservers?
16:41 chrowe: nope
16:41 silbe is too slow again :-/
16:41 _bernie silbe: ns1.codewiz.org (aka trinity.develer.com) is our master
16:42 chrowe ole.org is our domain
16:42 and we are using godaddy for dns
16:42 we have other domains running on sunjammer as well
16:42 _bernie silbe: all slaves use axfr  and notifies with IP-based authorization (me bad)
16:42 silbe _bernie: kthx
16:42 chrowe right now we are in need of at least one VM if not 2
16:43 _bernie silbe: it would be cool to switch to dnssec, but it's such a complicated protocol, I have no time/willingness to study it
16:43 silbe: perhaps we could live with a symmetric key and IP authorization
16:43 silbe _bernie: i don't think it's a good idea anyway, but that's a somewhat open-ended topic :)
16:43 dogi for me the question is more for what will u use them
16:43 chrowe unless we can use sunjammer for testing our applications
16:44 silbe _bernie: I've been using ssh for DNS updates for ages, but then I usually have root on most of the slaves :)
16:44 chrowe dogi: 1 we need for migrating our library.ole.org stack to so we can package it as a VM
16:44 _bernie chrowe: the original purpose of sunjammer was to be a shell server for developers and host applications temporarily before moving them to proper VMs
16:44 chrowe dogi: The other is just for testing out other library and repository stacks
16:45 dogi #link http://library.ole.org/
16:45 ups
16:45 chrowe we want to test out the software that runs cnx.org and curriki.org
16:45 silbe chrowe: what do you actually plan to do on those VMs? Just build stuff? Install the software locally to try it out? ...
16:45 _bernie chrowe: eventually, sunjammer became our best maintained machine, and after the move it was also performing well and quite reliable, so we never got around to split services around
16:46 lfaraone chrowe: a friend of mine works at curriki, they might be able to help with that.
16:46 dogi chrowe, can u plz post always the links
16:46 chrowe cnx.org is based on plone and curriki is based on xwiki
16:47 dogi: sure
16:47 djbclark (when there is a good pause, I can talk about hosting opportunities I know of)
16:47 dogi cool
16:48 chrowe http://cnx.org/ runs on http://rhaptos.org/
16:48 _bernie djbclark++
16:48 chrowe http://www.curriki.org/xwiki/b[…]view/Main/WebHome runs on http://curriki.xwiki.org/xwiki[…]view/Main/WebHome
16:48 dogi thx chrowe
16:49 so to you want more test/devel?
16:49 ups
16:49 chrowe so if it is ok to install those packages on sunjammer for testing that would work for me
16:50 for http://library.ole.org we really need our own VM so we can package it up for other to use.
16:50 dogi so do you want more test/devel ok
16:50 _bernie chrowe: in my past experience, these python web frameworks tend to be very invasive and are often not packaged in distros.
16:50 chrowe we have very low traffic at this point so we don't need much in the way of resources
16:51 _bernie chrowe: are you sure we can find the things you need on ubuntu jaunty?
16:51 silbe chrowe: sounds like a good fit for a VM then
16:51 dogi silbe, +1
16:51 hi scyrma
16:51 _bernie chrowe: if it's the same for you, I'd suggest using an independent karmic vm where you can even give root to any helpers
16:52 chrowe _bernie: that is would be great
16:52 _bernie scyrma: http://me.etin.gs/treehouse/tr[…]0100119_1559.html
16:53 chrowe scyrma is one of our developers so he may have some insight into this
16:53 scyrma hi
16:53 sorry, I'll have to keep lurking silently for a few more minutes
16:53 dogi #action oleVM bernie/chrowe
16:54 chrowe, i like to have documentation somewhere :)
16:54 is that fine with u?
16:54 _bernie lfaraone: my contacts at mit failed: <danjared> _bernie: How permanently? As much as I'd like to, I can't make colo space appear out of nowhere.
16:54 chrowe dogi: documentation is good, just let me know where you would like it?
16:55 dogi _bernie, knows :)
16:55 lfaraone _bernie: maybe djbclark has alternative options.
16:55 dogi: shall we switch to new hosting locations and machines?
16:55 dogi jupp djbclark
16:55 chrowe dogi: we are using Redmine and could post it there with a link to the sugarlabs wiki, or vise vera
16:55 lfaraone (change the topic, that is)
16:55 _bernie chrowe: we document our machines like this: http://wiki.sugarlabs.org/go/I[…]esources#Machines
16:55 dogi #action hosting possibilites
16:55 djbclark now? Okay, so I can def get servers in at BUilds, BU's new hackerspace where the guy running it is a good friend. I'd have a RFid and probably access whenever (there is an external door bur but I am told there is almost always at least one open). But IP addresses > 1 might be hard, and currently only 10baseT but they may get update to 100baseT soon (but BU admin is paranoid about bandwidth use, so
16:55 prob best not to put anything that intense there). For MIT at the moment best I could def get would be a friend's desk at CSAIL for several months guranteed (until Phd defense :-). The guy I knew well there resigned and is going to work at Yahoo in Cali.
16:56 _bernie chrowe: that would work, sure. put it where you prefer and link it from the other location
16:56 dogi #action hosting possibilites
16:56 lfaraone djbclark: okay. as I mentioned, we'd probably be hosting ASLO on these machines. They are rack servers.
16:57 dogi: I think you mean #topic
16:57 dogi #topic hosting possibilites
16:57 lol
16:57 is still human :P
16:57 djbclark lfaraone: what is the bandwidth footprint of those machines like? There is a ton of unused rackspace at BUilds, so rack-mount not a problem.
16:58 _bernie djbclark: a friend at CSAIL? does this friend have a long beard?
16:58 djbclark lfaraone: it's possible we'll like end up with some kind of radio link with MITERS, but that wouldn't be for a whle.
16:58 _bernie djbclark: and maybe likes parrots too? ;-)
16:58 djbclark _bernie: no; that friend might be up for it, and would be longer term, but he has rather stringent hardware and software requirements :)
16:59 lfaraone djbclark: _bernie would know best, but imagine the entire student body of Uruguay connecting to your computer at once :)
16:59 djbclark _bernie: eg PMON or coreboot, Trisquel of gNewSense.
16:59 dogi by the way lfaraone u are next with the update about the 12 wikimedia machines
16:59 lfaraone, :)
16:59 lfaraone dogi: that's what we're talking about racking :)
17:00 djbclark _bernie: also he's into privacy, so doubt we'd be able to get office keys; so server could be down for 3-4 weeks worst case.
17:00 _bernie: whereas friend I am thinking of isn't even going to be in the country, and I'll be looking after other boxes for him anyway.
17:01 IMHO BUilds would be best for a box that either mostly does monitoting/config mgmt or lots of processing but little communications (eg build machines)
17:02 dogi djbclark thx will think about sound interresting :P ... let switch to lfaraone
17:02 lfaraone djbclark: _bernie tells me we already have enough build machines.
17:03 silbe at least until i add another bunch of VMs :-P
17:03 _bernie silbe: yum-plugin-security.noarch
17:03 dogi silbe, where are the buildslaves hosted?
17:03 silbe _bernie: thx!
17:03 djbclark lfaraone: yeah the question is are the exisinting build machines in hosting locations that can stand more network i/o... if so perhaps would make sense to shuffle stuff around.
17:04 silbe dogi: Develer (bernie's old company)
17:04 dogi ahhhh
17:04 ok :)
17:05 lfaraone again defers to _bernie
17:05 _bernie djbclark: we may need to relocate bender from italy one day, I'll take BUilds into account for then
17:05 djbclark: for the aslo cluster, we need to look elsewhere
17:05 dogi lets hope in RIT :)
17:05 _bernie dogi:  http://www.develer.com/
17:06 dogi: yeah, let's hope. I'll try to get on a confcall with them... since they love telephones so much... as you know, I totally hate them
17:06 they are from 1800
17:06 lfaraone _bernie: if you can give me a short writeup of our needs, I can do the confcall with RIT.
17:07 is better on the phone than he is in real life :)
17:07 dogi _bernie, conference call which number an when?
17:07 _bernie #link http://en.wikipedia.org/wiki/F[…]896_telephone.jpg
17:07 lfaraone dogi: we don't know yet, they just said "we should do it sometime"
17:07 dogi ahh
17:07 _bernie lfaraone: sure, you would rock
17:08 lfaraone: you're very professional on the phone.
17:08 lfaraone _bernie: okay then. dogi, we have an #action item :)
17:08 dogi can somebody tell me something about rgs and scyrma  :P
17:08 is curious
17:08 scyrma is a developper working with chrowe on ole sites
17:09 _bernie lfaraone: I guess you know already what our needs are: rack 3 (4) machines, possibly with independent public IPs and a private lan interconnecting them. IPv6 is a big plus but not a  showstopper
17:09 scyrma ...and quite interested in VMs :)
17:09 dogi scyrma, :)
17:09 welcome scyrma
17:09 _bernie lfaraone: the machines need to be installed with our existing aslo-* images, which can be done by clever hackery or by installing karmic and then rsyncing the changes on top of it
17:10 dogi: rgs is my boss here at pyeduca
17:10 dogi: he's checking on me :)
17:11 lfaraone basti: can we give them a bare-bones xen hypervisor, and have ASLO be the sole VM? (so we can redeploy new configurations easilly)
17:11 * _bernie
17:11 _bernie: or is that too much overhead?
17:11 silbe lfaraone: do you know if there are free slots in the WMF servers?
17:11 dogi hi rgs
17:11 lfaraone silbe: I do not. You mean for HDDs?
17:12 dogi _bernie, :)
17:12 silbe lfaraone: for ethernet cards (so we could daisy-chain them if there are no switch ports left for a private LAN segment)
17:12 _bernie silbe: yum-updatesd provides a daemon which checks for available updates and
17:12           : can notify you when they are available via email, syslog or dbus.
17:12 silbe: the yum security thing is just for telling security updates from other updates
17:12 silbe: I also have a cron script which I use to send bug mail about pending rpm updates. do you want to install it?
17:13 silbe _bernie: ah, ok. so i still need to do the updates manually? no unattended upgrades like Debian can do?
17:13 _bernie silbe: you can find a copy in /usr/src/devtools/sysadm/fedora/ of any machine I touched
17:13 dogi #topic update 12 wikipedia servers
17:13 lfaraone silbe: very probably.
17:14 silbe lfaraone: great!
17:14 _bernie silbe: no, I think there was also something to run the updates for real... but I never trusted it so I kept using my old cronjob
17:14 lfaraone It looks like total it'll cost us 220USD to ship the servers are we currently planned. 6x to DC (APS, TWIX, Obscure), 3x to RIT, and 3x to Walter for stashing somewhere in FSF/MIT/etc.
17:15 _bernie lfaraone: both xen and kvm are low overhead for many VMs, but for just 1 VM it doesn't make sense to use them
17:15 silbe _bernie: at least for the buildslaves i'd say let it * the system ;)
17:15 lfaraone (make that 224)
17:15 _bernie lfaraone: you end up using almost twice as much memory and there's a lot more unneeded complexity and configuration work to be done
17:15 lfaraone: (you'd need 2 IPs, etc.)
17:16 dogi I m always for VM on every machine cause so u can ask for more IPs
17:16 lfaraone Since RIT is currently holding us up on shipping, I'll confcall with them to make sure they can meet our needs. Otherwise, we can send 6 to MA and 6 to DC.
17:16 dogi and can later bring a better server
17:17 lfaraone Does anybody have alternative locations we should consider sending these to? djbclark? silbe?
17:17 silbe lfaraone: not within the US I'm afraid
17:17 and international shipping is going to get prohibitive I guess
17:18 _bernie dogi: those are really small machines
17:18 silbe lfaraone: thx for doing the phone call!
17:18 _bernie dogi: and they already told us they don't want to give us many IPs anyway :-(
17:18 silbe hates phones as well
17:18 _bernie dogi: they == rit
17:18 dogi _bernie, i know
17:19 lfaraone silbe: no problem.
17:19 _bernie lfaraone: I'm sure walter's house would also work as a temporary storage, but we risk paying shipping twice
17:19 lfaraone _bernie: yep. or having somebody drive them down <_<;
17:20 _bernie silbe: what do you mean by "let it * the system" ?
17:20 lfaraone _bernie: so we're #agreed that I'll contact RIT? If they agree to rack what we have, can I tell WMF we're ready to "order", or should we wait till next meeting?
17:20 _bernie lfaraone: iirc, rit is 8 hrs drive away from boston. a little bit too much I guess
17:21 dogi could do this ... if i get a car :P
17:21 silbe _bernie: *=ruin/blast/whatever
17:21 _bernie <_bernie> danjared: very permanently. so I guess it's no game.
17:21 <danjared> _bernie: yeah, probably so. I could hold onto them for a bit in the same style as housetree, but can't give any guarantees on anything else
17:21 <danjared> _bernie: you'd have to pursue the same routes I recommended last time
17:21 silbe: haha yes
17:22 lfaraone: I think you have my approval. and walterbender's credit card
17:23 lfaraone: you only miss sunglasses to be a perfect yuppie :)
17:23 dogi but i liked last time plan send 6 to DC and 3 to MA (so that dogi can look onto them ...) and 3 to RIT
17:23 silbe hmm, maybe we could add an OpenRD on top of it if walter's paying ;)
17:23 _bernie still does not understand why 6 to DC
17:23 dogi <_bernie> lfaraone: I think you have my approval. and walterbender's credit card .....hui
17:24 ok bernie 6 to rit?
17:24 or 6 to MA
17:25 _bernie dogi: a clarification: shall we create 1 or 2 vms for http://ole.org/ ?
17:25 dogi: we said 3 to RIT
17:25 dogi, lfaraone: what are the DC machines for?
17:25 silbe i think they'd panic f we ship them 6 :)
17:25 _bernie (unless we're still in touch with twix
17:26 silbe: they're supposed to be a research computing lab
17:26 silbe: one would expect them to have seen computers before
17:26 silbe: http://wiki.sugarlabs.org/go/F[…]rlabs_systems.jpg
17:26 silbe _bernie: not if they're any like ours :-/
17:27 _bernie: the one with the big wheel? :)
17:27 dogi #topic IdeaVM
17:27 lfaraone _bernie: like I said, TWIX, ACC, and Obscure.org.
17:27 _bernie silbe: great careers!
17:28 dogi ups sorry
17:28 lfaraone _bernie: of which, TWIX is the most equipped to handle the load.
17:28 dogi #topic olpcorps.org and ideaVM
17:28 silbe lfaraone: do you think they'd be willing to host a.sl.o?
17:29 dogi #link http://olpcorps.org
17:29 dogi is searching volunteers to reactivate this AGPL service
17:30 _bernie lfaraone: oh, TWIX is still a possibility? I'd send them *all* the machines if they'd take them :-)))
17:30 lfaraone: remind me to talk to you about an offer to work on the web site by a local colleague, btw
17:31 rgs dogi: hi dogi, sorry for the intrussion! curiosity..
17:31 :)
17:31 lfaraone _bernie: will do.
17:32 dogi #topic other
17:32 sure rgs
17:32 lfaraone _bernie: yes, they are still a possibility, but I don't think they can host ALL of them.
17:32 dogi +1 lfaraone
17:33 lfaraone _bernie: I was told they can handle at least 3U, maybe more.
17:33 _bernie lfaraone: do you still have the skatch of what you wanted to do?
17:33 silbe lfaraone: if they'd host at least 3 (i.e. a.sl.o), that would be awesome!
17:33 _bernie lfaraone: since he's also on jabber, I'll pass you his jid
17:33 lfaraone _bernie: yeah, it's a SVG now :)
17:34 _bernie: I asked my contact about giving us 12U, his reply was "lmao".
17:35 _bernie lfaraone: just talked with this guy (cesar aka "saddor").
17:35 lfaraone: what's "lmao"?
17:35 lfaraone _bernie: also, same contact might be able to get us a VM + b/w at a cloud provider he works for.
17:35 _bernie silbe: you know, we could get rid of bender if we moved the vms to housetree. dogi, would you approve?
17:36 lfaraone _bernie: "laughing my ass off". a higher form of lol.
17:36 _bernie silbe, dogi: those vms use a lot of cpu time though
17:36 we can nice them
17:36 silbe _bernie: do we trouble develer with that machine?
17:36 _bernie lfaraone: you never finish learning l33t talk
17:37 lfaraone: now you also need to unlearn it before you can go to business school with your iphone and sunglasses
17:37 lfaraone _bernie: (another Dulles, VA IT group: http://opsource.com/ )
17:37 dogi _bernie, one step after an other ... lightwave first :P
17:37 _bernie silbe: not at all... I was just trying to consolidate our infrastructure on fewer hosts if possible
17:38 silbe _bernie: i'd prefer to keep it in IT, sunjammer's got pretty high RTT
17:38 dogi the 3 to MA i asked walter if I could help him there
17:38 _bernie silbe: and it disturbs me to depend on a machine which I can't physically access any more (although some of my best friends are taking care of it)
17:38 silbe: ok then
17:39 dogi and he agreed thats why I am so curious about to know
17:39 silbe _bernie: but if it really bothers you we can move then - though i'd hate to bog a server doing other useful work
17:39 _bernie lfaraone: ok then let's ship some (6? 3?) of those to your place and then we'll decide where to rack them
17:39 silbe: no, it was just a very minor issue
17:39 silbe _bernie: ok
17:40 _bernie silbe: they even moved bender to a better rack case while I wasn't looking!
17:40 lfaraone _bernie: so I should tell the WMF folks "we'll take half of them now, half later"?
17:40 _bernie: I might be able to get them to ship 6 to DC, 3 to MA, and hold on to the remaining 3 for a bit until RIT gets their things in order.
17:40 _bernie silbe: fwiw, we could also have kept the master dns on trinity, but I wanted to make the Sugar Labs infrastructure less dependent on me
17:40 dogi +1 lfaraone
17:41 _bernie silbe: so others can change the dns (you would be my #1 hostmaster candidate as you're very paran^Wconsiderate about security)
17:41 lfaraone _bernie: what should I tell TWIX our new bw reqs are?
17:41 _bernie silbe: and you live in a different TZ
17:41 silbe _bernie: DNS is low resources, high importance, so makes a lot of sense to host somewhere with at least some physical access
17:41 _bernie lfaraone: more or less the same, 'cause we wanted to host aslo there anyway
17:42 lfaraone: but I guess it wouldn't hurt doubling them just in case
17:42 lfaraone _bernie: will do.
17:42 silbe _bernie: buildslaves OTOH are low-importance, high (local) resources, so perfect for some warehouse server :)
17:43 _bernie silbe: yeah
17:43 silbe: we can reconstruct them easily
17:43 dogi what should i write into #action ?
17:43 _bernie silbe: actually, bender has been incredibly useful for me to learn how to manage many vms with libvirtd
17:44 silbe: and libvirt in fedora is still more mature/polished than it is on ubuntu
17:44 lfaraone dogi: #action lfaraone to request shipping of 9 machines to MA/DC. #action lfaraone to f/u and confcall with RIT folks
17:44 _bernie I need to drop off this meeting
17:44 dogi #action lfaraone/bernie/dogi to f/u and confcall with RIT folks
17:44 #action lfaraone to request shipping of 9 machines to MA/DC
17:45 +1 _bernie
17:45 me too
17:45 last words?
17:45 _bernie <_bernie> dogi: a clarification: shall we create 1 or 2 vms for http://ole.org/ ?
17:45 dogi: my last words: this meeting rocked, thanks for chairing it
17:46 dogi start with one ...
17:46 thx _bernie
17:46 #endmeeting

Index | Today     Channels | Search | Join

Powered by ilbot/Modified.