19:00 < jdrake> Meeting time?
19:00 < Smerdyakov> Sure looks like it.
19:00 < ntk> yes
19:00 < mwolson> present
19:00 < Smerdyakov> ntk, are your official logs started?
19:01 -!- You're now known as ntk_logger
19:01 < Smerdyakov> OK, then I bring this meeting to order!
19:01 < megacz> "do you swear to tell the truth, the whole truth, and nothing but the truth"?
19:01 < ntk> can we take attendance?
19:01 < Smerdyakov> Let's have everyone say his HCoop username, by way of roll-taking.
19:01 < Smerdyakov> adamc
19:02 < megacz> megacz
19:02 < docelic> docelic
19:02 < jdrake> jdrake
19:02  * mwolson 
19:02 < ntk> ntk/Nathan K
19:02 < jroes> jroes
19:02 < iriefrank> frank
19:02 < mattam> mattam
19:02  * mwolson wonders if empty "/me" statements even get sent
19:02 < Smerdyakov> mwolson, they do.
19:02 < ntk> yes, thanks
19:02 < ntk> is that it?
19:03 < Smerdyakov> We should probably go on, in any case.
19:03 < Kuril> ryan
19:03 < ntk> yes, justin can't make it
19:03 < Smerdyakov> I guess official appointment of new officers is easy and should go first.
19:03 < ntk> yes
19:03 < Smerdyakov> docelic, ntk, do we want to change our votes from last year?
19:03 < ntk> i move we retain the same roles
19:03 < docelic> mwolson, depends on the client 
19:03 < Smerdyakov> I move thusly in addition!
19:03 < ntk> done
19:03 < ntk> sorry docelic
19:04 < docelic> Tough love from you two ;)
19:04 < Smerdyakov> OK, now, I think the dead disk is next.
19:04 < docelic> Yes,
19:04 < ntk> if the disk isn't up, then it can probably be hotswapped per docelic
19:04 < Smerdyakov> I will submit a support request shortly.  Would someone please tell me exactly what to write in it to describe the problem? :P
19:04 < ntk> and it should be dead-easy to identify
19:05 < mwolson> "Fix our disk.  Now."
19:05 < Smerdyakov> OK, I guess the message I'm getting is that this isn't so complicated.
19:05 < ntk> how's this.   "Dear Interserver folks, The first disk on fyodor.hcoop.net, /dev/hda1, is dead.  Please hotswap it with an identical model
19:05 < ntk> and let us know when it is done."
19:05 < Smerdyakov> If they say there will need to be downtime, anyone disagree that we should warn members with sufficient notice?
19:06 < mwolson> we should indeed warn members
19:06 < Kuril> ntk, it would only be /dev/hda.  /dev/hda1 is the first partition of hda.
19:06 < jdrake> Members don't need to know!
19:06 < Kuril> or the second.
19:06 < ntk> no.  there will probably just need to be a brief switch to single-user mode to redo it
19:06 < Smerdyakov> There was also a suggestion that they might need login access.  Should we create such an account only if they ask?
19:06 < ntk> er
19:06 < ntk> to put it online
19:06 < ntk> the rebuilding of /dev/hda is done in the background
19:07 < ntk> if they can do it with the system up then they don't need login access at all.  i only mentioned that if we wanted them to be able to safely shut it down without us intervening
19:07 < Smerdyakov> OK.  I will wait for requests from them before suggesting account creation.
19:07 < docelic> Yes, the disk itself should not incur any outage. Outage will happen only if the machine does not support hotswap.
19:08 < ntk> (Kuril: the 1 came out as an odd tic)
19:08 < Smerdyakov> So now we have a fork in the road.  Y'all can discuss amongst yourselves while I submit the request, or we should look to other business and I will submit the request when we're finished.
19:08 < mwolson> if they are granted login access, and want to run "shutdown" as root, they should have access only to that command (via /etc/sudoers), and only until everything is up
19:08 < ntk> yes, that's exactly what i had in mind
19:08 < mwolson> s/access/root access/
19:08 < docelic> No.. listen.
19:08 < jdrake> Is it in anyway possible for this to be extended downtime?
19:09 < docelic> If they need to turn off the machine, they can just press the power off button, and ACPI will trigger software shutdown. In the case that ACPI isn't enabled, they can press ctrl+alt+del on the console turn it off.
19:09 < docelic> So they don't need access.
19:09 < ntk> jdrake: it's always "possible."  if they swapped the wrong drive and trashed it, we'd have extended downtime, but we'd be incompetent
19:09 < Kuril> what about ACPI button state?  If that's in the kernel, then they only have to press the off button and the system will initiate shutdown
19:09 < ntk> ok
19:09 < ntk> good point
19:09 < mwolson> docelic: true
19:09 < ntk> anyway
19:10 < Smerdyakov> OK, guys, should I submit it now or after the meeting proper?
19:10 < ntk> the only concern i have is, is the mbr and /boot on /dev/hda,
19:10 < ntk> or is that mirrored too
19:10 < ntk> i guess it must be
19:10 < ntk> since it came up
19:10 < jdrake> If there is the possibility of that, is there alternative arrangements we can make for DNS? (I only suggest this as I have a domain that is reporting an ip of my home machine, so I can easily do webdev).
19:10 < ntk> and hda is down right now
19:10 < mwolson> i don't care when it is sent, just that it is sent
19:10 < ntk> s/we'd/they'd
19:10 < Smerdyakov> jdrake, Abulafia is running fine and should be your secondary if you followed our instructions.
19:11 < Smerdyakov> OK, no one but mwolson is even acknowledging my question, so I guess we're moving on.
19:11 < jdrake> Smerdyakov: ns and ns2 I believe I have setup. I do not recall. Otherwise 127.0.0.1 is nice.
19:11 < ntk> alright, let's talk about peer1 now
19:12 < mwolson> what in particular about peer1 needs discussion?
19:12 < ntk> keep things moving along
19:12 < jdrake> Is peer1 the new or old?
19:12 < ntk> jdrake: new
19:12 < Smerdyakov> We have this "where do log files live" issue.
19:12 < Smerdyakov> Correlated is my uncertainty about whether it makes sense to use mod_waklog anymore, given this new understanding about logs.  For that, I ask megacz's opinion.
19:12 < megacz> could we put them on local disk and consider moving them to AFS after the migration?
19:13 -!- ali4728 [n=ali4728@adsl-68-127-208-205.dsl.pltn13.pacbell.net] has joined #hcoop
19:13 < ntk> I would like a brief report on what is working and what is blocking on deleuze & mire right now.  e.g. of the portal, domtool2, apache, exim, bind.
19:13 < Smerdyakov> Hm... I want to have centralized log statistics on the portal.
19:13 < megacz> the issue isn't mod_waklog -- it's that Apache assumes that the logs are stored on a filesystem that does permission checks on open()
19:13 < megacz> AFS is not such a filesystem
19:14 < Smerdyakov> megacz, but, you see, without waklog, the Apache processes can stay authenticated as www-data throughout their lifetimes.  It doesn't matter if additional checks occur.
19:14 < megacz> mod_waklog never calls setuid()
19:14 < Smerdyakov> megacz, I'm talking about AFS credentials.
19:15 < Smerdyakov> The default suexec approach to shared Apaches never has any process switch between protection domains, except for suexec wrappers that switch to users on startup.
19:16 < ntk> mod_waklog is what allows apache to run as an afs user?
19:17 < megacz> ntk, yes
19:17 < Smerdyakov> ntk, one tool of that kind, yes.
19:17 < ntk> but we're only allowing CGI scripts at peer1?  no mod_perl/mod_php etc?
19:17 < Smerdyakov> We haven't allowed mod_perl/mod_php in a long time, and there's been no motion to change that.
19:17 < ntk> okay.
19:18 < Smerdyakov> megacz, does my description at least make sense now?
19:18 < jdrake> Why don't we use fastcgi? It has the speed advantages of mod_php/perl yet will use separate daemons for processes.
19:19 < megacz> if you guys want to go this way, it's cool, but I personally would not be willing to support such a feature
19:19 < Smerdyakov> jdrake, off-topic.  Easy to add support.
19:19 < ntk> i agree, let's get running first
19:19 < megacz> no, it's actually not
19:19 < ntk> whether or not it is easy
19:19 < megacz> one of the problems you're going to run into if you k5start every single CGI request is that it will be agonizingly slow
19:20 < Smerdyakov> megacz, how does waklog avoid that?
19:20 < megacz> it caches tokens
19:20 < megacz> very intelligently, too
19:20 < mwolson> that sounds like a win
19:21 < Smerdyakov> megacz, given that users are already used to process-start overhead via suexec modules, how much change will they notice?
19:21 < ntk> so what exactly is this log issue, I thought we had decided where to put them
19:21 < megacz> the same delay you notice running "kinit;aklog"
19:21 < Smerdyakov> megacz, I don't generally notice any!
19:21 < megacz> If we keep the logs on local disk as a temporary measure and rsync them to some central location every ~60seconds, would that work?
19:22 < Smerdyakov> That would be fine with me.
19:22 < mwolson> likewise
19:22 < Smerdyakov> As long as I'm not architecting it.
19:22 < Smerdyakov> There are other current direct AFS uses that I wouldn't mind doing that way, too, as long as someone else does the admin legwork.
19:22 < ntk> every 60 seconds seems like overkill, maybe every 10 minutes?
19:23 < ntk> but anyway
19:23 < mwolson> we should have a simple (and easy) try-fail locking mechanism for that, is my only concern
19:23 < megacz> BTW, I see this only as a temporary measure.  Long term we really ought to have a separate process that Apache pipes its logs to -- that process can do much more intelligent things.
19:23 < megacz> (one process per machine, not one per user)
19:23 < Smerdyakov> megacz, does Apache's module system allow you to make such changes?
19:23 < docelic> The k5start/aklog takes below 0.020 seconds on idle system
19:23 < megacz> Smerdyakov, changes==log-piping?
19:23 < Smerdyakov> megacz, yeah.
19:24 < megacz> No modules needed; there's a directive that says "instead of sending logs to disk, fork a process and pipe them to its stdin"
19:24 < Smerdyakov> Oh, neat.
19:24 < ntk> mwolson, you are the Apache guy on TAskDistribution, can you do set up this save-to-disk-then-rsync system?
19:24 < docelic> Wait. Why not go with log piping from the start ?
19:25 < mwolson> ntk: i could do that, sure; though i want to hear the result of log piping discussion first
19:25 < ntk> ok
19:25 < megacz> docelic, possible; i'm just concerned it might take more than a day or two, and I'm pretty busy this week
19:25  * Smerdyakov imagines writing a log piping program that sends docelic a text message on his phone whenever something interesting happens. ;)
19:25 < megacz> =)
19:25 < ntk> my only concern is that it will turn into another grand initiative, especially if it requires writing software that does not yet exist
19:25 < megacz> ntk, yes
19:25 < docelic> Smerdyakov, in fact, I have this in mind ;)
19:25 < ntk> we don't have time for any more grand initiatives
19:26 < Smerdyakov> I will go along with whatever the admins want to do, with more say going to people who volunteer to do more work. ;)
19:26 < megacz> the log piper would mainly need to (1) kinit;aklog as the "i-write-to-logfiles user", (2) figure out which user each log-line belongs to, (3) write said log-line to the appropriate logfile in afs
19:26 < docelic> Ok. so be it rsync. Once you get the parameters right, cron job will take care of everything.
19:26 < Smerdyakov> I just need to know the function for calculating which Apache config lines go into each vhost to handle logging.
19:26 < ntk> rsync is an intense process, and there can be a lot of logs, why 60 seconds?
19:27 < ntk> crap
19:27 < ntk> sorry pasted the wrong line ignore
19:27 < docelic> 60 seconds was just idea.. we can run rsyncs for every 10 sites in a batch
19:27 < megacz> Smerdyakov, in the rsync setup we can have Apache put them in separate files and rsync the files
19:27 < megacz> I don't think we need to parse loglines until we move to the log-piping mechanism
19:27 < Smerdyakov> megacz, you're saying there will be no vhost configuration lines about logging?
19:28 < Smerdyakov> megacz, like CustomLog/ErrorLog in the current fyodor system
19:28 < docelic> yeah, it would be ideal if apache would output the directory structure that should be in afs, so we just directly rsync it
19:28 < ntk> megacz: is that something that can and will realistically be written, tested, and working in half a day?
19:28 < megacz> With rsync: there will be configuration lines.  With log-piping: there will not.
19:28 < ntk> otherwise, I am against going with that scheme
19:28 < Smerdyakov> megacz, OK, so I need to know what the lines should be! :D
19:28 < megacz> Smerdyakov, use whatever lines you would use for AFS, but s_/afs_/var/logs-to-be-rsynced_
19:28 < Smerdyakov> megacz, yes, yes, I need to know the exact paths.  That's all I'm saying.
19:29 < megacz> Oh, ok.  Suggestions anyone?
19:29 < docelic> s/afs/var\/log/apache2/ ?
19:29 < megacz> I second that.
19:29 < Smerdyakov> No point in using them until it's implemented, but eventually this info is required. :)
19:29 < docelic> same directory structure, same everything, just relocated to /var/log/apache2/
19:30 < megacz> Okay, I propose the motion: "we agree to go with rsyncing logfiles periodically, and to move to log-piping at some point after the user migration"
19:30 < mwolson> and what's the corresponding AFS location?
19:30 < megacz> mwolson, you mean with log-piping?
19:30 < docelic> ~user/logs/apache2/ 
19:30 < mwolson> megacz: i mean with rsync-ing
19:31 < docelic> mwolson, ah, no
19:31 < megacz> I like docelic's suggestion; somewhere in ~user...
19:31 < Smerdyakov> The layout Domtool assumes now involves ~user/apache/logs/SERVER_NAME/VHOST_NAME/error.log, etc..
19:31 < docelic> As I said, we'll make the same structure, just relocated
19:31 < docelic> so mwolson , the right rsync line
19:31 < mwolson> which is /afs/hcoop.net/user/apache/logs/... ?
19:31 < Smerdyakov> E.g. ~adamc/apache/logs/mire/www.schizomaniac.net/error.log.
19:32 < docelic> would be something like rsync from /var/log/apache2/ to /afs/hcoop.net/user/
19:32 < docelic> from those points in filesystem, paths will match
19:32 < mwolson> docelic: thanks, that is what i wanted to know
19:33 < ntk> okay
19:33 < megacz> btw, with log-piping we can do any user-proof bandwidth counting in the log-piper and not have to worry about users modifying their own logfiles.
19:33 < mwolson> i will put the syncing script in /afs/hcoop.net/common/etc/scripts/apache-sync-logs
19:33 < megacz> mwolson, cool
19:34 < megacz> btw, I need to convert that dir to CVS
19:34 < Smerdyakov> mwolson, and don't forget to check into 'misc' CVS repository.
19:34 < ntk> are we settled as to logfiles going forward now?
19:34 < megacz> ntk, IMHO yes
19:34 < docelic> mwolson, cool. Note that the sync script will need write permissions in AFS
19:34 < mwolson> Smerdyakov: ok, will do, now that i've checked out all of the hcoop repos
19:34 < docelic> so the way to do that, for the moment
19:34 < ntk> can someone update me as to bind and exim then?
19:34 < docelic> is k5start -U -f /etc/keytabs/root.admin.keytab; aklog
19:34 < ntk> what's the status and the transition plan?
19:34 < megacz> I believe exim is working, no?
19:35 < Smerdyakov> megacz, it has been for sending mail, at least, but things change so fast that I never know anymore.
19:35 < mwolson> megacz: nope, exim is not working anymore
19:35 < megacz> I am 99% sure that exim is working properly for inbound mail, and courier is working for IMAP
19:35  * megacz takes that back
19:35 < mwolson> i just tested it last hour
19:36 < megacz> oh wow, the daemon isn't even running
19:36 < mwolson> after migrating it to use U/US/USER type paths for delivery (also modified create-user script to make mail folders in U/US/USER style paths)
19:36 < ntk> which daemon?
19:36 < megacz> whatever listens on deleuze:25
19:36 < ntk> exim is running but it says "relaying not permitted" when i do rcpt to:ntk_admin@hcoop.net
19:36 < mwolson> deleuze:~# ps -e | grep exim
19:36 < mwolson> 29566 ?        00:00:00 exim4
19:36 < mwolson> 30620 ?        00:00:00 exim4
19:36 < ntk> nope, it's running
19:36 < megacz> hrm
19:36 < docelic> hcoop.net is not in list of allowed domains, I suppose only deleuze.hcoop.net is
19:36 < ntk> unless i'm talking to a ghost daemon from where i am
19:36 < megacz> ah dammit my ISP is filtering port 25
19:36 < ntk> :-/
19:37 < docelic> dpkg-reconfigure exim4-config
19:37 < ntk> but it is NOT set up to receive mail for hcoop.net
19:37 < mwolson> docelic: no need for that
19:37 < Smerdyakov> That's right.
19:37 < ntk> what is right?
19:37 < Smerdyakov> I didn't include mail stuff in hcoop.net Domtool config there yet.
19:37 < ntk> okay, so that aspect of exim is managed by domtool2
19:37 < Smerdyakov> So thanks for testing that it works. ;)
19:37 < mwolson> docelic: we write the list of acceptable domains into a separate place in /etc/exim4/conf.d/
19:37 < Smerdyakov> ntk, yes, same as always.
19:38 < ntk> is it receiving mail for any domain except deleuze.hcoop.net ?
19:38 < docelic> mwolson, ok , thanks
19:38 < Smerdyakov> ntk, Domtool (old and new) write domainlists to be used in Exim config.
19:38 < ntk> if not, we can't be sure it's actually working
19:38 < ntk> okay
19:38 < mwolson> here's the error i'm saying in exim's mainlog:
19:38 < mwolson> 2007-04-15 14:42:12 1Hd9fw-0007h7-Q3 <= mwolson@deleuze.hcoop.net U=mwolson P=l\
19:38 < mwolson> ocal S=402
19:38 < mwolson> 2007-04-15 14:42:12 1Hd9fw-0007h7-Q3 == mwolson@deleuze.hcoop.net R=local_user \
19:38 < mwolson> T=maildir_home defer (13): Permission denied: cannot chdir to /afs/hcoop.net/co\
19:39 < mwolson> mmon/email/m/mw/mwolson
19:39 < mwolson> gah, line-continuation characters
19:39 < Smerdyakov> It's probably best to identify to-do items before transitioning to debugging.
19:39 < Smerdyakov> Have we already identified them all?
19:39 < mwolson> fixed version:
19:39 < mwolson> 2007-04-15 14:42:12 1Hd9fw-0007h7-Q3 <= mwolson@deleuze.hcoop.net U=mwolson P=local S=402
19:39 < mwolson> 2007-04-15 14:42:12 1Hd9fw-0007h7-Q3 == mwolson@deleuze.hcoop.net R=local_user T=maildir_home defer (13): Permission denied: cannot chdir to /afs/hcoop.net/co\
19:39 < mwolson> mmon/email/m/mw/mwolson
19:40 < megacz> mwolson, ok, I need to modify create-user to use the /a/ab/abcdef convention for /common/email
19:40 < megacz> it's still using the old convention
19:40 < ntk> i do have a concern about migration.  when we update the MX records, we have to have ALL the users who use fyodor for mail to be created on deleuze and their maildirs copied over
19:40 < mwolson> megacz: oh ... didn't realize there was more than one place in that file that i had to change
19:40 < Smerdyakov> ntk, we can update hcoop.net last.
19:41 < Smerdyakov> ntk, ask people to set forwards to deleuze.hcoop.net on fyodor.
19:41 < ntk> you mean not migrate mail until all users have migrated?
19:41 < Smerdyakov> Yes.
19:41 < ntk> okay.  i'm guessing that will take a while
19:41 < Smerdyakov> Rather, people using their own domains to receive mail can do that on their own schedules.
19:41 < ntk> true
19:42 < Smerdyakov> We also still don't have any working rails at Peer 1....
19:42 < Smerdyakov> ntk, are you thinking of that as your job to coordinate, or mine?
19:43 < ntk> I think we should go with Graham's rails.  I'm not going to get these rails fixed, and nobody else is going to do that either.  graham's are cheap, and once he sends them i can just hop downtown during a lunch break and remount deleuze
19:43 < mwolson> as for bind status on mire and deleuze, i'm pretty sure that justin has it all working
19:43 < ntk> if we're going to do that, we should do that soon too
19:43 < ntk> because it requires some (brief) downtime)
19:43 < Smerdyakov> ntk, I asked a meta-question. :P
19:44 < mwolson> ntk: i second that particular solution
19:44 < ntk> i think it's mine, but i think you are awfully insistent on fixing the old ones
19:44 < megacz> ntk, can you let me know before you head to Peer1?  I'd like to have deleuze reboot with a recompiled AFS client.
19:44 < ntk> i think i should sell them for scrap
19:44 < docelic> fixing the old ones should be easy from what I've seen.. a hammer should do
19:44 < Smerdyakov> ntk, BTW, something else I've been wondering about: what legal restrictions do we have on having different members pay different dues/shares of one-time costs?
19:44 < docelic> but if you're not that type.. 
19:44 < ntk> i will obviously email -sysadmin a day ahead of time
19:44 < mwolson> yes, for pity's sake, let's now be so cheap as to insist on fixing them
19:44 < mwolson> s/now/*not*/
19:45 < mwolson> just toss them or sell them for scrap
19:45 < docelic> mwolson, it;s not the 'cheapness' issue.. if ntk was up for fixing them, it would be a quick action and we wouldn't remember them any better than last year's snow.. 
19:45 < Smerdyakov> BTW, it may please the audience to know that I will probably move near the hosting facility in the fall. ;)
19:45 < u_l-lap> mwolson: if they're steel what is there to lose by fixing them?
19:45 < u_l-lap> bent aluminium OTOH...
19:45 < docelic> u_l-lap,  ;-)
19:45 < ntk> Smerdyakov: we can set prices however we want, that includes allocating one-time costs to current/past/future members however we like.  the only thing is that using the voluntary "shares" system, any amount over 1 share per member must be recorded as gift income, rather than dues income
19:46 < ntk> not aluminum
19:46 < megacz> Smerdyakov, new job?
19:46 < mwolson> docelic: it's a pain in the ass to fix things like that.  let's not make ntk's life more painful than it needs to be
19:46 < ntk> they are solid STEEL
19:46 < Smerdyakov> megacz, yup.  adamc goes to Wall Street.
19:46 < u_l-lap> which means they can be bent back without issue
19:46 < ntk> you bend them, then.
19:46 < u_l-lap> I am heman
19:46 < ntk> you're always around, my house is open
19:46 < Smerdyakov> u_l-lap, don't you live near there now?
19:46 < megacz> Smerdyakov, you a quant now?
19:46 < u_l-lap> Smerdyakov: nope I'm in NC now as I escaped the clutches of my evil father
19:46 < Smerdyakov> megacz, most likely OCaml hacker, but all of my options aren't decided yet by the Powers That Be.
19:47 < mwolson> are there any more items left on the meeting agenda?
19:47 < Smerdyakov> ntk, I will go along with whatever The People think is reasonable for rail obtaining.
19:48 < ntk> alright, unless anyone says it's out of line, I'm going to let graham know we're taking up his offer of $50 hcoop credit or whatever it is and having him mail his rails to peer1
19:48 < Smerdyakov> ntk, that leaves one more set of rails.
19:48 < mwolson> ntk: please do
19:48 < docelic> ntk, good.
19:48 < ntk> i'll do that after the meeting, while you email interserver
19:48 < Kuril> ntk, I hope it isn't too late.
19:48 < ntk> no, i'm sure it isn't
19:48 < u_l-lap> we ought to save the bent rails in case we find a way to fix them later on
19:48 < Smerdyakov> Wait until we have our own offices, and then hang them on the wall!
19:48 < ntk> they're in the basement of the condo i moved into.  it's less urgent that i dispose of them now that they're not taking up roomspace
19:49 < u_l-lap> if I could get them down to my basement my ex-roommate left all of his metalworking tools so I'm sure there's some type of press in there somewhere
19:49 < mwolson> one additional item i want to bring up: the database creation script fails for me when creating the "mwolson" user
19:50 < Smerdyakov> mwolson, I think I got held up on the log file stuff and didn't update my bit as I should have.
19:50 < Smerdyakov> mwolson, is it an error about a database user not existing?
19:50 < mwolson> + sudo -u postgres psql -c 'CREATE TABLESPACE user_mwolson OWNER mwolson LOCATION '\''/afs/hcoop.net/common/databases/mwolson/postgres'\''' template1
19:50 < mwolson> ERROR:  role "mwolson" does not exist
19:50 -!- u_l-lap [n=clinton@cpe-071-077-050-183.nc.res.rr.com] has quit [Read error: 104 (Connection reset by peer)]
19:50  * Smerdyakov nods.
19:50 < docelic> Ah, right !
19:50 < docelic> So even tablespace should go to user-creation path
19:50 < ntk> Smerdyakov: the rails for the Sun?
19:51 < ntk> i forgot about that
19:51 < Smerdyakov> ntk, yes.
19:51 < docelic> It's my mistake, I didn't think of it
19:51 < Smerdyakov> docelic, why not do it as part of dbtool?
19:51 < docelic> Smerdyakov, you wanted me to move it out
19:51 < docelic> tablespace creation involves AFS stuff
19:51 < ntk> we can order those from sun, they are expensive, or do without them, since the sun is very lightweight and the dell doesn't mind it on its back ata all
19:51 < Smerdyakov> docelic, oh, if it's the CREATE TABLESPACE that creates the directory, yeah.
19:51 < ntk> but i have found no other sources of the rails
19:52 < docelic> Smerdyakov, yes, the moment you run create tablespace, it creates a bunch of files in the tablespace dir
19:52 < ntk> none significantly cheaper anyway
19:53 < Smerdyakov> I'm sure we can work something out among the subset of members with ready disposable income, as long as the law allows.
19:53 < ntk> donations from members are fine, if anyone wants to sponsor the sun rails
19:53 < ntk> let me find the model number again
19:54  * offby1 jerks awake
19:54 < Smerdyakov> Is that the end of the agenda?
19:54 < offby1> "ready disposable income + weak brain" -- that's me
19:55 < Smerdyakov> I'm more than willing to let admins handle various details amongst themselves, but don't be shy about bringing something up now if it would do some good.
19:55 < mwolson> megacz: you're planning on fixing create-user to get rid of the exim problem, then?
19:55 < docelic> What is the solution on tablespace ?
19:55 < megacz> mwolson, yes
19:55 < Smerdyakov> docelic, I don't know, but I think you guys come up with it and then tell me what to do in Domtool code. ;P
19:56 < megacz> mwolson, we should coordinate further, though.  it appears that exim is no longer obtaining tokens properly
19:56 < megacz> I will investigate this evening
19:56 < mwolson> ok
19:56 < docelic> Smerdyakov, supposedly, tablespace creation should stay in plugin 
19:56 < docelic> there's really no other way, if you don't want that I create the user in advance
19:56 < ntk> it is X9267A or X9274A
19:56 < mwolson> that concludes the meeting items that i wanted to bring up
19:57 < Smerdyakov> docelic, it should be OK, as long as the permissions and quota for the volume are set in adduser and inherited, yes?
19:57 < Smerdyakov> That is, where "OK" includes leaving system policy details out of Domtool as much as possible.
19:57 -!- bkhl [n=bkhl@nl104-206-203.student.uu.se] has joined #hcoop
19:57 < ntk> hello latecoming swede
19:57 < bkhl> Sorry.
19:58 < bkhl> Had company here.
19:58  * Smerdyakov really like show AFS directories inherit ACLs from their parents by default.
19:59 < docelic> Smerdyakov, yes, that's SO big problem solver re. permissions
19:59 < docelic> Smerdyakov,  I have good news, you can change owner of tablespace at later time
19:59 < docelic> so I;ll modify create-user to assign tablespace owner to postgres.. then when user is created, you issue SQL alter tablespace owner to $USER
20:00 < docelic> I'll take care of it. 
20:00 < Smerdyakov> Well, OK!  I'll await an e-mail from you detailing what you want me to change from current Domtool/dbtool implementation.
20:00  * mwolson fixes himself a late lunch, will check back periodically
20:00 < docelic> Ok. 
20:01 < Smerdyakov> So, I ask again: Are we through the agenda?
20:01 < docelic> Was a pretty short agenda. 
20:01 < ntk> so there's nothing blocking migration at all now except this log business?
20:01 < megacz> exim needs fixed
20:01 < ntk> um
20:01 < ntk> a few screens up the consensus was that it was working
20:01 < docelic> When Smerdyakov gets logs going and apache working, I can continue with migrating user 'jure'. Just the basic account creation and creation of two domains for him.
20:01 < jdrake> Is there any sort of time frame where things should be near finished?
20:01 < mwolson> ooh, wait, one more thing
20:01 < ntk> and that you couldn't get in due to your firewall
20:02 < docelic> jdrake, no, 'as fast as possible' is the policy
20:02 < jdrake> heh
20:02 < docelic> jdrake, otherwise, speaking of 'near finished', they are near finished for weeks.
20:02 < mwolson> is it ok if i modify root's .emacs on deleuze and mire so that it places backup files in /root/.emacs.d/backup instead of the current directory?
20:02 < ntk> or was there some 'old convention' in the log that needed to be modified
20:02 < jdrake> 10% taking 90% of the time I take it
20:02 < docelic> mwolson, please. 
20:02 < docelic> yes
20:02 < mwolson> i find having "file~" files scattered around the place rather hideous
20:03 < docelic> mwolson, sure. I thought you were generating them 
20:03 < ntk> can someone demonstrate the ease and wonderfulness of the create-user process on deleuze by creating my ntk account right now?
20:03 < Smerdyakov> ntk, that would be breaking procedure!
20:03 < ntk> i had an ntk account before, but it mysteriously disappeared
20:04 < mwolson> ok; since i don't hear any objections, i will make this change to .emacs
20:04 < ntk> do it
20:04 < megacz> ntk, put your password somewhere that I can find it and I will create-user you
20:04 < ntk> root backup files in /etc are annoying
20:04 < mwolson> i will email the necessary snippet to the sysadmin list as well so that individuals can add this to their own .emacs files
20:04 < Smerdyakov> I think the meeting is over.  Y'all admins have some things to do with setting up software.  ntk is going to make these rails happen.  I am going to contact InterServer about a new disk and wait for instructions from admins at any time about changes I should make in Domtool toolset.
20:04 < ntk> yes
20:04 < ntk> this wasn't really a board meeting anyway
20:05 < ntk> but wait
20:05 < ntk> can we at least set a deadline where we intend to have those issues addressed
20:05 < ntk> how about, next weekend at the latest?
20:05 < Smerdyakov> I don't expect that would help.
20:05 < Smerdyakov> We're doing the best we can.
20:05 < docelic> ntk, *as fast as possible*. 
20:05 < ntk> okay, fine
20:05 < ntk> my tail is between my legs
20:06 < ntk> I'm done with this meeting
20:06 < Smerdyakov> I will be around most of the time until 4 hours from now, to help in any debugging sessions folks might want to start to take advantage of combined presence.
20:06 < ntk> i will be be downstairs doing laundry
20:07 < Smerdyakov> </meeting>