Wednesday, 2010-11-24

*** dubsquared has quit IRC00:20
*** allsystemsarego has quit IRC00:21
*** dizz has joined #openstack00:23
*** GasbaKid has quit IRC00:27
*** nelson__ has quit IRC00:33
*** nelson__ has joined #openstack00:33
*** 50UAABCVA has quit IRC00:34
*** rds__ has quit IRC00:38
*** dizz is now known as dizz|away00:41
winston-dHi, All01:00
winston-dI met some problem with bzr01:00
winston-dI cannot branch NOVA or SWIFT from launchpad01:02
*** ar1 has joined #openstack01:08
*** sirp2 has quit IRC01:09
*** jpipes has quit IRC01:15
uvirtbotNew bug: #680730 in swift "Adding custom mime.types file for proxy server" [Undecided,New] https://launchpad.net/bugs/68073001:26
*** 94SAADXVK has joined #openstack01:27
*** rlucio has quit IRC01:34
*** dragondm has quit IRC01:37
*** sirp1 has joined #openstack01:45
*** greenisus has quit IRC01:47
*** miclorb_ has joined #openstack01:58
*** maplebed has quit IRC02:09
*** utauu has joined #openstack02:17
*** utauu has left #openstack02:17
*** rlucio has joined #openstack02:19
*** adjohn has joined #openstack02:19
*** jinesh has joined #openstack02:25
*** 94SAADXVK has quit IRC02:25
*** rlucio has quit IRC02:28
*** stewart has quit IRC02:33
*** adjohn has quit IRC02:38
*** jinesh has quit IRC02:57
*** nelson__ has quit IRC02:57
dendrobates winston-d: bzr branch lp:nova should work03:10
winston-dDendrobates, Thanks, but it doesn't.03:12
winston-dthere's error output: bzr: ERROR: exceptions.NotImplementedError: should resend request to http://feeds.edge.launchpad.net/bazaar/, but this isn't implemented03:13
winston-dI'm using 'bzr' behind proxy03:17
HugoKuowhat's the meanning od workers=2     in /etc/swift/account-server.conf03:24
notmynameHugoKuo: workers=2 means that 2 child processes will be spawned03:40
*** ksteward has joined #openstack03:43
*** filler has quit IRC03:46
*** filler has joined #openstack03:48
*** ArdRigh has quit IRC03:50
odyiHmm...had to kill dnsmasq and run it in the foreground to get it to respong to the instances dhcp requests.04:06
*** ptremblett has quit IRC04:07
*** krish has joined #openstack04:14
*** sirp1 has quit IRC04:16
*** ptremblett has joined #openstack04:20
*** HugoKuo has quit IRC04:20
*** HugoKuo has joined #openstack04:41
*** rlucio has joined #openstack04:42
*** hazmat has quit IRC04:45
*** omidhdl has joined #openstack04:47
*** miclorb_ has quit IRC04:50
*** daleolds has quit IRC04:55
*** suchitp has joined #openstack05:05
*** suchitp has left #openstack05:05
*** hornbeck has quit IRC05:08
*** hazmat has joined #openstack05:26
*** kashyapc has joined #openstack05:28
*** hazmat has quit IRC05:31
HugoKuohow to manage user account in SWIFT05:34
HugoKuoe.g.   show a users list05:35
creihtHugoKuo: There isn't that capability really built in for that yet (since it is just a dev auth)05:38
creihtsqlite3 /etc/swift/auth.db 'select account, user, password from account'05:38
creihtwill give it to you though :)05:38
HugoKuothanks ...05:38
creihtnp05:39
*** f4m8__ is now known as f4m805:42
*** hornbeck has joined #openstack06:03
*** hornbeck has joined #openstack06:04
*** hornbeck has quit IRC06:06
*** ramkrsna has joined #openstack06:09
*** ramkrsna has joined #openstack06:09
HugoKuoSWIFT question :  while I use st -A http://172.16.3.133:11000/v1.0 -U hugo:hugo -K 12345 list hugo  on SWIFT machine, it work normally to list objects06:32
HugoKuoBut I run the same cmd on another computer socket.error [Errno 111] connection refused06:33
creihtHugoKuo: First guess is maybe do you have any firewalling going on?06:33
HugoKuoLet me check06:35
HugoKuoTraceback (most recent call last):06:52
HugoKuo  File "./st", line 1551, in <module>06:52
HugoKuo    error_queue)06:52
HugoKuoFile "./st", line 1168, in st_list06:52
HugoKuo    prefix=options.prefix, delimiter=options.delimiter)[1]06:52
HugoKuo  File "./st", line 784, in get_container06:52
HugoKuo    full_listing=full_listing)06:52
HugoKuo  File "./st", line 735, in _retry06:52
HugoKuo    rv = func(self.url, self.token, *args, **kwargs)06:52
HugoKuo  File "./st", line 361, in get_container06:52
HugoKuo    conn.request('GET', '%s?%s' % (path, qs), '', {'X-Auth-Token': token})06:52
HugoKuo  File "/usr/lib/python2.6/httplib.py", line 910, in request06:52
HugoKuo    self._send_request(method, url, body, headers)06:52
HugoKuo  File "/usr/lib/python2.6/httplib.py", line 947, in _send_request06:52
HugoKuo    self.endheaders()06:53
HugoKuo  File "/usr/lib/python2.6/httplib.py", line 904, in endheaders06:53
HugoKuo    self._send_output()06:53
HugoKuo  File "/usr/lib/python2.6/httplib.py", line 776, in _send_output06:53
HugoKuo    self.send(msg)06:53
HugoKuo  File "/usr/lib/python2.6/httplib.py", line 735, in send06:53
HugoKuo    self.connect()06:53
HugoKuo  File "/usr/lib/python2.6/httplib.py", line 716, in connect06:53
HugoKuo    self.timeout)06:53
HugoKuo  File "/usr/lib/python2.6/socket.py", line 514, in create_connection06:53
HugoKuo    raise error, msg06:53
HugoKuosocket.error: [Errno 111] Connection refused06:53
edayHugoKuo: please use http://paste.openstack.org/ for big pastes :)06:53
HugoKuosorry about that06:53
edayHugoKuo: no worries... but that error message means it can't reach the server06:54
edayso, wrong port/firewalled/something along those lines06:54
HugoKuoThanks for your suggestion06:55
*** omidhdl1 has joined #openstack06:59
*** omidhdl has quit IRC07:00
*** guigui1 has joined #openstack07:08
*** rlucio has quit IRC07:08
*** GasbaKid has joined #openstack07:23
*** GasbaKid has quit IRC07:33
*** befreax has joined #openstack07:47
*** GasbaKid has joined #openstack07:54
*** prgmrcom has quit IRC07:54
sorenwinston-d: Did you work out your bzr problem?07:56
*** almaisan-away is now known as al-maisan08:05
*** aimon has quit IRC08:07
*** aimon has joined #openstack08:08
*** jkakar has joined #openstack08:09
*** MarkAtwood has quit IRC08:09
*** alekibango has quit IRC08:17
*** guigui1 has quit IRC08:18
*** alekibango has joined #openstack08:18
*** calavera has joined #openstack08:19
*** ibarrera has joined #openstack08:28
*** GasbaKid_ has joined #openstack08:29
*** GasbaKid has quit IRC08:31
*** al-maisan is now known as almaisan-away08:35
*** almaisan-away is now known as al-maisan08:39
*** GasbaKid has joined #openstack08:44
*** ianweller has quit IRC09:06
*** jkakar has quit IRC09:08
*** ianweller has joined #openstack09:09
*** miclorb has joined #openstack09:10
*** allsystemsarego has joined #openstack09:19
sorenWe need more core people in this timezone.09:20
*** jkakar has joined #openstack09:22
*** seife has joined #openstack09:23
sorenjoshuamckenty: Oh, got a minute to review a branch?  https://code.launchpad.net/~soren/nova/doc-build-dir/+merge/41693 It's a 0-line patch. Should be real quick.09:23
seifehi09:23
sorenHello.09:23
seifeis there a way to join the mailinglist without a launchpad account?09:24
sorenNo.09:24
seifeI'd really like to avoid launchpad if possible..09:24
seifeooops. Ok, that's bad.09:24
sorenWhy?09:24
seifeI hate it ;)09:24
soren*shrug*09:25
seifeI'd really like to not give my personal details to canonical ;)09:25
sorenThen don't?09:25
alekibangoseife: i understand. i would not join google to get maillists there...09:25
sorenseife: Or lie or whatever.09:25
seifeyeah, I'll try that.09:26
alekibangosoren: btw do you play golf?  :)09:26
sorenalekibango: :(09:26
sorenalekibango: No.09:26
alekibangoyou would not be here, right09:26
sorenThis is more fun than golf.09:27
alekibango:)09:27
alekibangosoren: i have seen some articles about Soren the golf player, just curious if you are family with him...09:28
sorenalekibango: I've gotten his fan mail occasionally.09:28
sorenalekibango: srsly.09:28
alekibangothat must be evil...09:28
soren"Soren Hansen" is one of the most common names in Denmark. It's *really* annoying.09:29
sorenalekibango: Tag line of my blog: "The blog of Soren Hansen (not the golfer)"09:29
alekibangobetter than  '(not the gay porn star)' still09:30
matttsoren: could always change your first name to chuck09:31
ttxalekibango: oh, you know about *that* soren hansen as well ?09:31
alekibangottx: :) no, tht was another experience09:31
sorenmattt: err?09:31
seifehm. I cannot login to launchpad with my openID...09:32
sorenYou can read the archives on-line if you're really that concerned about keeping your name and e-mail secret.09:33
alekibangoseife: try ssh key09:33
matttsoren: how about 'chip' then?09:33
seifeargh: "09:33
seifeCan I log into Launchpad with my existing OpenID?09:33
seifeNot at this time. Launchpad is currently an OpenID Provider (OP) and not a consumer (RP).09:33
alekibangoat least for bzr :)09:33
seife"09:33
alekibangoseife: you know, every software has few bugs and design flaws09:33
sorenmattt: Are these names supposed to mean anything to me? :)09:34
seifeIts not about keeping my email secret - it's about givin personal data to a random US company. :) But yeah, I probably have to bite the bullet.09:34
sorenseife: Canonical isn't a US company.09:34
sorenseife: Incidentally, Rackspace is.09:35
ttxseife: the amount of "personal data" you have to disclose to create a LP account is rather minimal. If not null.09:35
alekibangogoogle is other story09:35
sorenseife: And, as ttx points out, the only personal data you have to give is your e-mail address, so it's very much about your e-mail.09:35
matttsoren: guess it's too early to be talking crap :)09:36
soren:)09:36
*** seife has quit IRC09:39
*** seife has joined #openstack09:40
*** seife has quit IRC09:44
*** arthurc has joined #openstack10:07
*** HugoKuo has quit IRC10:07
*** perestrelka has quit IRC10:08
*** dkocher has joined #openstack10:10
*** HugoKuo has joined #openstack10:12
*** perestrelka has joined #openstack10:14
*** miclorb has quit IRC10:16
*** dizz|away is now known as dizz10:30
*** dizz is now known as dizz|away10:31
*** 92AABND4C has joined #openstack10:43
*** rds__ has joined #openstack10:51
*** jc_smith has quit IRC11:05
*** irahgel1 has joined #openstack11:26
sandywalshsoren, did I answer your question re: motivation on merge?11:31
sorensandywalsh: /me reads11:32
sorensandywalsh: It does. I just got confused by the lack of bug report and/or blueprint and that your example was about limiting access to flavours :)11:33
sorensandywalsh: Perhaps you could add something about the motivation to the description? Just for future reference?11:34
sandywalshah, yes, sorry, I'll try to be more generic11:34
sandywalshof the bp?11:34
sorenThere's a bp?11:34
sandywalshyup11:34
sorenOh, so there is.11:34
sorenI somehow missed that. Sorry.11:35
sandywalshnp ... easy to do that on lp I find11:35
sorenTotally.11:42
sorenI'm not sure why I failed so miserably.11:43
sorensandywalsh: Followed up.11:44
*** befreax has quit IRC11:47
sandywalshsoren, thanks for feedback ... good points.11:50
sorensandywalsh: np11:50
sandywalshone counterpoint, before I write notes ...11:50
sandywalshif we are replacing/fixing gflags to allow macro replacement could we not simply define the set of admin ops and user ops there?11:51
sandywalshand then configuration is simply "%admin_commands" or "%user_commands" in .conf11:51
sandywalsh(or whatever the syntax is)(11:51
sandywalshspecifically, keep the ops set out of the code completely11:52
sorenIn the current implementation of the macro replacement thing, that would look odd.11:55
sorenYou'd have to have flags called admin_commands and user_commands.11:56
soren...which would show up in --help.11:56
sandywalshah. yuck11:56
soren..and the effects of changing them would be surprising.11:57
sandywalshthis was my argument against the gflags refactoring ... we end up building a DSL11:57
sandywalshor severely limit ourselves11:57
sorenFor something as complex as Nova, flags are surprisingly difficult.11:57
sandywalshwhich is why I'm an advocate of .py configuration11:58
sandywalsha la django, etc.11:58
sorenI'm undecided.11:58
sandywalsh:)11:58
sorenWhatever we do, it's going to suck somehow.11:59
sandywalshso, you're proposing binary admin functionality ... this or that. No grey areas. Let me run it up the flag pole.11:59
sorenYup, that's my proposal.12:00
sandywalshan easy change and I understand the motivation ... thx for the feedback!12:00
sorenIf you can show me an example of something that may or may not be considered admin-only, I'm likely to accept the current proposal.12:00
sorenRight now, it just seems like a very unambiguous split.12:01
sandywalshright ... no idea at this point. Which makes the current implement sort of anti-YAGNI ... to your point.12:01
sorenYet Another... what?12:02
sandywalshYa Ain't Gonna Need It (don't build for the future too much)12:02
sorenAh.12:02
sorenYes, good point :)12:02
sandywalshk, thanks again. Let's see what happens ... (I may just make the change assuming it will be the case)12:03
patri0tsoren: when I want to install Nova in MultipleServer, what exactly do I need to change for networking12:06
patri0tusing PPA (so I shoudl change the db as well)12:07
patri0tand by the change I mean setting interfaces for Cloud Controller node12:08
*** ramkrsna has quit IRC12:10
sorenpatri0t: Do you need to change anything?12:12
patri0tgood question :D, so why I got this? http://paste.openstack.org/show/184/12:13
sorenpatri0t: Good question. Is that from nova-compute?12:13
patri0tyeao12:14
sorenpatri0t: Can I see your flagfile?12:14
patri0twhich one? compute?12:14
sorenYes.12:14
patri0thttp://paste.openstack.org/show/185/12:14
sorenOh, hang on.12:15
sorenWhich version of the packages are you using?12:15
patri0thmm, I used PPA12:15
sorenYes, but which version are you running.12:16
soren?12:16
patri0t2011.1~bzr396, lucid12:17
patri0tdo mean this?12:17
*** kashyapc has quit IRC12:17
patri0t2011.1~bzr396-0ubuntu0ppa1~lucid112:18
*** krish has quit IRC12:22
patri0tsoren: any idea?12:23
*** al-maisan is now known as almaisan-away12:24
sorenOh, that's old.12:24
sorenSeveral days old.12:24
soren:)12:24
sorenThere's a new PPA for trunk packages.12:25
sorensandywalsh: Oh, would you mind reviewing https://code.launchpad.net/~soren/nova/doc-build-dir/+merge/41693 ?12:25
sorensandywalsh: Shouldn't take more than a few seconds :)12:26
*** omidhdl1 has left #openstack12:26
patri0tis this one also related to same problem ? http://paste.openstack.org/show/186/12:26
sorenpatri0t: Not at all.12:27
sorenpatri0t: You should be using this PPA instead: https://launchpad.net/~nova-core/+archive/trunk12:27
patri0tthanks, I update it12:28
sorenpatri0t: Mind you, it consolidates your flagfiles into one.12:29
sorenpatri0t: Just so you know.12:29
sorenpatri0t: In the sense that it will make all workers look at the same flagfile. It won't actually touch your existing flagfiles.12:30
sorenpatri0t: So you should move everything into /etc/nova/nova.conf12:30
patri0tsoren: perfect, at least now I have one config file to deal with12:31
patri0tsoren: any other side effect?, I should update the workers as well12:31
sorenpatri0t: It's just a general update.12:32
*** Podilarius has left #openstack12:34
*** krish has joined #openstack12:34
*** arthurc has quit IRC12:40
*** ctennis has quit IRC12:47
*** ar1 has quit IRC12:47
sorensandywalsh: https://answers.launchpad.net/nova/+question/132732  <--- Sounds like something you'd be able to handle well.12:49
alekibangowhat do you think about this one? https://answers.launchpad.net/nova/+question/13438912:52
patri0tsoren: Im confused, should I put all the flags in nova.conf now? like fake_subdomain, cc_host, rabit_host ...?12:53
sorenpatri0t: Yes.12:54
sorenpatri0t: (FAKE_subdomain is gone, by the way12:54
soren)12:54
sorenalekibango: I think someone with sufficient motivation should respond. That's about all.12:55
sorenpatri0t: Back when I initially did this packaging, workers would reject unknown flags, so they couldn't all use the same flagfile (since they didn't accept the same flags). That's changed now, so they can all use the same flagfile, which makes everything much simpler.12:56
zykes-alekibango: good thing regarding SAN, OpenStack isn't an option for where i work without12:59
alekibangoi mean, what is bad on sheepdog ;)13:00
alekibangoi dont think that generic filesystem can be more effective13:00
patri0tsoren: change in DB as well??!13:01
patri0tI should drop it?13:01
sorenpatri0t: Huh?13:06
sandywalshsoren, I'll take that question13:06
sorensandywalsh: ta13:07
*** GasbaKid has quit IRC13:10
*** ctennis has joined #openstack13:16
*** ctennis has joined #openstack13:16
*** joshuamckenty has left #openstack13:17
*** stewart has joined #openstack13:19
*** hggdh has quit IRC13:20
zykes-alekibango: is sheepdogg meant to be usd for san ?13:23
sorensandywalsh: While you're around, mind reviewing https://code.launchpad.net/~soren/nova/doc-build-dir/+merge/41693 ?13:24
sandywalshsoren, got it13:26
*** krish has quit IRC13:33
*** krish has joined #openstack13:34
sandywalshsoren, do I just do an Approve comment? What is Review Type?13:34
patri0tsoren: is there any change in db schema?13:35
patri0thttp://paste.openstack.org/show/187/13:35
*** jdarcy has joined #openstack13:37
*** ttx has quit IRC13:48
sorenpatri0t: I don't think so.13:49
sorenpatri0t: But I'm not sure.13:49
sorensandywalsh: Just ignore review type.13:49
*** ttx has joined #openstack13:49
*** ttx has joined #openstack13:49
sorensandywalsh: It's for cases where a merge proposal may have UI changes, DB changes, logic changes and different people will review different things.13:50
sorensandywalsh: Stuff like that.13:50
sandywalshsoren, gotcha13:50
zykes-soren: which UI ?13:51
*** seife has joined #openstack13:52
*** hazmat has joined #openstack13:52
sorenzykes-: eh?13:53
*** almaisan-away is now known as al-maisan13:53
zykes-just saw the comment above13:54
*** westmaas has joined #openstack13:57
*** ttx has quit IRC13:58
*** ttx has joined #openstack13:58
*** ttx has joined #openstack13:58
openstackhudsonProject nova build #261: SUCCESS in 1 min 17 sec: http://173.203.104.204:8080/job/nova/261/14:00
*** ttx has quit IRC14:01
*** ttx has joined #openstack14:01
*** ttx has joined #openstack14:01
sorenseife: https://launchpad.net/~nova should have a "join this team" sort of button.14:01
sorenseife: Ah, no it doesn't.14:01
sorenseife: We're moving stuff from nova and glance into the openstack team.14:02
sorenseife: Just join that.14:02
*** dkocher has left #openstack14:02
seifeok14:02
sorenseife: Ah, you can't join that either, can you?14:02
sorenProbably not.14:02
seifeI'll try. one moment14:02
sorendendrobates: What's the plan for this?14:02
sorendendrobates: Right now, both ~nova and ~openstack are restricted teams.14:03
seifeno "join" button on any of the projects swift, nova, openstack14:03
alekibangozykes-: ??usd?14:03
exltspeaking of "where things are in launchpad", where is the nova debian/ branch kept?14:03
sorenexlt: It's well hidden.14:05
exltheh - I looked, and you are correct :)14:05
alekibangohttps://code.launchpad.net/~nova-core/nova/debian-packaging14:06
sorenexlt: https://code.launchpad.net/~openstack-ubuntu-packagers/ubuntu/natty/nova/ubuntu14:06
sorenalekibango: Nope, not anymore.14:06
*** kashyapc has joined #openstack14:06
alekibangosoren: ok14:06
alekibangoso ubuntu package should work for debian also?14:07
alekibango:)14:07
alekibango(the source one)14:07
alekibangoi was using sources for now14:07
sorenDon't know.14:07
sorenHaven't tried.14:07
alekibangoi can try that...14:08
exltalekibango: as long as package build is correct - the contents of the debian/ directory specifies package build14:08
sorenalekibango: Nope, won't work.14:08
sorenalekibango: Debian doesn't have eventlet.14:08
sorenalekibango: Blame mtaylor. I do.14:08
alekibangomtaylor: ??? eventlet in debian?14:08
alekibangosoren: i suppose we can repackage it for debian14:09
sorenalekibango: Sure.14:09
alekibangoi mean, squeeze is right before release14:09
exltdebian has an ITP for eventlet, which I've pinged a few times..14:09
sorenalekibango: Exactly.14:09
sorenSo too late.14:09
alekibangocould be nice target to run on14:09
alekibangosoren: for me, its not essential to get it into debian14:09
alekibangothat is step #114:09
alekibangojust having debian compatible packages and tools would be nice14:10
alekibangoi mean step #2 is pushing it into debian14:10
sorenYou can probably install the Ubuntu packages on Debian and they'll run just fine.14:10
alekibangoi did that with freenx recently14:10
*** jaypipes has joined #openstack14:10
alekibangosoren: will try tonight, if i will not fall asleep too soon14:11
alekibangosoren: which packages should i try on squeeze?14:11
alekibangolucid ones?14:11
*** kashyapc has quit IRC14:11
sorenalekibango: No idea.14:11
exltalekibango: that should be fine14:11
sorenalekibango: I'd be building from source if it were me.14:11
sorenBuilding packages from source, that is.14:12
alekibangojsut for the heck of the test...14:12
exltalekibango: and I'm curious if something doesn't work14:12
*** gondoi has joined #openstack14:12
alekibangoi would like fai to install it for me tonight14:12
exltalekibango: actually, the maverick packages would be the one's I would attempt14:12
alekibangofai-project.org14:12
alekibangowhy, what is the difference, exlt?14:13
alekibangoi didnt follow them much14:13
exltlucid released a while back, while many changes have gone into squeeze since lucid release14:13
openstackhudsonProject nova build #262: FAILURE in 9.4 sec: http://173.203.104.204:8080/job/nova/262/14:13
sorenopenstackhudson: I don't like you anymore.14:14
exltstart with squeeze install and try maverick packages for nova and swift -14:14
zykes-alekibango: ?14:14
alekibangozykes-: you are asking me something i cant grok14:14
alekibango0-11-24 14:23] <zykes-> alekibango: is sheepdogg meant to be usd for san ?14:14
alekibangowhat is usd?14:14
zykes-used.14:14
alekibangoah...14:14
alekibangono, its scalable storage for VIRTUAL IMAGES -- compatible with KVM/QEMU14:15
alekibangonot generic storage14:15
alekibangosan - most people imagine storing files...14:15
sorenGawd, this is embarassing.14:16
alekibangozykes-:  see web, pictures etc http://www.osrg.net/sheepdog/14:16
alekibangowell, maybe in some context, it might be - SAN for virtual images for KVM/QEMU14:16
zykes-alekibango: what's the use of having bought a san if you can't use it ?14:17
alekibangobut its not storing files...14:17
alekibangozykes-: the man asking the question in nova questions in launchpad said he has no san atm14:17
zykes-ah14:17
alekibangoand i also would like to avoid paying server tax14:17
alekibangoi mean - array of commodity hardware -- with little better network here and there - should be enough14:18
alekibango(at best)14:18
zykes-alekibango: Compellent ain't that expensive, contra EMC14:18
* soren has to run14:18
zykes-or Eqalogic14:18
alekibangozykes-: i like floss solutions14:18
alekibangocoming with freedom14:19
zykes-cheap solutions ? : )14:19
alekibangofreedom is not cheap14:19
*** ppetraki has joined #openstack14:19
alekibangobut is worth it14:19
zykes-hehe, well we allready have a SAN, and doesn't rackspace also use one for storing vm's ?14:20
alekibangozykes-: well, openstack is still lacking reliable, integrated storage for running guests....14:20
alekibangoi mean highly  available one14:21
zykes-and that's sheepdogg for example ?14:21
alekibangoyes14:21
zykes-or a cluster filesystem i guess14:21
alekibangoand i dont know about much other solutions14:21
zykes-on a shared san14:21
alekibangoright14:22
*** dendrobates is now known as dendro-afk14:22
alekibangobut i dont believe they will be more effective than sheepdog14:22
alekibangoeven if they might be fun to hack...14:22
zykes-what fs can be used on a san then i wonder14:23
*** dendro-afk is now known as dendrobates14:23
*** krish has left #openstack14:31
openstackhudsonYippie, build fixed!14:40
openstackhudsonProject nova build #263: FIXED in 1 min 12 sec: http://173.203.104.204:8080/job/nova/263/14:40
_0x44zykes-: No, cloudservers is not SAN backed.14:41
*** spectorclan has joined #openstack14:43
*** jinesh has joined #openstack14:45
zykes-_0x44: hmm, ok i thought i saw that in a vid of the datacenter but14:45
zykes-_0x44: what's used then ?14:45
*** jinesh has joined #openstack14:45
dendrobateszykes-: local disk14:46
_0x44zykes-: ^^ What dendrobates said14:46
*** f4m8 is now known as f4m8_14:46
*** dubsquared has joined #openstack14:46
_0x44zykes-: If there was a SAN in a video, it's something Managed uses or sells14:46
*** Podilarius has joined #openstack14:46
creihtzykes-: yeah there are other SAN offerings at RS14:47
zykes-ah ok, what would be a good setup to use with openstack / kvm / san then ?14:48
*** Ryan_Lane has quit IRC14:48
alekibangozykes-: local disk? :)14:49
*** laurensell has joined #openstack14:49
alekibangoi know, bad joke...14:49
* creiht doesn't know enough about san14:49
zykes-heh :o14:49
_0x44zykes-: If you are going to use a SAN why wouldn't you just configure your dom0 to treat your SAN VG/LVs as local disks/devices?14:49
_0x44zykes-: Then you'd just use OS the way you would on local disks, and configure your SAN the way you always would14:50
alekibango_0x44: you have to understand that people asking such questions do want to have HA setup, but they dont know much about how to run your own SAN....14:50
zykes-alekibango: .. i never used san except in RHEV (RedHat Enterprise Virt) and VMware ESX/vCenter..14:51
alekibangozykes-: and i just tested such systems on 2 computers, without real life usage in production14:52
zykes-which kind of system ?14:53
alekibangoi mean glusterfs, sheedog, etc14:53
zykes-ah14:53
zykes-is glusterfs distributed as well as ceph / sheepdogg ?14:53
alekibangoits different... very interesting, simple and lego like14:53
alekibangobut its slow14:54
zykes-ok14:54
zykes-i guess i could check with rhev uses14:54
alekibangolustre is much mroe complex14:54
alekibangoand has central metadata servers14:54
alekibangoeasier to get corrupted badly somehow14:54
zykes-alekibango: i guess lustre is more like panFS14:54
alekibango(in theory)14:55
zykes-that i mentioned earlier14:55
alekibangoi like glusterfs because its simple14:55
alekibangoits almost as simple as nfs -- but few times slower14:55
dubsquaredlusters distributed filesystem doesn't break files apart, but rather puts whole files on different target locations14:55
alekibangoglusterfs also stores whole files14:56
alekibangoits interesting when you have higher number of servers - total throughput is good... but each and every request is slow14:56
*** schisamo has joined #openstack15:01
*** rbergeron has quit IRC15:02
*** rbergeron has joined #openstack15:02
spectorclanDesign Summit Event Page with Sides and Videos Available - http://openstack.org/events/openstack-design-summit-2010/15:05
creihtinteresting...15:05
creihthttp://www.theregister.co.uk/2010/11/23/darily_ubuntu_updates/15:05
creiht:)15:05
*** damon__ has joined #openstack15:05
*** al-maisan is now known as almaisan-away15:16
*** almaisan-away is now known as al-maisan15:17
sandywalshsoren, :( I don't have any admin command ... my unit tests are no-ops15:18
*** laurensell has quit IRC15:20
*** dubsquared has quit IRC15:21
*** jinesh has quit IRC15:21
*** ksteward has quit IRC15:21
*** cory_ has quit IRC15:21
*** pquerna has quit IRC15:21
*** timrc has quit IRC15:21
*** jero has quit IRC15:21
*** odyi has quit IRC15:21
*** dubsquared has joined #openstack15:24
*** jinesh has joined #openstack15:24
*** ksteward has joined #openstack15:24
*** cory_ has joined #openstack15:24
*** pquerna has joined #openstack15:24
*** timrc has joined #openstack15:24
*** jero has joined #openstack15:24
*** odyi has joined #openstack15:24
*** Ryan_Lane has joined #openstack15:24
*** winston-d has quit IRC15:25
*** sirp1 has joined #openstack15:26
ttxcreiht: it's about the parallel lightweight application delivery mechanism15:30
ttxcreiht: not really about making ubuntu a rolling distro.15:30
ttxSee https://wiki.ubuntu.com/PostReleaseApps15:32
creihtI just thought it was interesting :)15:32
*** al-maisan is now known as almaisan-away15:35
*** GasbaKid has joined #openstack15:36
*** sirp1 has quit IRC15:38
*** almaisan-away is now known as al-maisan15:41
*** abecc has joined #openstack15:42
*** abecc has joined #openstack15:43
*** sirp1 has joined #openstack15:47
*** blpiatt has joined #openstack15:51
*** sirp1 has quit IRC15:56
*** MarkAtwood has joined #openstack15:57
spectorclanalexibango: are you there?15:58
*** jdarcy has quit IRC16:05
*** rsampaio has joined #openstack16:10
*** joshuamckenty has joined #openstack16:13
*** seife has quit IRC16:13
spectorclanIs there a way to search into the IRC Logs which are listed by day?16:19
Ryan_LaneI'm noticing that the LDAP schema has a requirement on the uid attribute... is there any reason to force the use of the nis schema?16:21
Ryan_Lanethis doesn't totally exclude the use of Active Directory, but adds an odd unneeded requirement16:22
*** jaypipes has quit IRC16:23
Ryan_Laneit would likely be better to require the use of cn, default to using cn, but have the default configuration use uid. that way people could use sAMAccountName instead of uid16:23
Ryan_Laneand the schema would be compatible16:24
Ryan_LaneI guess I should put this in as a bug16:24
*** jaypipes has joined #openstack16:27
*** kevnfx has joined #openstack16:28
dubsquaredis there anything going on with the ppa this morning?  im getting some package not found errors...16:30
*** dizz|away has quit IRC16:32
*** dizz has joined #openstack16:32
exltspectorclan: google 'site:http://eavesdrop.openstack.org/irclogs/ <keyword(s)>'16:36
exltspectorclan: is that what you meant?16:36
spectorclanexlt: thanks! Will try it now16:36
*** dizz has quit IRC16:37
spectorclanexlt:  found the link I wanted but the content was wrong. Thanks four your help on search. Will write that down16:41
uvirtbotNew bug: #681030 in nova "Nova's LDAP schema has an unneeded requirement on the nis or bis schema" [Undecided,New] https://launchpad.net/bugs/68103016:42
*** iammartian has joined #openstack16:43
dubsquaredhttps://launchpad.net/~nova-core/+archive/trunk/ is the new location of the packages :D  — documentation update needed (multiserver install)16:49
annegentledubsquared: heh, thanks for the heads up! I'll change it now16:50
dubsquaredman, talk about service!16:51
dubsquaredthanks, anne :D16:51
*** calavera has quit IRC16:53
annegentledubsquared: I know, crazy huh? Well it's in two places, the wiki and the .rst (http://nova.openstack.org/adminguide/multi.node.install.html) so I'll get it both places16:57
dubsquaredThanks! I was update the script that JordanRinke and I put together from the summit, and I was testing and all the sudden it was not working…freaked me out for a sec :)16:59
dubsquaredupdating, too17:00
*** 92AABND4C has quit IRC17:02
*** al-maisan is now known as almaisan-away17:04
sandywalshseeing too_many_processes test failing sporadically17:04
*** Dweezahr has quit IRC17:05
*** Dweezahr has joined #openstack17:07
sandywalshThere was a new test added to trunk nova.tests.misc_unittest ... fails for me17:07
*** daleolds has joined #openstack17:09
sandywalshsoren, looks like your test_authors_up_to_date test17:11
*** jinesh has quit IRC17:13
*** laurensell has joined #openstack17:18
*** jkakar has quit IRC17:25
mtaylorsoren: the maverick ppa should work on debian...17:26
*** jinesh has joined #openstack17:30
*** schisamo has quit IRC17:32
*** maplebed has joined #openstack17:39
*** Ryan_Lane has quit IRC17:58
*** ibarrera has quit IRC17:59
*** kevnfx has quit IRC18:02
*** laurensell has quit IRC18:03
*** joearnold has joined #openstack18:04
*** dendrobates is now known as dendro-afk18:05
*** BK_man has joined #openstack18:05
*** dendro-afk is now known as dendrobates18:05
BK_manhi all! short question about Nova deps. How strict are deps which specified in tools/pip-requires with exact version number?18:06
BK_manshould I install exact that version of python module or I can install latest (with a few exceptions)18:07
*** rlucio has joined #openstack18:07
*** greenisus has joined #openstack18:08
edayBK_man: those are really minimum, versions after *should* work18:08
edayBK_man: and if they don't, let us know :)18:08
*** laurensell has joined #openstack18:08
BK_maneday: indeed!18:08
*** jc_smith has joined #openstack18:08
BK_manis there are any formal tests to ensure that my build is ok?18:09
BK_manI'm running on VM - should I use UML?18:09
dendrobatesBK_man: I would use UML, but you could use qemu too18:11
*** spectorclan has quit IRC18:11
sandywalshBK_man, there are unit test suites (./nova/run_tests.sh)18:11
sandywalshBK_man, I use qemu, haven't tried uml18:11
dendrobateswhen zul finishes the lxc work, that will be my default instead of uml18:11
zuloooh....pressure :)18:12
dendrobatesuml requires images that support it, but it is faster18:12
BK_manbtw - where can I host a small yum repo for free? does somebody knows a such service?18:13
* BK_man is going to create a RHEL packages for nova18:13
dendrobatesBK_man: you might want to check with the fedora guys, they have been working on packaging it already18:14
BK_maneday: I can remember that I failed to run nova code with latest python-lockfile version. You might want to check this18:15
*** rsampaio_ has joined #openstack18:15
BK_mandendrobates: OK. But I need to create my own repo for RHEL6, not Fedora to ensure that I have a local repo for nova18:16
*** rsampaio has quit IRC18:16
*** laurensell has quit IRC18:16
*** laurensell has joined #openstack18:17
*** seshu has joined #openstack18:22
*** laurensell has quit IRC18:24
*** ppetraki has quit IRC18:25
*** westmaas has quit IRC18:26
*** ppetraki has joined #openstack18:26
*** laurensell has joined #openstack18:30
*** spectorclan has joined #openstack18:31
*** westmaas has joined #openstack18:38
sorenmtaylor: Really?18:45
sorenmtaylor: Personally, my money would have been on the lucid ppa.18:45
mtaylorsoeren: no reason why not18:45
mtaylorsoren: we should probably test it...18:46
* mtaylor keeps wanting debian ppas from launchpad18:46
sorenmtaylor: I totally want a Debian maintainer to pick this up.18:46
* mtaylor keeps offering to find machines for the task...18:46
sorenmtaylor: I'd like to make a case study out of openstack.18:46
mtaylorsoren: well, I was going to do it - but then debian freeze sort of sapped my will for new packages :)18:47
sorenmtaylor: Flipping the flow of packages between Ubuntu and Debian upside down.18:47
mtayloryes. it is a good case study candidate18:47
*** stewart has quit IRC18:47
mtaylorsoren: I should do one on drizzle- similar thing, except where upstream uploads to both debian and ubuntu directly, so there is no flow from one to the other18:48
sorenmtaylor: I get the impression drizzle is more self-contained than Nova is.18:48
mtaylorsoren: not at all - I have several dependency packages that I maintain solely because we need them for drizzle18:49
mtaylorand others that I maintained in ppa before adequate versions hit debian/ubuntu mainline18:49
mtaylorso it's essentially the same or similar overall bucket of issues18:49
*** rbtz has joined #openstack18:51
*** westmaas has quit IRC18:51
sorenmtaylor: Ok.18:53
sorenmtaylor: I forget... Are you a DD?18:55
jaypipes_0x44, soren: mind taking a look at https://code.launchpad.net/~jaypipes/glance/clients/+merge/41498? thx in advance :)18:58
*** iammartian_ has joined #openstack18:59
sorenjaypipes: Sure.19:00
jaypipessoren: cheers19:00
*** irahgel1 has quit IRC19:03
_0x44jaypipes: Looking19:03
*** iammartian_ has quit IRC19:08
openstackhudsonProject nova build #264: SUCCESS in 1 min 13 sec: http://173.203.104.204:8080/job/nova/264/19:10
*** seshu has quit IRC19:13
sorenI've haven't quite grokked Glance's discovery bit.19:20
sorenIs it supposed to go look at various image stores to find images that can be imported by nova?19:21
sorenIs that what it's about?19:21
_0x44soren: Nova gives glance the URL of a teller endpoint with an image id, then glance queries that teller for the real image URL, then queries the endpoint for that URL and proxies the data back to nova.19:24
_0x44soren: So it doesn't look at various datastores so much as queries whatever datastore it's asked to.19:25
vishyBK_man: lockfile and boto are broken with the latest, not sure what else might be.19:26
_0x44soren: Errr... dammit, s/teller/parallax/ s/glance/teller/19:26
BK_manvishy: will test it soon! :)19:26
soren_0x44: So "Nova gives teller the URL of a parallax endpoint with an image id, then teller queries that parallax for the real image URL, then queries the endpoint for that URL and proxies the data back to nova." ?19:27
_0x44Yessir19:28
_0x44teller being the data proxy/caching layer, parallax being the image repository end-point.19:28
_0x44(and image registry)19:28
mtaylorsoren: no. I've been in process to be one for like, forever. I'm gonna try to get per-package uploader status there too..19:28
*** burris has quit IRC19:31
*** joearnold has quit IRC19:34
sorenmtaylor: ok19:34
*** joearnold has joined #openstack19:35
soren_0x44: I may have gotten confused by the description on Launchpad. It strongly suggests it does discovery.19:38
_0x44soren: Yeah, that's wrong and needs to be changed.19:39
jaypipes_0x44: thx mate :)19:40
*** rlucio has quit IRC19:41
*** jaypipes has quit IRC19:41
sorensoeren: Um... hi?19:44
soren:)19:44
*** kakoni has joined #openstack19:44
kakonig'day.19:44
kakoniquestion regarding swift replication.19:45
annegentlekakoni: sure, go ahead. or g'ahead :)19:46
sandywalshsoren, did you see my note above about your authors/contributors test failing?19:46
sorensandywalsh: Sorry, no.19:46
sandywalshnot sure if there is some other setup I need to do to make it pass, but ./run_tests.sh fails test_authors_up_to_date19:47
kakonilets say I do updates to same object within short timespan.  First PUT goes to server X, second goes to server Y. Replicator process starts on both of these servers? Filehashes for the directories are being count and compared and.. well.. what happens next?19:48
sorensandywalsh: What's the exact error?19:48
*** seshu has joined #openstack19:49
sandywalshsoren, I'm not listed in Authors19:50
sorensandywalsh: So.. Fix it :)19:50
sandywalshhaha19:50
*** BK_man has quit IRC19:50
sorenReally. That's th eintent.19:50
sandywalshis it required that much that a test breaks?19:51
annegentlekakoni: I'll ask chuck or gholt to chime in. are you asking what happens to the object or what happens as a next process?19:52
sorensandywalsh: Tests can either fail or not. :)19:52
creihtkakoni: Each object is stored with its timestamp, latest timestamp wins19:52
sandywalshsoren, fair enough :)19:53
sorensandywalsh: It ensures the accuracy of the Authors file. I thought that was important. Two others agreed enough to approve the check.19:53
*** Ryan_Lane has joined #openstack19:53
creihtkakoni: short answer is that the most recently written version of the object will get replicated19:54
sandywalshsoren, isn't there a Contributors License Agreement (CLA)? Wouldn't that be sufficient?19:54
*** irahgel1 has joined #openstack19:54
kakonicreiht: oh yes, thanks. This is the way how understood it from the code, just wanted to doublecheck.19:54
sorensandywalsh: Well, we have an Authors file. Might as well make sure it's up-to-date.19:55
* sandywalsh smiles and nods 19:56
*** laurensell has quit IRC19:57
kakonicreiht: second question. Whats going into to the bexar release? Apart from the bugfixes?19:57
soren_0x44: Ok, so if I want to understand how Glance is supposed to fit into Nova. What do I need to read. I can read the code all I want, but it's hard to work out how it's meant to be used and fit into everything else.20:00
sorenSome of that should have been question marks.20:01
creihtkakoni: the main thing will be large file uploads20:02
creihtpossibly an s3 compatilibity layer20:02
*** spectorclan has quit IRC20:03
_0x44soren: I'd look at the etherpad doc from the first summit for the presentation that sirp and dubs presented on glance. The names of the sub-packages have changed since then, but the architecture should still be the same.20:03
*** irahgel1 has left #openstack20:03
creihtkakoni: https://blueprints.launchpad.net/swift/+spec/client-side-chunking20:03
creiht^ large files blueprint (aka client side chunking)20:03
uvirtbotcreiht: Error: "large" is not a valid command.20:03
* creiht smacks uvirtbot 20:04
*** rsampaio_ has quit IRC20:04
_0x44creiht: Oh that'll be really nice, I'll be able to get rid of the client side chunking in my upload code for glance.teller20:05
creiht:)20:05
_0x44(That isn't in a branch on lp yet)20:05
soren_0x44: You woulnd't happen to have the URL for it, would you?20:05
kakonicreiht: why not multipart like in s3?20:06
sorenAFAIK, I can't search on etherpad.20:06
_0x44soren: One moment20:06
creihtkakoni: It is funny as s3 announced their s3 multipart after we finalized the design for the client side chunking20:07
creihtBut we like our way better :)20:07
annegentlesoren: the Etherpad you seek is probably linked from here: http://wiki.openstack.org/Summit/Austin20:07
creihtthe s3 multipart is still limited to 5GB20:07
creihtyou have to track a transaction id20:07
sorenannegentle: Lovely! Thanks!20:07
_0x44soren: http://etherpad.openstack.org/Images20:08
_0x44soren: That is... a bit disjointed from all the discussion.20:08
_0x44soren: Also: http://wiki.openstack.org/ImageNotes?highlight=(image)20:10
_0x44soren: Though it's really about something else20:10
_0x44soren: Though you have just highlighted a serious deficiency on the glance side (lack of adequate documentation), I guess there needs to be a bug report.20:11
*** Podilarius has left #openstack20:11
dubsquaredis anyone aware….i installed the nova packages and there is now only two config files, instead of one for each service…is that the new standard?20:14
kakonicreiht: right. ok thanks.20:14
sorendubsquared: Yes.20:14
sorendubsquared: There's only supposed to be one, really.20:15
sorendubsquared: What are the two?20:15
dubsquarednova.conf, and nova-manage.conf20:15
sorennova-manage.conf shoulnd't be used anymore.20:15
sorenEverything should be using nova.conf20:15
dubsquarednice, i see it's much more put together…and all my services are good20:16
dubsquaredthanks for the update20:16
zulsoren: when i installed the package last night i noticed that the old configuration files were left around still20:17
*** brd_from_italy has joined #openstack20:17
annegentle_0x44: definitely, feel free to log lack of doc on Glance as a bug. Do we need to register glance.openstack.org or does it fit within Nova's site?20:17
sorenzul: Yeah. We need to do something clever about them. Not sure what exactly.20:18
zulsoren: ditto20:18
_0x44annegentle: I just submitted and triaged it as a bug. I dunno where it best sits on the site, if they start getting unweildy then maybe we should carve out its own space.20:20
annegentle_0x44: yeah I'm just trying to figure out who will use Glance - would you use glance only if you already have nova installed? Or would you use glance to store images for other controllers to access? (is that even possible)20:22
soren_0x44: There's an awful lot of ideas and "we might do this"s and "we're probably going to do that"s.20:23
pvoannegentle: I think you could use glance standalone without nova or swift20:24
sorenpvo: Cool. for what?20:24
_0x44annegentle: Ostensibly, you could have nova installed, and I could have glance installed and users of your cloud could install images from my image store/repo20:24
pvosoren: ^^20:25
_0x44soren: Yeah, it was a lot less disjointed before the people started typing. I think sirp might have a copy of their original presentation without the comments/annotations.20:25
*** dubsquared has quit IRC20:26
annegentleah so you're building a standalone "storefront" for vm images. Thanks that helps.20:26
*** kakoni has quit IRC20:26
pvopublic images, paid private images, user uploaded images20:26
_0x44soren: 20:24 <+_0x44> annegentle: Ostensibly, you could have nova installed, and I could have glance installed and  users of your cloud could install images from my image store/repo20:26
uvirtbotNew bug: #681117 in glance "Inadequate documentation of glance architecture." [Medium,Confirmed] https://launchpad.net/bugs/68111720:26
_0x44Whoa... I didn't know uvirtbot checked glance stuff20:29
creihtI think it will look up any bug20:30
creihtoh new bugs20:30
creihtheh20:30
creihtbug #66620:30
uvirtbotLaunchpad bug 666 in malone "can't file a bug on Ubuntu" [Medium,Invalid] https://launchpad.net/bugs/66620:30
creihtlol20:30
pvobug #120:31
uvirtbotpvo: Error: Could not parse data returned by Launchpad: The read operation timed out20:31
pvoconquered20:31
creihtbug #99999999999999999999999999999999999999999920:31
pvoyou broke it20:31
* creiht brings down both uvirtbot and launchpad :)20:31
sorenIt does a lot of other stuff, too.20:33
sorenSamba bug #3420:33
creihtuvirtbot: make me a sandwich20:33
uvirtbotcreiht: Error: "make" is not a valid command.20:33
creihtuvirtbot: sudo make me a sandwich20:33
uvirtbotcreiht: Error: "sudo" is not a valid command.20:33
creihtworthless :)20:34
sorenWeird.20:34
sorenhttps://bugzilla.samba.org/show_bug.cgi?id=405220:35
uvirtbotbugzilla.samba.org bug 4052 in Build environment "Interface detection code fails if build host has no interfaces" [Normal,New]20:35
sorensamba bug 405320:35
uvirtbotLaunchpad bug 4053 in alsamixergui "alsamixergui: merge new debian version" [Medium,Fix released] https://launchpad.net/bugs/405320:35
sorenfigures.20:35
sorenNever do demos. Ever.20:35
creihthah20:35
sorenbugzilla.samba.org bug 405320:35
sorenuvirtbot: I don't like you anymore.20:36
uvirtbotsoren: Error: "I" is not a valid command.20:36
creihtuvirtbot: list20:36
uvirtbotcreiht: Admin, Bugtracker, Channel, Config, Misc, Owner, RSS, Sayto, and User20:36
sorenuvirtbot: Bugtracker list20:36
uvirtbotsoren: bugs.winehq.org, bugzilla.kernel.org, bugzilla.mindrot.org, bugzilla.redhat.com, bugzilla.samba.org, cups, debian, developerbugs.linux-foundation.org, django, freedesktop, freedesktop2, gnewsense, gnome, gnome2, issues.apache.org, kde, launchpad, lp, malone, mozilla, openoffice, sourceforge, supybot, trac, ubuntu, and ximian20:36
creihtuvirtbot: sayto soren I don't like you either20:36
creihtheh20:36
creiht:)20:36
sorenI'm the only one who can do that :)20:37
creihtfigures :)20:37
jk0#65823420:37
jk0LP65823420:37
sorenI can't see the responses, though.20:37
sorenbug 65823420:37
uvirtbotLaunchpad bug 658234 in nova "Image type is not taken into consideration when generating image ID" [Low,Triaged] https://launchpad.net/bugs/65823420:37
sorenjk0: About that bug..20:44
sorenjk0: Bah, I'll follow up on the bug.20:44
jk0:)20:44
jk0I was going to take a crack at it until I saw that snippet in the code. figured I'd get some feedback before spending time trying to reproduce20:45
*** jdarcy has joined #openstack20:47
sorenjk0: The problem is that it's not the objectstore that assigns the id.20:50
sorenjk0: The API server does.20:50
sorenjk0: And the API server doesn't inspect the manifest to actually figure out if it's a kernel, ramdisk or machine image.20:50
* soren calls it a day20:50
jk0ah20:51
*** jbaker has joined #openstack21:07
*** jimbaker has quit IRC21:07
*** jbaker is now known as jimbaker21:07
*** kevnfx has joined #openstack21:18
*** ctennis has quit IRC21:26
*** jdarcy has quit IRC21:27
*** pothos_ has joined #openstack21:35
*** ctennis has joined #openstack21:38
*** pothos_ has quit IRC21:38
*** pothos has quit IRC21:38
*** pothos has joined #openstack21:38
Ryan_Lane"nova-manage user create" creates and modifies users using uid, but displays them using cn, when using the LDAP driver :D21:46
Ryan_Lanethat's awkward, to say the least :)21:47
*** burris has joined #openstack21:50
*** jimbaker has quit IRC21:53
*** EdwinGrubbs has quit IRC21:55
*** damon__ has quit IRC21:58
*** joearnold has quit IRC21:58
dendrobatesRyan_Lane: file a bug about that21:59
Ryan_Lanewell, the way the code is written right now doesn't make it problematic22:00
Ryan_LaneI'm modifying the code to work with existing servers. that's when it becomes problematic22:01
Ryan_Laneso I'll just fix that while I'm doing the other changes22:01
*** BK_man has joined #openstack22:01
*** BK_man has quit IRC22:02
*** allsystemsarego has quit IRC22:06
vishyRyan_Lane: you are more than welcome to fix / take over ldap code22:10
Ryan_Laneheh. I dunno about take over, but I'm about to send a merge request :)22:11
vishyRyan_Lane: I wrote it and I'm an ldap n00b22:11
Ryan_Lanethe ldap code is pretty good22:11
Ryan_Laneit's written under the assumption that nova owns the server though, so I'm making mods there22:11
Ryan_Lanefor most of the objects, that's true. it's unlikely for user accounts though22:12
Ryan_LaneI also entered a bug to modify the schema, so that we can interoperate with AD easier too22:12
Ryan_Lanethat'll need further code changes, so I'll push this change first, then work on that22:13
vishyI'm not sure of the value of having NovaProject be a separate object22:14
vishyprobably could have just made it a GroupOfNames22:14
Ryan_Laneprobably22:15
Ryan_Lanethough you do need extra attributes on it, right?22:15
vishyalso project_manager should probably be a role22:15
vishyRyan_Lane: not really, project manager is the only addition, which could be done using owner or a role22:16
Ryan_Laneyeah, true22:16
Ryan_Lanewell, it's still early on in the schema22:16
vishyRyan_Lane: it does make searching for projects easier since you can filter on objectclass22:17
*** miclorb has joined #openstack22:17
vishywhereas if we switched you would maybe have a hard time filtering projects vs roles22:17
vishyRyan_Lane: if you see any other issues, go at it.  I didn't really do any testing of trying to integrate with AD, although our basic plan was to use a translucent proxy22:20
Ryan_Laneah. that would work too22:20
Ryan_Lanethough it's a pretty simple fix22:20
vishyRyan_Lane: interoperability is important so the fix seems good to me22:20
Ryan_Lanemake the attribute configurable, and change the MUST uid to MUST cn in the schema22:21
vishyRyan_Lane: also there should be a differernce between user/project name and user/project id22:21
Ryan_Laneright now it is cn/uid22:22
Ryan_Lanefor user22:22
Ryan_Lanethe awkward thing is that user list displays the cn, which may not be the same as the uid22:23
*** gondoi has quit IRC22:23
vishyyes we treated them as the same for a long time22:23
Ryan_LaneI changed __to_user to set name and id to uid, instead of using cn. is that a bad idea?22:23
vishyand there are still places in the tests that assume they are the same22:24
vishy(and in the code in general)22:24
* Ryan_Lane nods22:24
Ryan_LaneI'll set that change back then22:24
vishyinetorg uses uid in its dn, not cn right?22:25
vishys/inetorg/inetorgperson/22:25
Ryan_Lanethat is configurable on the part of the ldap admin22:25
Ryan_Laneyou can set the naming attribute to a number of things22:25
vishyeventually it would be nice to enforce id as unique but allow duplicate names22:26
Ryan_Laneyou can use cn+email, for instance22:26
vishyso it seems like name should be cn22:26
Ryan_Laneis cn multi-value?22:26
vishyand id should be whatever is used in the dn22:26
Ryan_Lanehmm. id should be configurable, and likely shouldn't be what's in the DN22:27
vishyi suppose as long as you can convert from an id to a dn reliably22:27
Ryan_Lanesome people would want it to be uid, some would want samaacountname22:27
Ryan_Laneto get a DN, you should always do a search22:28
vishyor are you thinking of doing a filter on uid instead of constructing the dn directly22:28
vishyah ok22:28
Ryan_Lanebecause you can't reliably determine a DN from an attribute without doing a search22:28
vishythen probably that change should be added22:28
Ryan_Laneespecially since that can be changed on a whim from the ldap admin's perspective22:28
vishybecause right now it is just constructing the dn based on the id22:28
Ryan_Laneah22:28
Ryan_Laneyeah, that's problematic22:29
Ryan_LaneI'll make that change soon22:29
vishysee these are things I don't know because i'm not an ldap admin22:29
vishycool22:29
Ryan_Lanelet me enter a bug for it, so that I can remember22:29
Ryan_Lanewhen creating entries from scratch, it's necessary to create the DN from scratch22:31
Ryan_Lanethat should be configurable, as well as the objectclass lists22:31
*** seshu has quit IRC22:31
Ryan_Lanemost of the configurable items will only be for users though, since they are the only objects that would intersect already existing entries22:32
*** brd_from_italy has quit IRC22:38
uvirtbotNew bug: #681164 in nova "Use a search to find DNs instead of creating them directly from attributes" [Undecided,New] https://launchpad.net/bugs/68116422:41
*** londo_ has joined #openstack22:53
*** joearnold has joined #openstack22:56
edayvishy: I got a good size patch I'd like you to review when you have a moment: https://code.launchpad.net/~eday/nova/compute-abstraction/+merge/41805   :)22:59
edayvishy: not sure where you are on pushing IP allocation down, but it will conflict/hopefully help with that22:59
*** matiu has joined #openstack22:59
*** joearnold has quit IRC23:00
*** joearnold has joined #openstack23:02
*** joearnol_ has joined #openstack23:04
*** joearnold has quit IRC23:04
vishyeday: I'm hoping to get a chance to code soon23:07
vishy:(23:07
vishyeday: i don't understand 463 vs 47223:08
vishywhy does one check for none and the other just run show23:08
vishy?23:08
edayvishy: good q, I should probably check for none on all of those23:09
vishyso show() returns none instead of raising NotFound like the old code does?23:10
vishyalso the exception should probably be NotFound or ApiError rather than just Exception23:10
edayvishy: hmm, well we could have show() throw instead I suppose23:10
edayahh, in images API, it says it's supposed to raise NotFound, so I'll fix that23:11
edayvishy: that check is from the openstack API, I was trying to merge the logic of each and missed a few things I guess :)23:12
vishycoolness23:13
vishyother than that it looks good23:13
edayok, I just removed those 2 lines... anything else?23:13
vishywe could use some better testing23:13
vishyin general of api commands23:13
vishypreferrable without race conditions like the ones i added for public ips :(23:14
edayyeah.. in theory this should preserve all behavior, but I know coverage isn't probably all that good with the unittests23:14
vishywe need to start running smoketests!23:14
vishyand figure out how to poll properly for cloud uinttests23:15
edayand we need to start reporting on coverage so we know what to add tests for23:15
vishyso they work without greenthread.sleep()23:15
vishythose seem like the most important additions23:15
vishyi'm sure there are a bunch of api commands which don't get tested23:15
vishyaustin threw number of argument exceptions in describe_addresses :(23:16
*** joearnol_ has quit IRC23:17
vishyeday: I assume all quota checking will eventually move into the shared code?23:17
edaywe focused so much on new features at the summit, I don't think we talked much about making what we have more solid23:17
edayyeah, I hope to push almost everything into worker APIs, like this create_instance patch23:18
*** Ryan_Lane is now known as Ryan_Lane|away23:19
edaythis was just the big one, but now we can go through and abstract more API requests23:19
vishyeday: yeah seems good23:19
edaythe API server code in nova.api should be a simple kwargs translation, should have little-to-no logic23:19
*** joearnold has joined #openstack23:20
*** kevnfx has quit IRC23:20
vishywe should make a bug for cleaning up cloud unittests, fixing polling issues and adding tests for all of the api commands23:21
edaywell, I'd love to get everything converted to nosetests first with code coverage enabled23:21
vishyi suppose if we are breaking things out like this, we could just stub/mock the managers23:21
sorenYou know we have code coverage checks already, right?23:22
sorenJust sayin'.23:22
edayoh?23:22
vishyand the testing won't depend on eventlet/greenlet23:22
sorenI'd tell you about it, but I'm nothere.23:22
vishypolling of rpc23:22
soreneday: http://hudson.openstack.org/job/nova-coverage/23:22
edayahh, the hudson job is finally running :)  last I checked it was dead23:22
soreneday: I fixed it weeks ago.23:23
sorenPossibly at the sprint.23:23
sorenErr... summit.23:23
sorenThe fix was ridiculously simple.23:23
edayahh, must have missed it23:23
soren"python-coverage run_tests.py"  --> "python-coverage $(pwd)/run_tests.py"   or something to that effect.23:23
vishysoren: that's sexy23:24
vishysoren: i didn't know that was working either23:24
sorenvishy: I didn't feel very sexy went I worked it out. Hair all ruffled, sweating..23:24
jk0soren: that was probably the time someone asked you a question during a talk, and you weren't paying attention :P23:24
sorenvishy: Oh, the graph? I thought you meant the fix :)23:24
edaysoren: can you get nosetests we run in there too?23:25
soreneday: I... err...23:25
sorenno.23:25
soren:)23:25
sorenProbably, but I'm not sure it'd be worth the effort, if we're consolidating our test frameworks.23:25
edaysoren: we're consolidating to the nosetests though :)23:26
sorenjk0: That never happened :)23:26
soreneday: It's simple for nosetests. It's also (as it turns out) simple for non-nose tests.23:26
soreneday: Both at the same time? Less so.23:26
jk0don't worry, we all assumed you were busy being productive :)23:26
soren:)23:26
edaysoren: ahh, I was hoping you could just dump both commands on the args list and be done with it23:27
sorenWell, it's way past my bed time, just stopped by to add to the confusion.23:27
soreneday: I don't thinkg so.23:27
soreneday: If you have a python script that runs both, I guess it would work.23:27
soreneday: "python-coverage foo.py arg1" corresponds to "python foo.py arg1", just with code coverage checking enabled.23:28
soreneday: Hmmm..23:28
edayahh23:28
soreneday: Actually, it may be really easy.23:28
soreneday: If the running of the tests and the generation of the report are separate steps, it's simple as pie.23:28
sorenAnyways... Tomorrow.23:29
* soren -> sleep23:29
soreng'night.23:29
*** Ryan_Lane|away has quit IRC23:34
*** ppetraki has quit IRC23:37
*** abecc has quit IRC23:53
*** rsampaio has joined #openstack23:56
*** GasbaKid has quit IRC23:59

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!