Wednesday, 2010-12-15

*** schisamo has joined #openstack00:05
*** dagger has quit IRC00:11
*** dagger has joined #openstack00:11
*** dagger has joined #openstack00:11
*** schisamo has quit IRC00:12
*** _skrusty has quit IRC00:13
*** kashyapc has quit IRC00:17
*** kashyapc has joined #openstack00:17
*** sophiap_ has quit IRC00:19
*** sophiap has joined #openstack00:20
*** _skrusty has joined #openstack00:27
*** aliguori has quit IRC00:27
*** kashyapc has quit IRC00:28
*** kashyapc has joined #openstack00:29
*** AimanA has quit IRC00:30
*** kashyapc has quit IRC00:34
*** bobl has quit IRC00:35
*** aliguori has joined #openstack00:36
*** miclorb has quit IRC00:36
*** rlucio has quit IRC00:42
alekibangoreboot not working?00:42
alekibangobtw tr3buchet are u using nova-volume?00:42
alekibangotr3buchet: what us 'cloudservers' ?00:44
sandywalshalekibango, cloudservers is an openstack fork of a cmdline tool for the rackspace api00:46
alekibangoah, interesting00:46
sandywalshalekibango, reboot works ... we just have a situation in xenserver right now where we have to reboot instances after creation to get 'em going.00:46
alekibangoah, ic00:47
*** kashyapc has joined #openstack00:47
sandywalshand just showing off pause/unpause00:47
alekibangoare u using nova-volume?00:47
tr3buchetno not using volume00:47
alekibangobtw about that partitioning, if you would find some time tomorrow to talk with me about it, it would help a lot00:48
tr3buchetwe're on crazy different time zones however00:48
alekibangowhich time is ok for you?00:48
tr3buchet4:30pm gmt until around 12gmt00:49
alekibangothats good. i will eb here00:49
alekibangosandywalsh: some documentation for cloudservers would be nice00:51
alekibangoi will prolly start using it soon00:51
*** miclorb has joined #openstack00:51
alekibangodouble decorators = fun00:52
alekibangooh,. even 7 at once00:52
sandywalshalekibango, beyond the readme? (bottom of the github page)00:52
tr3buchetabout as fun as double pointers or nested lambdas00:52
alekibangoi hated lambda00:53
alekibangosandywalsh: ah, ty, was blind00:53
*** jasonb365 has joined #openstack00:54
tr3buchetalekibango: i had written some code a few months ago that had 3 levels of lambdas.00:55
*** miclorb has quit IRC00:55
tr3buchetsandywalsh can verify00:55
tr3buchetfixed now however..00:55
* sandywalsh shudders00:55
alekibangothats sick00:55
tr3buchetsee you guys tomorrow!00:56
sandywalshnight tr3buchet00:56
alekibangoi never loved lambda calcul... and lisp... and postscript...   even perl is more nice :)00:56
*** MarkAtwood has joined #openstack00:56
alekibangotr3buchet: gn800:56
*** aliguori has quit IRC00:56
tr3buchetif only perl had real object orientation.... if only..00:56
*** jasonb365 has quit IRC01:01
*** evilrob has joined #openstack01:03
*** kevnfx has joined #openstack01:04
* evilrob is sitting in ACUG meeting... may need to start using this at work with xenserver... 01:06
*** BK_man has joined #openstack01:08
*** ar1 has joined #openstack01:10
*** jasonb365 has joined #openstack01:12
dabotr3buchet: "if only perl had real object orientation"... it would be another Ruby! :)01:12
zulvishy: ping at one point did you say that anso labs was going to open source a provisioning tool at the summit?01:13
vishyexplain provisioning tool01:13
zulvishy: a tool for deploying openstack :)01:14
zuli guess similar to puppet and the likes01:14
vishywe put our puppet recipes in contrib01:15
vishywith some info stripped01:15
vishywith the hope that people would modify/improve them01:15
zulah cool ok thanks01:15
*** dirakx has quit IRC01:21
*** sophiap has quit IRC01:24
*** sophiap has joined #openstack01:24
*** jasonb365 has quit IRC01:32
*** dragondm has quit IRC01:33
*** rds__ has quit IRC01:34
*** jc_smith has quit IRC01:39
*** vvuksan has joined #openstack01:44
*** vvuksan has quit IRC01:49
*** piken has quit IRC01:53
*** winston-d has left #openstack02:00
*** MarkAtwood has quit IRC02:03
*** MarkAtwood has joined #openstack02:08
*** sophiap has quit IRC02:09
*** sophiap has joined #openstack02:09
*** vvuksan has joined #openstack02:12
*** schisamo has joined #openstack02:18
vvuksanwho is in charge of Nova documentation ?02:35
*** joearnold has quit IRC02:38
*** miclorb has joined #openstack02:41
*** jdurgin has quit IRC02:44
*** Ryan_Lane has quit IRC02:47
*** sandywalsh has quit IRC02:51
*** vvuksan has quit IRC03:01
*** maplebed has quit IRC03:05
*** schisamo has quit IRC03:33
*** kashyapc has quit IRC03:38
*** rchavik has joined #openstack03:39
*** jasonb365 has joined #openstack03:42
*** jasonb365 has quit IRC03:47
*** jimbaker has quit IRC03:55
*** annegentle-web has joined #openstack03:55
*** BK_man has quit IRC04:05
*** daleolds has quit IRC04:05
*** kashyapc has joined #openstack04:08
*** BK_man has joined #openstack04:18
*** EdwinGrubbs has joined #openstack04:21
*** EdwinGrubbs has joined #openstack04:21
*** ccustine has quit IRC04:22
*** ccustine has joined #openstack04:22
*** ccustine has quit IRC04:25
*** herki has quit IRC04:29
*** BK_man has quit IRC04:30
*** schisamo has joined #openstack04:45
*** schisamo has quit IRC04:50
*** annegentle-web has left #openstack04:51
*** dubsquared has joined #openstack05:19
*** HugoKuo has joined #openstack05:32
*** f4m8_ is now known as f4m805:45
*** f4m8 is now known as f4m8_05:46
*** ccustine has joined #openstack06:03
*** ccustine has quit IRC06:03
*** dubsquared has quit IRC06:16
*** stewart has quit IRC06:35
*** reldan has joined #openstack06:43
*** apanasenko has joined #openstack06:45
*** guigui1 has joined #openstack06:46
*** stewart has joined #openstack06:52
*** kashyapc has quit IRC06:53
*** kashyapc has joined #openstack06:53
*** kashyapc has quit IRC06:58
*** kashyapc has joined #openstack06:58
*** almaisan-away is now known as al-maisan07:02
*** reldan has quit IRC07:05
*** ArdRigh has joined #openstack07:06
*** guigui1 has quit IRC07:08
*** stewart has quit IRC07:13
*** apanasenko has quit IRC07:16
*** arthurc has joined #openstack07:32
openstackhudsonProject nova build #303: SUCCESS in 1 min 13 sec:
*** sophiap has quit IRC07:53
*** allsystemsarego has joined #openstack07:54
*** GasbaKid has joined #openstack07:54
*** sophiap has joined #openstack07:56
*** brd_from_italy has joined #openstack08:00
*** arthurc has quit IRC08:01
*** miclorb has quit IRC08:12
*** larstobi has joined #openstack08:16
*** fabiand has quit IRC08:16
*** fabiand has joined #openstack08:16
*** arthurc has joined #openstack08:17
*** Podilarius has left #openstack08:18
*** zul has quit IRC08:23
*** arthurc has quit IRC08:26
*** ibarrera has joined #openstack08:29
*** calavera has joined #openstack08:30
*** arthurc has joined #openstack08:34
*** zul has joined #openstack08:35
*** sophiap has quit IRC08:55
*** sophiap has joined #openstack09:02
*** arthurc_ has joined #openstack09:17
*** arthurc has quit IRC09:19
*** nelson__ has quit IRC09:27
*** deshantm_cosi has quit IRC09:28
*** deshantm has quit IRC09:29
*** irahgel1 has joined #openstack09:34
*** deshantm_cosi has joined #openstack09:41
*** deshantm has joined #openstack09:41
*** reldan has joined #openstack09:53
*** pothos has quit IRC10:01
*** pothos has joined #openstack10:02
*** doude has joined #openstack10:05
*** guigui1 has joined #openstack10:09
*** olivier_ has joined #openstack10:14
olivier_Hi all, I've got a problem for starting my swift-proxy server and /var/log/message didn't show me error message. How can I troubleshoot it ?10:29
larstobiolivier_: you need to add catch_errors-filter10:30
olivier_ok, I will do that, thanks10:30
*** DigitalFlux has joined #openstack10:32
larstobiolivier_: also, did you make sure syslog is running? catch_errors-konfig excerpt:
DigitalFluxHi guys10:33
DigitalFluxI am trying to bootstrap nova using the nova script10:33
*** kashyapc has quit IRC10:34
DigitalFluxeverything is up and working from what i see on the console logs10:34
DigitalFluxhowever when i run the euca-add-keypair command, i just get a connection refused10:34
DigitalFluxwhat could this be related to ?10:34
olivier_Still nothing on my log (and rsyslogd is runnig)10:43
olivier_I've adapted my proxy-server.conf whith your excerpt, and try to start it, but still nothing10:44
larstobidid you start auth-server okay?10:45
larstobiwould you mind pasting your proxy-server.conf to
*** sebastianstadil has joined #openstack10:47
larstobihmm, it's almost exactly like mine10:47
olivier_Can it be related to a problem with the ring files ? (I meet an index error when I rebalance the ring files)10:48
larstobiI'm not sure about that10:49
larstobiwhat was the index error?10:49
larstobiolivier_: do you only have 1 zones? you should have a minimum of 310:51
larstobiolivier_: also, which version of swift are you running?10:53
olivier_ok, it's a lab test, then I beleive that with 1 zone it should be enough10:53
olivier_But if it's mandatory to have 3 zone minimum, I will create 310:53
olivier_How to know the swift version ?10:54
larstobihmm, did you install it from trunk or downloaded zip file?10:55
olivier_I'm using ppa10:55
larstobiok, then it's probably 1.1.010:55
larstobijust do a "apt-cache policy swift" to chech which is installed10:57
olivier_I've followe.1.0-1ubuntu1~lucid010:58
olivier_oups: .1.0-1ubuntu1~lucid010:58
larstobiI guess that should be 1.1.0-1ununtu1~lucid010:59
olivier_yes (bad copy/past)10:59
larstobiI figured :)10:59
olivier_Then, If I need a minimum of 3 zones with 3 storage nodes, I need to put each storage node in a different zone ?11:00
larstobiI'm not sure if this is the reason proxy-server won't start11:02
*** sebastianstadil has left #openstack11:03
*** sebastianstadil has joined #openstack11:04
*** sophiap has quit IRC11:04
*** sophiap has joined #openstack11:05
*** apanasenko has joined #openstack11:08
olivier_I've fixed the ring files, and still nothing with proxy11:16
*** nelson__ has joined #openstack11:35
*** nelson__ has joined #openstack11:35
larstobiolivier_: are you starting the proxy-server via swift-init?11:38
larstobimaybe if you try to start it like this, you might get some error message: "swift-proxy-server /etc/swift/proxy-server.conf"11:39
*** stewart has joined #openstack11:40
*** arthurc_ has quit IRC11:40
olivier_I'm using the command: swift-init proxy start11:40
*** arthurc has joined #openstack11:40
olivier_Error trying to load config /etc/swift/proxy-server.conf: Entry point 'catch_errors' not found in egg 'swift' (dir: /usr/lib/pymodules/python2.6; protocols: paste.filter_factory, paste.filter_app_factory; entry_points: )11:40
larstobiokay, maybe that's a new config option from trunk (I'm running trunk). Try to remove catch_errors11:41
olivier_now it didn't give my shell back: Is a good news ?11:42
*** schisamo has joined #openstack11:42
larstobiyes :)11:42
larstobiI think you'll have to kill −9 the process to exit11:43
olivier_ok great11:43
olivier_Thanks a lot's, now I can begin to test swift :-)11:45
larstobino problemos! :-) are you the freenas guy?11:45
larstobikudos! :)11:46
olivier_thanks :-)11:46
*** _skrusty has quit IRC11:46
olivier_But my swift tests are not related to FreeNAS11:46
larstobimy guess is that swift isn't a good fit for embedded systems anyway ;)11:48
olivier_You're right :-)11:48
*** stewart has quit IRC11:48
*** schisamo has quit IRC11:49
*** dantti_work has quit IRC11:52
*** rds__ has joined #openstack11:54
*** dagger has quit IRC11:55
*** dagger has joined #openstack11:55
*** dagger has joined #openstack11:55
*** stewart has joined #openstack11:59
*** dagger has quit IRC11:59
*** _skrusty has joined #openstack11:59
*** zul has quit IRC11:59
*** dagger has joined #openstack12:01
*** dagger has joined #openstack12:01
*** hggdh has quit IRC12:02
*** sophiap has quit IRC12:05
*** sophiap has joined #openstack12:07
*** schisamo has joined #openstack12:08
*** ArdRigh has quit IRC12:14
*** ar1 has quit IRC12:14
*** sandywalsh has joined #openstack12:18
*** matclayton has joined #openstack12:23
*** sophiap has quit IRC12:32
*** sophiap has joined #openstack12:33
* soren dances and jumps12:45
larstobi*jumps along*12:48
sorenIt's hard not to!12:49
* soren is calm again13:04
*** zul has joined #openstack13:09
*** westmaas has joined #openstack13:09
*** piken has joined #openstack13:11
sorenBacked by iptables.13:12
sorenIe. actually works.13:12
soren*sigh* Just the SOAP thing left now.13:18
alekibangoi use soap only in bathroom ;)13:19
sorenSensible choice!13:19
larstobivery nice :)13:19
alekibangosoren: we are in agreement on soap :)13:20
sorenI know one particular place in a bath room where I'd like to stick SOAP.13:20
alekibangosoren: i hope there will be some docs/examples for this feature.... that would be nice13:21
sorenalekibango: The soap thing?13:21
larstobiI like to rest in the bathroom and also use soap there13:21
alekibangono, the thing ou pasted13:21
sorenalekibango: Oh.13:21
alekibangosecuity groups13:22
sorenI'm sure there's a bunch already. Whatever you can find about EC2 and security groups should hold true for this, too.13:22
larstobinot so funny, I guess...13:22
*** GasbaKid has quit IRC13:23
alekibangosoren: can i ask you for help? it should be easy for you... please answer this question. i just dont feel confident enough:
alekibangosoren: linking relavant ec2 docs  would be nice too :)13:24
alekibangosoren: also relevant:
sorenalekibango: done13:27
alekibangosoren: imho docs will be critical issue around next release13:28
alekibangowe need to put them in shape13:28
alekibangoand its hard for people without know-how to write it13:29
alekibangoalmost impossible13:29
*** sophiap has quit IRC13:29
*** sophiap has joined #openstack13:30
sorenalekibango: That's why we have the super awesome annegentle  :)13:30
alekibangoshe is awesome, but not superman13:30
alekibangosoren: can you please answer one of those 2 linked questions ?13:31
*** sebastianstadil has quit IRC13:31
alekibangoyou will get some karma boost and thanks of many people who fight with it again and again... we will rewrite it soon into docs somehow13:31
sorenalekibango: Which ones?13:37
sorenI already did.13:37
soren13:27 <+soren> alekibango: done13:37
sorenThat was me having done so.13:37
alekibangoah i dont see it there13:37
sorenIt's like totally there. It says "10 minutes ago" and everything.13:37
alekibangoah its there now. sorry was prolyl cached somewhere13:37
alekibangosorry :D13:38
sandywalshtermie, you around?13:41
*** aliguori has joined #openstack13:42
*** ctennis has quit IRC13:42
sandywalshperhaps this is just a more recent trunk issue13:42
sandywalshattempting 'nova-manage project zipfile' and getting complaints of13:42
sandywalshopenssl.cnf not found13:43
sandywalshis this a new requirement?13:43
*** kashyapc has joined #openstack13:43
sandywalshsorry ... skipped a step ... my bad13:44
*** hazmat has joined #openstack13:45
reldanHi guys. Nova doesn't support guest windows OS (net and key injection), do it?13:55
reldanAre there additional complexity to add support windows server?13:57
sandywalshreldan, is there a particular hypervisor you're considering?13:57
*** hggdh has joined #openstack13:57
*** ctennis has joined #openstack13:58
*** ctennis has joined #openstack13:58
reldansandywalsh: No. I want to find openstack restrictions on guest OS.14:02
olivier_I've still a last problem with swift: swift-auth-add-user can't add a user (Update failed: 503 Service Unavailable)14:02
alekibangosoren: you are missing dependency on sudo14:02
*** vvuksan has joined #openstack14:02
alekibangoin your packages14:02
reldanalekibango: thanks for blueprint14:03
alekibangonp reldan, but please consider not using windows14:03
sorenalekibango: I'm whuh?14:03
alekibangosoren: your nova packages call sudo :)  but doesnt depend on it14:03
sorenalekibango: Oh.14:03
sandywalshreldan, I don't believe there are restrictions in nova itself. The restrictions would really come from the hypervisor.14:04
sandywalshreldan, the nova restrictions would come from the functions you can perform on the guest (a la, the bp alekibango cited)14:05
reldanalekibango: Rackspace supports Windows Server, I think nova also should :)14:06
sorenIt does.14:06
sorenWell, it will once my branch from this morning langs.14:06
sorenlands, even.14:06
alekibangobut note that by using windows you support racism, luciferianism and genocide (by giving bill gates money)14:06
alekibangojust search those words along his name14:06
reldanalekibango: :)14:07
apanasenkosandywalsh: a hypervisor does not allow to use Windows as a guest OS?14:09
reldansandywalsh: - are you sure, I can inject net into fs such way in Windows14:10
apanasenkoalekibango: Bill no longer works in MS :)14:10
*** rchavik has quit IRC14:10
alekibangoapanasenko: right but he still have some shares, doesnt he?14:10
*** sebastianstadil has joined #openstack14:11
daboreldan: that code is changing. the original design was linux-only.14:12
apanasenkoalekibango: Well, yes :)  But still all fault, Steve :)14:12
alekibango???  i dont get you now :)14:13
sandywalshreldan, yes, you're correct with the current implementation. It will support windows though. As soren mentioned, it is a priority for rackspace.14:13
reldandabo: sandywalsh: thanks14:13
*** hggdh has quit IRC14:13
sandywalshtermie is west-coast US, correct?14:14
sorensandywalsh: He's in Berlin, isn't he?14:14
sorensandywalsh: He used to be.14:14
sorenMaybe he moved. I don't know.14:14
sorenErk, I'm late.14:15
*** hggdh has joined #openstack14:17
*** hggdh has quit IRC14:23
*** gondoi has joined #openstack14:24
*** hadrian has joined #openstack14:24
* hadrian cannot find any doc on how to create images for nova14:24
*** ppetraki has joined #openstack14:27
*** Podilarius has joined #openstack14:34
*** hggdh has joined #openstack14:34
*** jasonb365 has joined #openstack14:35
*** sophiap_ has joined #openstack14:37
*** sophiap has quit IRC14:37
*** sophiap_ is now known as sophiap14:37
*** kevnfx has quit IRC14:39
*** reldan has quit IRC14:40
*** reldan has joined #openstack14:40
*** jasonb365 has quit IRC14:43
*** doude has quit IRC14:48
*** doude has joined #openstack14:48
*** larstobi has quit IRC14:53
*** larstobi has joined #openstack14:53
*** apanasenko has quit IRC14:54
*** apanasenko has joined #openstack14:55
*** apanasenko has quit IRC14:55
annegentlehadrian: does this help?
*** westmaas has quit IRC14:58
*** dirakx has joined #openstack14:59
hadrianannegentle: i was aware of that link, I was looking for ways to create your own image15:01
hadrianannegentle: I assume you are the author of the youtube videos, correct?15:02
hadrianawesome work, highly appreciated15:02
annegentlehadrian: does this readme help with "Do It Yourself" images?
*** hggdh has quit IRC15:03
smoserit probably doesn't help hadrian too much15:04
chmouelsandywalsh: by the way rscloud server are not implementing pause so that could confusing to get that method on the python-cloudservers binding15:05
smoserhadrian, i think the most documentation on building images that you're going to find is going to be related to building EC2 images.15:05
chmouelsandywalsh: I would vote to fork it and call something like with swift or nova in the name15:05
hadriansmoser: there's plenty of that15:05
hadrianbut from what i gather they won't work in nova, unless...15:05
* hadrian is looking for the the unless part :)15:05
smoserIf it fits your bill, I think that downloading an image from and starting from that is the path of least resistance.15:06
sandywalshchmouel, yes, that's a good point.15:06
smoserthere are lots of "gotchas" that qualify as your "unless".15:06
sandywalshchmouel, or we could make it a flag to enable openstack commands15:06
smoserthe biggest 2 differences from EC2 -> openstack are15:06
hadriansmoser: not the answer I was looking for, but very helpful nonetheless :)15:06
smosera.) devices: virtio nework and disk, and 'root=/dev/vda' rather than 'root=/dev/sda'15:06
chmouelsandywalsh: yeah true but eventually rackspacecloud would become openstack nova so people would just migrate to that library when the switch happen15:07
*** sebastianstadil has quit IRC15:07
hadriani need java and an osgi container on my vms, that's the reason i am asking15:07
*** hggdh has joined #openstack15:08
smoserb.) depending on the hypervisor you're using in openstack (likely kvm), the kernel and modules are not going to be be xen kernel/modules. they'll need to run on kvm.15:08
sandywalshchmouel, hmm, correct.15:08
*** allsystemsarego has quit IRC15:08
smoserhadrian, download uec image, extract, mount loopback, chroot, apt-get install, configure, exit, umount, and go.15:08
hadriansmoser: correct15:08
hadriani also have a local ubuntu mirror, so it's fast too15:09
smoseralternatively, you can start from something like vm-builder, but starting at the uec-image level just removes one step.15:09
*** hggdh has quit IRC15:09
*** allsystemsarego has joined #openstack15:10
hadriansmoser: is this process something you guys focus on improving/documenting in the near future?15:10
smoserwell... sort of. but help would not be denied.15:11
smoseri have lots of work to do there, and it is a goal to make the ubuntu images very good starting points for UEC , EC2, or Openstack.15:12
smoser covers some of the things that i'm hoping to do15:12
hadriansmoser: noted, but I didn't doubt that :)15:12
olivier_I need some help for using swift: I can't create a user with swift-auth-add-user15:12
*** reldan has quit IRC15:15
*** daleolds has joined #openstack15:15
*** allsystemsarego has quit IRC15:15
smoserhadrian, is there a part of that process that you think could stand for some help ?15:16
hadriansmoser: i have a bit more to learn to be an effective helper, but getting there...15:17
smoser might be helpful. a bit ec2 centric though.15:18
hadrianright now i am trying to fine tune the process of setting up nova on my system and getting the vms running with a minimum of time spent15:18
hadriansmoser: but i am reporting here issues i find15:19
* hadrian should get a launchpad account15:19
*** hggdh has joined #openstack15:20
smoserhadrian, yes you should. it comes with 2G of cloud storage (Ubuntu One) and lets you open bugs against me.15:21
*** jfluhmann has joined #openstack15:22
* hadrian will, but he has too many accounts already15:22
smoseri don't see mention of 'uec-publish-tarball'  or uec-publish-image ( a text search on . but that is a very easy way to publish Ubuntu images, assuming euca2ools set up.15:22
annegentlesmoser: I can add that info to the "GettingImages" page if you think that's a good place15:23
smoserwell, launchpad functions as a openid provider, so you can ditch all the others. (no, i don't reallly think that is true)15:23
*** allsystemsarego has joined #openstack15:26
*** allsystemsarego has joined #openstack15:26
*** colinnich has joined #openstack15:30
sandywalshwhat is the default logfile location if you don't supply the --logfile option?15:30
chmouelsandywalsh: current directory i think15:31
chmouelwould need to double check tho15:31
sandywalshchmouel, sadly, nothing.15:33
hadriansmoser: there you go: hzbarcea@launchpad :)15:35
* smoser assigns dozens of bugs to hzbarcea15:36
smoseri'm 'smoser' there, so if you open some, feel free to subscribe me.15:36
hadriansmoser: thanks15:37
larstobiolivier_: are you getting error messages?15:38
annegentlesmoser: thanks!15:38
*** pvo_school is now known as pvo_in_oz15:40
olivier_larstobi: Yes15:42
larstobiolivier_:  can you do a paste?15:42
larstobisounds familiar15:45
*** dfg_ has joined #openstack15:45
olivier_do you have any clue ?15:47
larstobiolivier_:  have you specified auth ip's like this: ?15:47
olivier_I've forgot this step !15:48
olivier_I'm following the guide here:
larstobiyes, I believe there are some parts missing15:49
*** reldan has joined #openstack15:49
*** dfg_ has quit IRC15:50
larstobiI must go home now, dinner time :)15:50
*** larstobi has quit IRC15:50
olivier_thanks for your help15:51
*** reldan has quit IRC15:51
*** reldan has joined #openstack15:52
annegentleolivier_: feel free to log a bug against the multi-node install instructions. I'll work on adding that15:54
*** rbergeron has quit IRC15:55
*** rbergeron has joined #openstack15:55
olivier_ok, I will do that once I will reach to use it15:55
annegentleolivier_: ok, thanks for testing the instructions and letting us know what you see :)15:56
*** apanasenko has joined #openstack15:59
xtoddxjaypipes: around?16:01
*** mdomsch has joined #openstack16:01
jaypipesxtoddx: yup16:05
xtoddxjaypipes: I can't branch  lp:~jaypipes/openstack-common/options16:05
xtoddxbzr: ERROR: Not a branch: "bzr+ssh://".16:06
jaypipesxtoddx: hmm...must have something to do with the reorganization of the core teams... ttx?16:06
ttxlet me have a look16:07
ttxopenstack-core was removed. but it had branches.16:08
ttxthey are now under ~registry instead of ~openstack-core16:08
jaypipesttx: registry?16:09
ttxthat the owner for orphaned things16:09
* ttx looks at ways to undo damage16:09
jaypipesttx: I can push the branch to some other location... I have it locally of course.16:09
*** dubsquared has joined #openstack16:10
ttxjaypipes: if we are to follow we should have ~openstack-common-core16:11
ttxjust a sec, looking at how openstack-common project is currently organized16:11
ttxjaypipes: do you plan to migrate openstack-common under ~openstack-hudson at some point ?16:14
*** sophiap has quit IRC16:14
*** sophiap has joined #openstack16:14
ttxjaypipes: I can create an openstack-common-core group, make you the admin of it, and then you can push the branch there and we can make it the main branch16:16
*** dragondm has joined #openstack16:16
*** jimbaker has joined #openstack16:17
jaypipesttx: go for it :)16:18
*** westmaas has joined #openstack16:18
*** sophiap has quit IRC16:18
colinnichHi. Anyone from swift up for a problem with stats-collector?16:21
ttxjaypipes: --> (let me know when the branch is pushed, I'll fix the lp:openstack-common link)16:23
ttxjaypipes: you could even "bzr branch lp:openstack-common" then "bzr push lp:~openstack-common-core/openstack-common/trunk", I suppose16:26
*** jasonb365 has joined #openstack16:28
ttxthat said, all this will just restore your ownership of the main branch... not really fix old references to bzr+ssh:// ...16:28
creihtcolinnich: ask away, and someone will answer when they can :)16:29
colinnichok :-)16:30
colinnichRunning a multi server Austin install on Lucid. Have several hundred Gb stored and am trying to get the stats working.16:30
colinnichI've followed the SAIO stats instructions and adapted for multi-server and the account stats are being generated and uploaded as are the proxy logs16:31
colinnichBut when stats-collector runs, it outputs an error to syslog and seems to hang - I can leave it for hours and it doesn't progress past file #116:31
creihtolivier_: yes you need to have at least 3 zones16:32
colinnichany ideas?16:32
daborunning out to donate blood. back in a few.16:32
creihtand just saw that you got it figured out... great!16:33
*** kashyapc has quit IRC16:34
creihtcolinnich: notmyname just walked in and should be on shortly and be able to help you16:34
notmynamecolinnich: hi16:35
*** zul has quit IRC16:35
*** jbryce has joined #openstack16:36
colinnichnotmyname: Hi John16:36
notmynamecolinnich: give me a minute to catch up16:36
colinnichnotmyname: np16:36
notmynamecolinnich: that traceback is a bug that has been fixed. let me check to see when it was added16:37
notmynamethe lazy-load of the internal_proxy attribute had a bug16:38
creihtcolinnich: are you running the 1.1 release?16:38
colinnichpossibly not16:38
creihtthe 1.1 release wouldn't have the patch16:39
creihtnotmyname: we probably should backport that16:39
notmynamenot a bzr expert, but bzr blame says fixed in118.2.116:39
colinnichnotmyname: it is 1.116:39
*** sebastianstadil has joined #openstack16:39
*** romohr has joined #openstack16:39
*** sophiap has joined #openstack16:40
notmynamecolinnich: creiht:
uvirtbotLaunchpad bug 671236 in swift "LogProcessor fails first attempt to use it's InternalProxy" [High,Fix committed]16:43
jk0eday: did you get a chance to look over the latest updates on ? :)16:44
sandywalshjk0, have you tried to run that branch yet?16:45
jk0not yet16:45
jk0haven't had a chance16:45
sandywalshjk0, I can't authenticate to it (euca-describe 403's)16:45
jk0interesting. I'll try now16:45
sandywalshI think I see what's happening16:47
sandywalshthe flagfile isn't getting passed in since it uses a new launcher (instead of the screen approach)16:47
sandywalshyup, that was it16:48
jk0whew :)16:48
sandywalshheh, yeah16:48
sandywalshk, off to lunch!16:48
*** kashyapc has joined #openstack16:51
colinnichnotmyname: It's flying along now thanks. Doesn't like any lines in the access files (20116 bad, 20116 total) but I'll try and figure that one out. Unless you know :-)16:51
colinnichnotmyname: "stats" files don't have any errors logged though, so fingers crossed16:51
colinnichnotmyname: swift Log processing done (0.17 minutes)16:52
notmynamecolinnich: are you using a separate access log or are you using /var/log/syslog? If it were me, I'd pull down one of those access logs from the swift cluster and see what it looks like16:53
*** normanrichards has joined #openstack16:54
colinnichnotmyname: using the syslog-ng instructions from the docs. I'll get a file down16:54
*** jasonb365 has quit IRC16:57
colinnichnotmyname: here's a section I'd say was representative -
*** brd_from_italy has quit IRC16:58
*** seshu has joined #openstack16:58
notmynameat first glance they look ok16:59
*** calavera has quit IRC16:59
colinnichnotmyname: every file processed had every line bad16:59
notmynameit could be the server name17:00
notmynamelet me check17:00
*** cloudmeat has joined #openstack17:00
notmynamecolinnich: in /etc/swift/log-processing.conf, what do you have for server_name under [log-processor-access]17:00
notmynameif it's not there, it defaults to "proxy"17:00
notmynamewhich will not match your logs and reject every line17:01
colinnichnotmyname: don't have anything.17:01
notmynameok, that's the issue then17:02
notmynameadd "server_name = proxy-server"17:02
notmynamethen remove the processed_files object in the log_processing_data container and rerun the stats (this will reset everything and reprocess all files)17:03
*** daleolds has quit IRC17:03
colinnichnotmyname: seems to be working now17:03
colinnichnotmyname: proxy-server isn't my machine's name though, so where is that from?17:03
notmynamethat's the name of the process that was logging17:04
notmynamein this case, "Server" is software, not a machine17:04
jk0sandywalsh: fwiw, I'm running the eventlet_merge branch and it's working fine -- although I am not using the new launcher17:05
jk0so it gets my thumbs up (cc eday :))17:05
colinnichnotmyname: so is that a bug then, or is it fixed already? :-)17:05
notmynamejust submitted a bug for the default server_name (
creihtcolinnich: yeah the logging was changed a while back to me more consistent in the names of the processes that were logged17:07
uvirtbotLaunchpad bug 690740 in swift "change stats default server name to proxy-server" [Undecided,New]17:07
creihtunfortunately I missed the fact that the stats stuff looked for that name17:07
colinnichnotmyname: creiht: ok, thanks guys17:07
*** lvaughn has quit IRC17:07
notmynamecolinnich: the other bug (with internal_proxy) will be backported to 1.117:08
notmynameworking on that now17:08
*** lvaughn has joined #openstack17:08
*** kevnfx has joined #openstack17:08
creihtttx: did we ever make any headway on what we want to do for releases between major releases?17:08
colinnichnotmyname: cool. I'll be getting some decent hardware soon to set up a production cluster, so expect to hear from me :-)17:08
ttxcreiht: I mentioned it to dendrobates, he couldn't elaborate but his answer was "we don't"17:09
ttxcreiht: I still need to catch-up with him on that.17:10
*** aliguori has quit IRC17:10
creihtwell *we* (swift) need to :)17:10
creihtWe have people running on 1.1 already17:10
creihtWe need to be able to backport crucial fixes, and have releases for those fixes17:11
ttxcreiht: I understand that, we'll have to corner him so that he explains his point to you17:11
doudeHi all, I don't understand how the nova-network defines the block subnet to NAT in the VlanManager mode ? Is it defined by the flag 'fixed_range' ?17:11
creihtttx: I don't care what his point is, we have software that we need to support :)17:12
ttxcreiht: I think it was mostly about letting distributors package up stable release updates17:12
* creiht sighs17:13
*** sebastianstadil has quit IRC17:13
ttxcreiht: but I'd love to discuss that in more detail with both of you ;)17:14
creihtttx: we need to formalize this stuff17:14
*** aliguori has joined #openstack17:14
*** zul has joined #openstack17:15
ttxcreiht: first we need a common view on what we should or should not do. Having a stable branch where we push stable release fixes is different from formally releasing a .117:15
creihtSo when we adopt a new project into openstack, are we going to force them into the same release schedule that we do?17:15
uvirtbotNew bug: #690740 in swift "change stats default server name to proxy-server" [Low,Confirmed]
ttxcreiht: inside openstack-core, yes. In the "compatible" projects, no.17:17
creihtttx: The problem is we release 1.1 (or austin, or whatever you want to call it), people install it on a production cluster, and it may be non-trivial to upgrade 1.1-1.2 to get a bug fix, and even more so when 1.2 isn't released uet17:17
ttxcreiht: that's what johnpur and the POC are working on17:17
jaypipesxtoddx: should be able to branch lp:~jaypipes/openstack-common/options now :)17:18
*** rds__ has quit IRC17:18
creihtso we have a bug in austin, people are running austin in production, and we say sorry, you will have to wait for bexar for a fix17:18
ttxcreiht: I understand that, but is it our work to do those tarballs ? Or should we rely on distributions to package bugfix branches ?17:18
ttxcreiht: I have no strong opinion on that, Rick and you seem to have very strong diverging opinions on that17:19
ttxcreiht: we should setup a 3-way call and see if we can come up with something sane17:19
ttxcreiht: gtg now, sorry17:19
* creiht sighs17:20
*** romohr has quit IRC17:21
notmynamecolinnich: just proposed a merge for the 1.1 branch for that bug. should be taken care of this afternoon17:23
colinnichnotmyname: thanks. But from the argument that's just gone on, it may not see a release :-)17:26
colinnichnotmyname: was an easy enough code change for me to make anyway17:26
colinnichnotmyname: but I'd agree with creiht that I shouldn't have to manually apply patches for stuff that was fixed ages ago17:27
colinnichnotmyname: got to go now, home time for me. Thanks again for the help17:29
*** lvaughn has quit IRC17:33
*** Ryan_Lane has joined #openstack17:33
*** lvaughn has joined #openstack17:34
*** colinnich has quit IRC17:35
*** jdarcy has joined #openstack17:37
*** reldan has quit IRC17:41
*** brd_from_italy has joined #openstack17:44
*** guigui1 has quit IRC17:52
*** schisamo has joined #openstack17:58
*** aliguori has quit IRC18:02
*** jdurgin has joined #openstack18:03
*** aliguori has joined #openstack18:04
*** maplebed has joined #openstack18:05
*** lvaughn_ has joined #openstack18:08
xtoddxjaypipes, ttx: thanks for repo fix18:08
jaypipesxtoddx:  no worries, buddy.  how's the lovely weather in KY btw?  it's a balmy 19 degrees F here...18:09
*** lvaughn has quit IRC18:11
vishyopinions on testing components that are dependent on time?18:12
termievishy: stub out the clock and move it yourself18:12
xtoddxjaypipes: it was 9 this morning when i woke up18:12
jaypipesvishy: agreed with termie.18:12
vishytermie: oh hai! yeah that is basically what i did18:12
termievishy: was why i was trying to get people to use utils.utcnow()18:12
termievishy: to get the time18:12
termievishy: so that tests can work it18:13
xtoddxjaypipes: warmer now, but lots of snow still on the ground, did you guys get snow over the weekend?  We missed  a day and a half of school.18:13
jaypipesxtoddx: ya, pretty friggin cold round our parts :)18:13
vishyI suppose i should propose that, not sure how valuable it is going to be to anyone else though18:13
jaypipesxtoddx: and about 4 inches of snow... really, it's been almost too cold to snow...18:13
termievishy: don't we already have a fake memcache?18:14
vishytermie: not that I am aware of18:14
termiei definitely remember writing one for the project at some point18:14
vishyfake rabbit, we had keeper for a while18:14
termieit must have been keeper stuff18:14
termieso, i wouldn't use the time function thing18:15
vishythe problem is testing against real memcached doesn't let me manipulate time18:15
termiei would instead have it import a common time function18:15
termieand stub that out18:15
*** reldan has joined #openstack18:15
Ryan_Lanevishy: had you had a chance to look at my comment on: ?18:15
termierather than handing the time to the client init18:15
*** al-maisan is now known as almaisan-away18:16
termieand provide a general purpose fake clock18:16
termiethat you can wind and rewind18:16
termiei have an example somewhere18:16
vishytermie: that makes sense18:17
vishymy first pass was to monkeypatch time.time in setUp and tearDown but that is pretty fragile18:18
termiethat is kind of hard to read18:18
termiebut basically i just pass in the time delta args18:18
termieactually i don't really like that implementation18:19
vishyi see where you are going though18:19
termieit worked fine for the tests but i like ones better that let you manipulate a single time18:19
termiethis one tends to creep forward18:19
termiein practice i just make three at a time18:19
*** jimbaker has quit IRC18:19
*** joearnold has joined #openstack18:19
*** abecc has joined #openstack18:19
termiefor the different times i will set it to and test18:20
vishyI could make a simple class in utils that essentially does what I'm doing in the test with the replacement function18:20
* vishy hacks it up18:20
termieit has a fake memcache implementation also, ftr, it looks a bit different than yours18:21
vishymine is very minimal for sure :)18:21
termieall fake memcache implementatiosn tend towards the real memcache18:21
termiein my experience ;)18:21
termiethey will all eventually ahve all the features, and send email18:22
termieshould not have eaten two burritos last night18:22
termiei am groggy mcgroggerson18:22
*** reldan_ has joined #openstack18:23
vishyRyan_Lane: sorry I read that in email and was waiting to be online so I could approve it then promptly forgot.18:23
*** reldan has quit IRC18:23
*** reldan_ is now known as reldan18:23
Ryan_Lanevishy: I should likely add the key schema back in18:23
Ryan_Laneat least the openssh-lpk one18:24
*** westmaas has quit IRC18:25
*** DigitalFlux has quit IRC18:25
termieuh oh, did somebody start the gettext stuff already?18:27
vishytermie: so If Timer is a class that is instantiated in fakememcache, how can the test override the time for that particular timer18:28
termieit isn't instantiated in fakememcache18:28
termiefakememcache ujust says "utils.utcnow()"18:28
termieto get the current time18:28
vishyI don't see how that has a huge advantage over monkeypatching.18:28
termieTimer might be class that you use in tests to manipulate the result that utils.utcnow() will give18:28
Ryan_Laneok. added lpk schema back into that branch18:29
vishyit still suffers from side effects18:29
termievishy: it is better than monkeypatching because you are explicitly expecting the call to be monkeypatched18:29
vishyi.e. if I forget to unset the override it breaks everything18:29
termierather than with "time.time"18:29
termiethat's what tests are fore18:29
termieand tearDown18:29
vishyThe dependency injection has the advantage of not having those side effects18:30
vishybut it is a bit ugly i agree18:30
termieand it doesn't affect the whole code base18:30
termieso you'll ahve side effects when two thigns are looking at clocks18:30
termieand getting different times18:30
termie(using dependency injection)18:31
vishygood point.  I guess it is a tradeoff18:31
termiedoesn't seem like much of a tradeoff to me, dependency injection doesn't work outside of large java frameworks18:31
termieseriously though, it doesn't, nobody else is going to make sure that the dependency is actually injected everywhere18:32
vishyso I could just monkeypatch utils.utcnow18:32
termiethat is exactly what i am suggesting18:32
termiebut make yourself a utility clock18:32
termieso you can manipulate the output18:32
vishyyeah, I have to rewrite the timeout to calculate timediffs instead of epochs though18:33
vishyepoch/unix timestamps18:33
termiei don't see why18:33
termieyou can make the clock give teh right time18:33
vishyutcnow returns a datetime18:33
termiewas my solution to part of that18:34
termie(so you would just call the utcnow_ts equiv)18:35
*** vvuksan has quit IRC18:36
*** vvuksan1 has joined #openstack18:36
*** DigitalFlux has joined #openstack18:39
*** jc_smith has joined #openstack18:39
*** ibarrera has quit IRC18:42
*** krish has joined #openstack18:44
DigitalFluxAny body working on the OpenStack puppet stuff here ?18:52
vishytermie: thoughts?
termiei really need to cut down on my underhanded bzr jabbing18:53
termieall my statements read like hate letters18:53
termievishy: non traditional but i like it18:54
vishyfunction attributes are fun18:54
sandywalshtermie, thanks for feedback on comments ... trying new branch now18:54
termiesandywalsh: thanks for testing xen stuff18:54
hazmatvishy, still globals though18:54
termiehazmat: needs to be global18:55
vishynamespaced global!18:55
termiehazmat: to override everything at the same time18:55
*** doude has quit IRC18:55
vishyall of the benefits with none of the drawbacks!18:55
hazmattermie, is it a startup/customization one time thing?18:56
termiei do tend to like the pattern of providing a handle when overriding18:56
termiethat allows you to clear the override18:57
termiehazmat: it is in tests so it will happen a bunch of times, but you want the tiem to be consistent across all parts of the system18:57
termiehazmat: as long as things call utcnow to get the time18:57
termiehazmat: then it doesn't matter when you change it18:57
hazmattermie, my concern with the global,  its a mutable global while there are ongoing ops that might use it when its being changed. if its primarily changed for tests and startup, then its not an issue, but else some sort of lock around usage would be appropriate.18:59
termiehazmat: it is _only_ changed for tests18:59
hazmatperhaps it should go into the tests package then, instead of top level utils.19:00
edayvishy: just curious, why do you need a fake clock, what's wrong with a real clock? Are you trying to introduce skew? shorten test timed?19:00
hazmati guess everything imports it from utils to get it19:00
vishyeday: for testing timeout19:00
vishyeday: unittests, don't want to sleep to wait for timeout to hit19:01
edayvishy: setting timeout to 1 sec isn't shot enough? I can see wanting tests to never actually sleep, but seems like a bit of extra complexity19:03
termieeday: i think it is worth the complexity, it will be needed more in teh future19:03
termieeday: sleeping in tests is just asking for a flaky test19:04
termieis there an adam waters in here?19:05
edaytesting in an unrealistic harness may also be a flaky test (in that it always passes because you give predicatable input) :)19:05
termieeday: not a flaky test in that case, just not a thorough test19:06
termieeday: flaky means it flaps from time to time19:06
edaysure, flaky/thorough/..., whatever you want to call it :)19:06
termieeday: flaky means it flaps from time to time19:06
edayyes, I know what flaky means :)19:07
termieeday: woops, that was a repeat but i guess it got the point across :p19:07
termiea test that passes but doesn't test something real is an altogether different beast19:07
edayI meant "not thorough" yes, but this is an issue with unittests in general19:08
edaybut it is possible to create non-flaky unittests based on real time too, you just need to make the window large enough (within reason) for what's passable19:08
*** dfg_ has joined #openstack19:09
termiefaking a clock for testing is done in almost every large system I am aware of, and where it isn't it should be19:09
termiei agree that there are a variety of ways to not test things that matter, but using real time isn't going to magically solve that either19:10
edayI've seen it a fair amount as well, but I've also seen many test systems using real time that worked quite well too. Not sure there is a correct way here, whatever works :)19:11
vishyi could set the flag to one second, but i do like being able to test reasonable timeouts19:11
vishyand i don't have a whole freaking second to spare!19:12
edayI was only questioning to make sure we relaly needed it, I'm not opposed to it19:12
edayif we just accepted everything at face value, we'd be in the world we are in today19:12
* eday thinks the sinus congestion may be getting to his head19:14
vishytermie: time.mktime(<datetime>) no worky19:14
* vishy looks for the right conversion19:14
*** kevnfx has quit IRC19:15
vishyah timetuple19:15
*** jimbaker has joined #openstack19:16
*** HugoKuo has quit IRC19:17
*** melrefaey has joined #openstack19:19
*** reldan has quit IRC19:19
*** k0p has joined #openstack19:25
sandywalshtermie, can't seem to get past bundle-image ... stuck in 'decrypting'19:28
sandywalshtermie, no clues in the logs19:28
termiesandywalsh: what does one need to do to get xen running19:29
termieor rather... to the channel as a whole19:29
termiewhat do i need to set up locally to test xen?19:29
sandywalshwe have a great step-by-step guide ... it's on the rackspace wiki. Let me see about getting on the openstack wiki19:29
sandywalshjk0, any thoughts on getting your install wiki page moved to openstack wiki?19:30
termiesandywalsh: there is a good chance that there is a blocking call we have to wrap, I can suggest the change if you want to check it19:30
sandywalshtermie, sure thing19:30
jk0sandywalsh: I can do that19:30
jk0I'll try to have it done by today19:31
termiewow, upgrading to maverick did something really crazy to my fonts in gvim19:31
*** joearnold has quit IRC19:31
sorentermie: good crazy or bad crazy?19:34
termiesoren: hehe, bad crazy but it has been resolved, i think they got rid of a font i was using or i somehow had one installed in a weird way19:35
sandywalshtermie, I would imagine this bundle-image issue would arise on other hypervisors too, no?19:35
termiei just switched back to Monospace19:35
termiesandywalsh: not sure, but there was definitely a place in the patch we were unsure of how xen would respond to19:36
termiesandywalsh: so i am trying out the alternate solution19:36
sorenvishy: I asked you a couple of questions yesterday morning.. I think. Maybe the day before.19:38
termiesandywalsh: this puts xen calls into threads19:39
sorenvishy: I don't think I saw a reply. It was about the necessity/rationale for the network concept.19:39
sandywalshtermie, gotcha ... trying19:39
sorenvishy: Did you see them or should I repost them?19:39
termiesandywalsh: did you say there was some info on a rackspace wiki for setting up xen?19:40
termiethis would probably go much more smoothly if i could test it locally :p19:40
sandywalshtermie, jk0 is looking into moving it onto the openstack wiki today19:41
termiesandywalsh: ah, cool19:41
vishysoren: don't think i saw them19:41
vishyeaten by scrollback!19:41
soren10:07 <+soren> vishy: I'm (again) staring at the network code. I'm having trouble wrapping my head around the network concept. Why is it that we need to chop it up into smaller IP subnets?19:42
soren10:07 <+soren> vishy: It it to be able to have simpler firewall rule sets?19:42
*** dendro-afk is now known as dendrobates19:42
vishythe subnets are for vlan mode19:42
vishywhere each project gets its own subnet/vlan19:42
vishyeach subnet has its own dhcpserver19:43
sorenOk, the VLAN thing I understand.19:43
sorenI'm just not sure why they need their own IP subnet.19:43
vishythe alternative would be one dhcpserver for everyone i suppose19:44
vishyso there are two things that the subnet simplifies19:44
vishy1) vpn is always at the same ip in each subnet19:44
vishy2) cross region firewall rules19:45
vishyi can't think of anything else that would be required19:45
sorenWe could have separate dhcp servers even though people aren't on the same subnet.19:45
vishybut I'm probably forgetting something19:45
sorenOur DHCP servers are hand out statically assigned IP's. It's not like they're picking from a pool anyways.19:46
vishysoren: dnsmasq has a start range that you specify for where it listens19:46
vishysoren: have to go to lunch.  I will consider and see if there is anything else19:46
sorenvishy: ok.19:46
sorenvishy: If it's not obvious, my motivation is to avoid allocating a bunch of IP's to a project that only needs a fraction while not restricting others to a too small set.19:47
sandywalshtermie, seems to be missing import again?19:47
termiesandywalsh: should be LoopingCall not LoopingTask19:48
melrefaeyHello guys! this is Mohamed ... and i am here to discuss the Nova diagram I've created, as I knew you have some feedback on it ..19:48
sandywalshtermie, k, fixing19:48
sorenmelrefaey: Yes! Can you post the link to it?19:49
melrefaeywhich link, i don't think it is on the doc site yet ...19:50
sorenmelrefaey: Oh, found it.19:50
sandywalshtermie, so far so good ...19:50
termietentative woo19:50
melrefaeygood ...19:50
sorenmelrefaey: Ok, so one thing that's inaccurate is that it looks like only the auth manager speaks to the DB.19:51
sorenmelrefaey: Every single component, except the object store, speaks to the DB.19:51
sorenWell, it depends :)19:52
sandywalshtermie, feel free to woot away ... we're golden!19:52
sorenThat's another thing. Using LDAP is optional.19:52
sorenYou can use the DB as your user database.19:52
termiesandywalsh: hot damn :) i'll make the fix in the code and push19:52
melrefaeyyeah, you are right ... but it was dificult to show that ... but i will think of a way to illustrate that link ..19:52
sorenThe object store speaks to the user database, which may be in LDAP or in the DB.19:53
sandywalshtermie, excellent ... thanks for the help on that!19:53
melrefaeyI see19:53
sorenmelrefaey: Also, it looks like euca2ools are somehow different from other clients (since it's inside the big square), but really it's exactly like the "other clients" in the box on the left.19:53
*** rds__ has joined #openstack19:54
*** nelson__ has left #openstack19:54
melrefaeyPerfect then! ... will change it19:55
melrefaeyrearding the network worker ... should i show the ATAoE and such stuff ?19:56
sorenmelrefaey: The network worker has nothing to do with ATAoE.19:57
*** DigitalFlux has quit IRC19:57
sorenmelrefaey: That's only used between the compute and volume controllers.19:57
melrefaeysorr, i mean the volume controller19:57
sandywalshtermie / soren so, I'm going to want to merge the xs-pause branch with eventlet branch. I imagine I'll want to merge xs-pause into eventlet (and not the other way around), correct?19:57
sorensandywalsh: No, the other way aroudnd.19:58
sorensandywalsh: Really, it doesn't matter. The result is the same.19:58
sorensandywalsh: But..19:58
melrefaeysoren: yes I meant the volume controller ..19:58
sorensandywalsh: Since your stuff if xs-pause, that's what you'd usually pull other stuff into.19:58
sorenmelrefaey: Right, ok.19:59
sorenmelrefaey: Yeah, there should be a link between the two, but it can also use iSCSI.19:59
sandywalshsoren, right ... gotcha19:59
sorenmelrefaey: I'd rather you put that, since ATAoE is not a very good idea to begin with.19:59
sorenmelrefaey: Another thing:19:59
melrefaeyYes, ATAoE/iscsi20:00
sorenmelrefaey: The "physical host" box suggests that we only support Xen.20:00
melrefaeydefinitely no, i just put xen as VMM example ...20:00
melrefaeybut if it wouldn't be confusing I will add the other supported hypervisors ..20:01
*** BK_man has joined #openstack20:01
sorenmelrefaey: Maybe just put "VMM (KVM/Xen/UML)" or something?20:02
sorenmelrefaey: What are you using to make the diagram?20:02
melrefaeyyes, i can do that suggestion as well20:03
melrefaeyVMM(KVM, Xen, UML)20:03
*** odyi has quit IRC20:04
melrefaeyregarding the Queueing engine .. do you think it is good idea to put the rabbitmq instead ... ?20:04
openstackhudsonProject nova build #304: SUCCESS in 1 min 16 sec:
melrefaeyor leave it abstract20:04
*** odyi has joined #openstack20:05
sorenmelrefaey: I'd put Rabbit.20:05
melrefaeyI see ... the same for LDAP? you mean i put OpenLDAP?20:05
*** reldan has joined #openstack20:06
melrefaeySoren: okay, I have another idea, we can leave this digram with an abstract componenet like it is now ... and I can do another one with the actuall used technologies and frameworks, what do you think ?20:06
sorenmelrefaey: I think people actually are using different LDAP servers.20:07
sorenmelrefaey: ...but I don't think anyone is using anything but rabbit.20:07
melrefaeyhmm, okay ...that make sense to me20:07
*** odyi has quit IRC20:07
*** odyi has joined #openstack20:07
*** zul has quit IRC20:09
fabiandsoren, what about apache's amqp implementation?20:11
sorenfabiand: What about it?20:11
fabiandsoren, you said raqbbit is in use, i wonder if you know some environments using apaeche's thingy ..20:13
sorenI don't. That's what I'm saying :)20:13
soren20:07 <+soren> melrefaey: ...but I don't think anyone is using anything but rabbit.20:13
fabiandmh kay. cheers :)20:13
melrefaeyanything else? do you have any other points to include?20:16
*** dendrobates is now known as dendro-afk20:18
*** hadrian has quit IRC20:19
*** dendro-afk is now known as dendrobates20:19
sorenmelrefaey: I think that's about it.20:22
fabiandI tried to find some api documentation, but the appropriate docs on are empty ...20:23
fabiandis the api documented somewhere else?20:23
melrefaeyokay, I will consolidate the points, work on them and will back to you ... Thank you20:23
fabiandah okay, just some are empty ..20:24
*** mdomsch_ has joined #openstack20:28
*** pothos has quit IRC20:29
*** daleolds has joined #openstack20:30
*** matclayton has left #openstack20:31
*** pothos has joined #openstack20:31
*** jasonb365 has joined #openstack20:32
dendrobatesdid you guys see the traceback from hudson after the last test run with reldan's 2 line patch?  It seems we have a problem unrelated to the patch20:33
*** BK_man has quit IRC20:33
vishysoren: upon further consideration, I can't think of anything else that needs a subnet.  The vpn thing isn't a biggie.  The cross region security groups will be a problem though.20:34
sorenCan you explain that?20:34
vishysoren: I don't know if you can have multiple dnsmasqs responding to leases for the same overall network, but you could break them up20:34
sorenvishy: They'd never see the same requests, so I don't think it's a problem.20:35
vishysoren: so our naive multiregion implementation is two separate installs20:35
vishysoren: how do you stop them from getting packets?20:35
*** westmaas has joined #openstack20:37
reldandendrobates: I try to fix it. install_venv installs TWISTED_NOVA=''. Is it just Twisted-10.0.0 library or something else?20:37
dubsquaredHey guys, is there a page that lists all the current available  network managers for nova?20:38
*** hadrian has joined #openstack20:38
dubsquaredi felt like i saw this somewhere…but cant find it in my bookmarks20:38
vishysoren: so if project a is in a given subnet in region A, it is possible to set up a rule in region B to allow access from the whole project.20:39
sorenvishy: You've lost me.20:39
vishysoren: that packet question was in regards to dnsmasq20:39
vishysoren: the first and 3rr: so our... and so if... were both in response to your question20:39
sorenvishy: Oh!20:40
sorenvishy: Sorry, I'm in the middle of three(!) other conversations, too, so I get confused easily.20:40
dendrobatesreldan: the test error was not caused by your patch.  It was this test :    test_public_network_association ...                                 [ERROR]20:40
sorenvishy: The dnsmasq instances could just listen on a vlan tagged interface.20:40
vishydendro: linked to logs?20:41
vishysoren: ah good call, the obvious solution didn't occur to me20:41
*** melrefaey has left #openstack20:41
reldandendrobates: Yes, seems so. It passed all test in my environment.20:42
vishydendrobates: that is can you post a link to the failed test output so I can see it?20:42
*** AimanA has joined #openstack20:42
vishyhmm it is running the ldap tests too it seems20:44
sorenIf python-ldap is available, it'll run them.20:44
vishyis ldap cleaned between runs?20:44
soren..but with fakeldap as the backend.20:44
vishyah right20:45
vishybacking to redis20:45
dendrobatesit looks like the environment was not cleaned properly20:45
*** _skrusty has quit IRC20:45
sorenvishy: Perhaps it ran at the same time as another test run.20:46
sorenvishy: This is the coverage test run. tarmac might have been running at the same time. Using the same redis instance.20:46
dendrobatesI'll try it again, and see if it fails again20:46
sorendendrobates: You have hudson power?20:46
*** irahgel1 has left #openstack20:46
vishyhmm that could be20:47
dendrobatessoren: no, but I can set it to approved and let it magically happen20:47
sorenIt merged just fine, didn't it?20:47
dendrobateshudson is pure magic to me, and I intend to keep it that way.20:47
vishyif netuser is leaking from a previous test it should have shown up previously20:47
sorenthis is not the tests from tarmac failing.20:47
mtaylordendrobates: ++20:47
mtaylordendrobates: making some progress on better hudson, btw20:48
sorendendrobates: I'll rerun it.20:48
dendrobatessoren: show off20:48
sorenYou know it.20:48
soren <--- If you want to follow the progress20:48
sorenLooks good so far.20:48
sorenvishy: Ok, so multi-region stuff..20:49
sorenvishy: I don't understand how that even applies. You're only passing out private IP's anyway?20:50
vishysoren: instances in region a can talk to instances in region b20:52
dendrobatescan another core dev review:
vishysoren: so we route 10.x... from region a to 10.y... in region b20:52
sorenvishy: I see. Hmm. Ok.20:52
vishyand then it is blocked at a host level if we don't have a security group rule20:53
sorenvishy: Ok.20:53
dendrobatesreldan: it's merged.20:53
reldandendrobates: Great! Thank you!20:53
vishysoren: if we can come up with a reasonable solution for that issue, the advantage of having large and small projects like you are thinking would be very useful20:54
*** mdomsch_ has quit IRC20:54
openstackhudsonProject nova build #305: SUCCESS in 1 min 12 sec:
sorenvishy: Can you elaborate on how you do it now?20:54
sorenvishy: You add a security group rule based on your knowledge of which network is assigned to you?20:55
*** rlucio has joined #openstack20:55
sorenvishy: How do you know which subnet is assigned to you? Derived from your IP and a-priori knowledge of size of networks?20:55
*** _skrusty has joined #openstack20:58
vishysoren: magic...
dendrobatesannegentle: around?21:00
*** cloudmeat is now known as bobl21:00
sorenvishy: ah.21:00
vishysoren: quick and dirty21:01
sorenvishy: So you do it from Nova. Gotcha.21:01
vishysoren: we were trying to do it externally to the system, so that the regions didn't really have to know about each other21:01
vishyso it is an external worker that runs21:02
vishyand hacks the dbs directly21:02
vishyand sends messages into the queue21:02
*** kevnfx has joined #openstack21:02
dendrobatesI heard back from him about it.21:04
sorenWrong channel :)21:04
vvuksan1dndrobates and soren: who is in charge of Centos nova documentation :-) ?21:05
sorenvvuksan1: Now that you asked, you are.21:05
vvuksan1there are a number of missing / incorrect references in it21:05
vvuksan1aha :-)21:05
vvuksan1i'm glad I volunteered21:05
sorenTag. You're it.21:05
sorenIsn't free software great?21:05
vvuksan1it's telling me it's an immutable page21:07
jbrycedendrobates: annegentle is out this afternoon at the team thing in austin. she'll also be offline some tomorrow for family stuff.21:08
dendrobatesjbryce: ah thanks.21:08
jbrycevvuksan1: are you logged in as a user in the wiki?21:10
*** rlucio has quit IRC21:10
*** rds__ has quit IRC21:10
vvuksan1i thought i was21:10
vvuksan1that was odd21:10
*** jdarcy has quit IRC21:14
sorenvishy: Ok, I can't come up with a good solution for that.21:16
sorenvishy: But, erm... It seems neither could you :)21:16
* soren ducks21:16
vishysoren: agreed21:16
vishysoren: the real solution is not have independent regions that need to be able to communicate21:17
sorenI'll probably try to work on this for Cactus. I have a rather limited set of public IP's and having to carve it up like this is really annoying.21:17
vishysoren: why are you using publics?21:17
sorenvishy: I'm not sure I agree.21:17
sorenvishy: Because that's how I like it :)21:18
vishysoren: some love needs to be given to flatdhcp21:18
sorenvishy: It's how Rackspace does it, for instance.21:18
vishysoren: but they don't need project vlans right?21:18
*** reldan has quit IRC21:18
sorenvishy: I don't know.21:19
vishysoren: it is hard for me to hack on flatdhcp because it isn't really our usecase, but it is close to being very useful for that style21:19
sorenvishy: They do use VLAN's now, so I guess they do.21:19
vishypublic ips and such21:19
*** lvaughn_ has quit IRC21:19
vishysoren: the only issue with it currently is that it need two interfaces21:19
*** lvaughn has joined #openstack21:19
vishysoren: to work on multiple machines21:19
sorenYes, that's rather annoying.21:20
vishysoren: I think I've figured out a way to make it work on one, but it means nova will have to get a little smarter21:20
sorenI'd love to hear your ideas. I've been thinking about the same thing.21:20
dendrobatesah, true21:23
dendrobateswrong channel again21:23
*** BK_man has joined #openstack21:24
vishysoren: I think that it could copy the ip from eth0 on to the bridge21:28
vishysoren: then remove the ip from eth0 and attach, then add the dnsmasq/gateway ip as well21:29
vishysoren: that way you have public internet and private on the same device.  The tricky part is doing that without dropping the user's ssh connection or otherwise destroying networking on the box21:29
vishysoren: for the compute nodes it is a bit easier because the bridge doesn't need a second ip for dnsmasq/gateway21:30
vishysoren: I think there is an alternative using a tun/tap, that may be a little more user friendly, but I don't know enough about how they work and I haven't had time to research it21:31
sorenvishy: Let me know if you decide to start working on it. I may be able to fill in some gaps.21:32
sorenFor now, though, I'll call it a day.21:32
soren..week, month, year, and decade, too!21:32
vishysoren: word21:32
*** rlucio has joined #openstack21:33
vishysoren: I'm hoping someone who actually uses it can take it over21:33
sorenvishy: I wouldn't mind adopting it.21:33
* soren wanders off21:33
*** jasonb365 has quit IRC21:34
*** westmaas has quit IRC21:35
*** miclorb_ has joined #openstack21:35
*** brd_from_italy has quit IRC21:36
*** BK_man has quit IRC21:42
*** GasbaKid has joined #openstack21:43
*** sophiap has quit IRC21:44
*** jasonb365 has joined #openstack21:46
*** burris has quit IRC21:47
*** mdomsch_ has joined #openstack21:56
*** allsystemsarego has quit IRC21:58
*** ctennis has quit IRC22:06
*** reldan has joined #openstack22:12
comstudsoren/vishy, we don't need to use vlan tagging at the host, if that's what you were talking about22:20
*** ctennis has joined #openstack22:20
*** ctennis has joined #openstack22:20
zykes-vishy: that uses what ?22:20
*** BK_man has joined #openstack22:20
comstuds/don't need to use/don't use/22:21
*** Podilarius has left #openstack22:21
zykes-22:33 <+soren> vishy: I wouldn't mind adopting it.22:26
*** kevnfx has quit IRC22:27
*** reldan has quit IRC22:30
*** KyleM1 has joined #openstack22:35
KyleM1Hey, when's the freeze date for Bexar?22:35
*** dendrobates is now known as dendro-afk22:36
*** apanasenko has quit IRC22:36
*** rlucio_ has joined #openstack22:37
*** termie has quit IRC22:38
*** rlucio has quit IRC22:41
*** rlucio_ is now known as rlucio22:41
*** seshu has left #openstack22:43
*** vvuksan1 has quit IRC22:44
*** ppetraki has quit IRC22:44
dabosandywalsh: did you get the eventlet branch working? I'm having issues with objectstore (twisted errors)22:45
*** sebastianstadil has joined #openstack22:46
jk0I put a copy of our XenSever guide up on the wiki:
*** dirakx has quit IRC22:49
*** aliguori has quit IRC22:49
*** daleolds has quit IRC22:52
*** daleolds has joined #openstack22:53
*** jimbaker has quit IRC22:53
*** daleolds has quit IRC22:59
*** daleolds has joined #openstack22:59
*** BK_man has quit IRC23:00
*** daleolds has joined #openstack23:00
*** daleolds has joined #openstack23:02
*** daleolds has joined #openstack23:02
*** reldan has joined #openstack23:03
*** BK_man has joined #openstack23:05
*** termie has joined #openstack23:10
*** termie has quit IRC23:10
*** termie has joined #openstack23:10
*** daleolds has quit IRC23:10
*** hazmat has quit IRC23:11
*** jimbaker has joined #openstack23:12
*** daleolds has joined #openstack23:12
*** GasbaKid has quit IRC23:14
*** daleolds has quit IRC23:15
*** daleolds has joined #openstack23:17
*** mdomsch_ has quit IRC23:30
*** reldan has quit IRC23:43
*** arthurc has quit IRC23:47
*** Ryan_Lane has quit IRC23:51
*** gondoi has quit IRC23:55
*** schisamo has quit IRC23:57

Generated by 2.14.0 by Marius Gedminas - find it at!