| *** bcwaldon has joined #openstack | 00:00 | |
| uvirtbot | New bug: #731684 in glance "Glance lazy loading issue -- still coupled with Swift" [High,New] https://launchpad.net/bugs/731684 | 00:01 |
|---|---|---|
| *** enigma1 has joined #openstack | 00:03 | |
| *** gregp76 has quit IRC | 00:05 | |
| *** enigma1 has quit IRC | 00:06 | |
| *** pvo has quit IRC | 00:06 | |
| *** enigma has quit IRC | 00:07 | |
| *** jaypipes is now known as jaypipes-afk | 00:10 | |
| *** troytoma` is now known as troytoman-away | 00:13 | |
| *** enigma has joined #openstack | 00:14 | |
| *** msassak has quit IRC | 00:14 | |
| *** hggdh has quit IRC | 00:21 | |
| *** hggdh has joined #openstack | 00:22 | |
| *** hggdh has joined #openstack | 00:22 | |
| *** ddumitriu has quit IRC | 00:26 | |
| *** pandemicsyn has quit IRC | 00:27 | |
| *** syn_ has joined #openstack | 00:28 | |
| *** ChanServ sets mode: +v syn_ | 00:29 | |
| *** syn_ is now known as pandemicsyn | 00:29 | |
| *** blakeyeager has quit IRC | 00:31 | |
| *** winston-d has joined #openstack | 00:38 | |
| *** bcwaldon has quit IRC | 00:41 | |
| *** bcwaldon has joined #openstack | 00:43 | |
| *** ccustine has quit IRC | 00:46 | |
| *** bcwaldon_ has joined #openstack | 00:49 | |
| *** bcwaldon has quit IRC | 00:50 | |
| *** Vek has joined #openstack | 00:50 | |
| *** ericrw has quit IRC | 00:52 | |
| *** MarkAtwood has quit IRC | 00:57 | |
| *** rnirmal has quit IRC | 00:57 | |
| *** mahadev has quit IRC | 01:01 | |
| *** armmer has joined #openstack | 01:01 | |
| *** mahadev has joined #openstack | 01:07 | |
| *** ivan has quit IRC | 01:09 | |
| *** littleidea has joined #openstack | 01:09 | |
| *** ivan has joined #openstack | 01:12 | |
| *** paltman has joined #openstack | 01:15 | |
| *** paltman has quit IRC | 01:18 | |
| *** ericrw has joined #openstack | 01:28 | |
| *** bcwaldon_ has quit IRC | 01:29 | |
| *** mahadev has quit IRC | 01:30 | |
| *** bcwaldon has joined #openstack | 01:30 | |
| *** mahadev has joined #openstack | 01:30 | |
| *** dragondm has quit IRC | 01:34 | |
| *** MotoMilind has quit IRC | 01:34 | |
| *** clauden_ has joined #openstack | 01:37 | |
| *** bcwaldon has quit IRC | 01:42 | |
| *** littleidea has quit IRC | 01:52 | |
| *** azneita has quit IRC | 01:58 | |
| *** rlucio has joined #openstack | 02:03 | |
| *** rlucio has quit IRC | 02:07 | |
| *** dprince has joined #openstack | 02:15 | |
| *** magritte has quit IRC | 02:16 | |
| *** dendro-afk is now known as dendrobates | 02:17 | |
| *** azneita has joined #openstack | 02:20 | |
| winston-d | any chance gholt is still here? | 02:21 |
| winston-d | gholt: hi | 02:21 |
| *** paltman has joined #openstack | 02:24 | |
| *** paltman has quit IRC | 02:27 | |
| *** hadrian has quit IRC | 02:32 | |
| *** ericrw has quit IRC | 02:38 | |
| *** dprince has quit IRC | 02:48 | |
| *** ericrw has joined #openstack | 02:53 | |
| *** mahadev has quit IRC | 02:55 | |
| *** mahadev has joined #openstack | 02:57 | |
| *** EndEng|Alt has quit IRC | 02:57 | |
| *** mahadev has quit IRC | 02:58 | |
| *** pvo has joined #openstack | 02:59 | |
| *** EndEng|Alt has joined #openstack | 03:11 | |
| *** jero has quit IRC | 03:12 | |
| *** azneita has quit IRC | 03:18 | |
| uvirtbot | New bug: #731750 in nova "nova compute race condition on truncate" [Undecided,New] https://launchpad.net/bugs/731750 | 03:22 |
| *** miclorb_ has quit IRC | 03:27 | |
| *** paltman has joined #openstack | 03:39 | |
| *** miclorb has joined #openstack | 03:39 | |
| *** kashyap has joined #openstack | 03:46 | |
| *** lamar has joined #openstack | 04:02 | |
| *** miclorb has quit IRC | 04:16 | |
| *** miclorb has joined #openstack | 04:28 | |
| openstackhudson | Project nova build #608: SUCCESS in 1 min 55 sec: http://hudson.openstack.org/job/nova/608/ | 04:38 |
| openstackhudson | Tarmac: Refactoring nova-api to be a service, so that we can reuse it in unit tests | 04:38 |
| HugoKuo | morning | 05:08 |
| HugoKuo | while I use flat manage mode , instance would not get the namesever address right ? | 05:09 |
| HugoKuo | the only way is turn to use FlatDHCPmanage mode. Am I right ? | 05:09 |
| *** enigma has quit IRC | 05:18 | |
| *** dirakx1 has quit IRC | 05:24 | |
| *** EndEng|Alt has quit IRC | 05:28 | |
| *** fayce has joined #openstack | 05:36 | |
| *** fayce has quit IRC | 05:38 | |
| *** f4m8_ is now known as f4m8 | 05:42 | |
| *** pvo has quit IRC | 05:59 | |
| *** dirakx has joined #openstack | 06:02 | |
| *** naehring has joined #openstack | 06:13 | |
| *** fayce has joined #openstack | 06:15 | |
| *** ramkrsna has joined #openstack | 06:19 | |
| *** fayce has quit IRC | 06:27 | |
| *** mahadev has joined #openstack | 06:31 | |
| *** mgoldmann has joined #openstack | 06:33 | |
| *** mahadev has quit IRC | 06:37 | |
| *** ericrw has quit IRC | 06:44 | |
| *** gasbakid has joined #openstack | 07:04 | |
| *** nerens has joined #openstack | 07:07 | |
| *** CloudChris has joined #openstack | 07:17 | |
| *** shoaib has joined #openstack | 07:20 | |
| shoaib | hi | 07:20 |
| shoaib | i wanted to ask if there is a way i can modify the disk size for m1.tiny instance? | 07:21 |
| *** ramkrsna has quit IRC | 07:29 | |
| *** fayce has joined #openstack | 07:34 | |
| shoaib | anyone? | 07:34 |
| naehring | shoaib, do you mean the size of your machine image? | 07:35 |
| shoaib | is there a way i can modify the disk size for m1.tiny instance? | 07:35 |
| shoaib | when we spawn an instance | 07:35 |
| btorch | fayce: did u get things working ? | 07:35 |
| shoaib | its give a disk size | 07:35 |
| naehring | shoaib, I think you mean the disk size inside the spawned instance? | 07:37 |
| *** zenmatt has quit IRC | 07:37 | |
| shoaib | yes | 07:38 |
| naehring | this depends on the image you've created and you are using. We're talking about instance types not about persistent volumes, right? | 07:39 |
| shoaib | yes | 07:39 |
| shoaib | lets say if i spawn an instance of type tiny what disk space will i get 1GB? | 07:40 |
| shoaib | doesnt the disk space change by the instance type? | 07:40 |
| fayce | btorch: almost ! I could ssh now into the instance ... using FlatManager, and an other UEC image... the get-console-output shows that there is still this "can't get metadata from 169.254.169.254" error message... that doesn't seem to affect the instance that much | 07:41 |
| *** matiu has quit IRC | 07:41 | |
| fayce | the only problem is the wrong DNS configuration (which prevents the instance from reaching internet) - we solved this problem by using FlatDHCPManager... but now we are facing another problem: we can't access the instance from a different machine than the one that hosts the VM | 07:43 |
| *** ramkrsna has joined #openstack | 07:43 | |
| *** guigui1 has joined #openstack | 07:43 | |
| shoaib | naehring -> there? | 07:47 |
| *** drico has quit IRC | 07:49 | |
| *** drico_ has joined #openstack | 07:49 | |
| *** naehring has quit IRC | 07:49 | |
| *** naehring has joined #openstack | 07:50 | |
| *** gasbakid has quit IRC | 07:51 | |
| shoaib | naehring -> need help | 07:51 |
| btorch | fayce: hmm not sure about the dns thing since I haven't care much about reaching the internet from the vms yet. | 07:53 |
| btorch | fayce: the two compute nodes are on that same localnet 172.16.2.x | 07:53 |
| naehring | shoaib, I think I know what you mean. AFAIR I saw in the debug log a "resizefs" anywhere when spawning an instance. At the moment I have no system where I can proof this but you can take a look on compute.log with --debug as a flag and then watch an instance spawning | 07:54 |
| fayce | btorch: I gave up on using 2 nodes... we are now just trying to make an all-in-1 install works correctly (with instance that can reach the net..) - later I will add more compute nodes and separate the components accross different HW.. | 07:55 |
| shoaib | ok | 07:58 |
| shoaib | let me see | 07:58 |
| btorch | fayce: I see, u still using the 172 or you using another one .. I remember you mentioned u wanted to keep them out of that corp net | 07:59 |
| *** dirakx has quit IRC | 08:00 | |
| *** paltman has quit IRC | 08:00 | |
| *** nerens has quit IRC | 08:01 | |
| fayce | btorch: yes, I actually took 2 machines and put them out of the corp net (192.168.0.0/24). everything works fine, in FlatManager mode (but we can't test internet connectivity) , and took another machine, put it in the corp net will all the components, a in FlatManager we couldn't access the net but we could access the VM from anywhere - in FlatDHCPManager, the correct DNS config is injected, the VM can access the Web but can't be access f | 08:03 |
| fayce | somewhere else than the host machine.. | 08:03 |
| *** nerens has joined #openstack | 08:05 | |
| *** Nacx has joined #openstack | 08:05 | |
| *** rcc has joined #openstack | 08:08 | |
| *** dirakx has joined #openstack | 08:14 | |
| *** daveiw has joined #openstack | 08:20 | |
| *** eikke has joined #openstack | 08:29 | |
| *** MarkAtwood has joined #openstack | 08:37 | |
| *** diegoparrilla has joined #openstack | 08:38 | |
| diegoparrilla | hi all | 08:39 |
| diegoparrilla | I'm having problems uploading images. I get this error http://paste.openstack.org/show/834/ | 08:41 |
| *** z0_ has quit IRC | 08:42 | |
| fayce | guys, I have a question. Does Nova-Network do the same job as Eucalyptus-Cluster-Controller ? and for those who are not familiar with Eucalyptus: Do I need to put my Nova-Compute machines in a private network and put a Nova-Network machine between the public and the private network (with 2NICs) , letting then all the traffic to/from my VM pass trough the Nova-Network machine ? | 08:44 |
| ttx | fayce: Eucalyptus's CC would be a combination of our scheduler and network nodes, I think. | 08:46 |
| ttx | fayce: and you don't *have to* use 2 NICs, though the exact setup depends on the network model you chose | 08:47 |
| diegoparrilla | fayce: nova-network should have a NIC in a public network and another NIC in the service/private network, correct | 08:47 |
| ttx | diegoparrilla: not in VlanManager mode, for example | 08:47 |
| diegoparrilla | ttx: correct | 08:48 |
| fayce | ttx: thanks ! .... the idea of a CC that handles all the traffic between public network and all the boxes and VMs in it's private network, disturbs me a little bit ... | 08:48 |
| ttx | fayce: right, it doesn't scale. Especially since you can't duplicate the CC. | 08:49 |
| fayce | what if that CC machine (nova-network + nova-scheduler) crashes ? a | 08:49 |
| fayce | ok... | 08:49 |
| diegoparrilla | ttx: with VlanManager you take a different approach. But I have found a lot of problems running VlanManager with single NIc | 08:49 |
| fayce | is there any other more scalable approach to that ? | 08:50 |
| fayce | ttx: | 08:50 |
| diegoparrilla | fayce: In stackops we deploy nova-network in a single node, nova-compute in single nodes, nova-volume in single node and nova-api,scheduler,objectstore in a different node | 08:50 |
| ttx | fayce: in Nova, you can have lots of schedulers and network nodes, so you scale that as you grow your cloud | 08:50 |
| fayce | diegoparrilla and ttx , but you still have a bunch of nova-compute boxes that depends on the same nova-network box . right ? | 08:51 |
| fayce | depends = when the nova-network box crashes, traffic can't be routed to VMs | 08:52 |
| ttx | fayce: I'm far from being the resident networking expert, so it may depend on the model used. It scales, but it's not necessarily resilient to failure (if a compute node crashes, you lose VMs) | 08:53 |
| diegoparrilla | fayce: that's exactly the kind of testing I'm executing now... what happens when a node fails and how to recover it. network, compute and volume behaves in a very different way | 08:54 |
| *** MarkAtwood has left #openstack | 08:55 | |
| diegoparrilla | fayce: by the nature of network and volumes they should recover their state in case of a system crash. | 08:55 |
| diegoparrilla | fayce: compute nodes... well... if nova detects that the VMs are dead, what actions should be triggered? automatic restart? a recovery of state? report to the user? not an easy answer | 08:56 |
| fayce | ttx: a nova-compte node that crashes an kills all the VM is fine.... this is where load balancers/HA are useful.. | 08:57 |
| fayce | I'm just worried about the network model, where many physical machines (and their many*4-8 virtual machines) all depends on ONE machine (the nova-network) | 08:58 |
| fayce | is there just a way to have them (the nova-compute node) depend on more than just one nova-network ? | 08:58 |
| ttx | fayce: maybe soren or vishy will have an answer for you here | 08:59 |
| *** miclorb has quit IRC | 09:02 | |
| *** MarcMorata has joined #openstack | 09:04 | |
| vishy | fayce: there are possibilities in vlan mode but it requires some complicated routing to make metadata work properly | 09:04 |
| eikke | I'm trying to run Swift, but encounter failures | 09:04 |
| eikke | in the proxy-server, at a certain moment when creating a connection to some other service, headers for the request are passed, and the header value of one of the items in the dict (Content-Type) is None | 09:05 |
| eikke | on which Python's httplib fails | 09:05 |
| eikke | sounds like a known bug to anyone? just worked around it using a simple patch | 09:06 |
| *** allsystemsarego has joined #openstack | 09:07 | |
| *** allsystemsarego has joined #openstack | 09:07 | |
| diegoparrilla | vishy: any clue of what's happenning to me? http://paste.openstack.org/show/834/ ;-) | 09:09 |
| vishy | diegoparrilla: using the wrong keys perhaps | 09:10 |
| vishy | haven't seen that one before though | 09:10 |
| diegoparrilla | the fact is that other tasks that need keys run smoothly :-( | 09:10 |
| vishy | are you getting credentials using project zipfile? | 09:11 |
| diegoparrilla | yes | 09:11 |
| diegoparrilla | it works like a breeze | 09:11 |
| vishy | diegoparrilla: only the bundle uses the key | 09:11 |
| vishy | it is possible you have an old set of keys | 09:11 |
| vishy | from a previous install or something | 09:11 |
| vishy | i would redo project zipfile / unzip / . novarc | 09:12 |
| diegoparrilla | vishy: It's a fresh install... that's the strange thing. | 09:12 |
| vishy | then bundle and upload into a new bucket | 09:12 |
| vishy | hmm that is odd | 09:12 |
| vishy | did you wipe the box? | 09:12 |
| vishy | because you could have some old keys sitting around | 09:12 |
| diegoparrilla | vishy: yes, I have an automated process to deploy a fresh single/dual/multi node configuration of nova | 09:13 |
| diegoparrilla | vishy: with a custom build distribution for nova | 09:13 |
| diegoparrilla | vishy: I think some library is playing a trick on me about the certificates | 09:13 |
| vishy | yes very strange | 09:14 |
| diegoparrilla | vishy: but... you know... may be in the channel can give me a different view | 09:14 |
| vishy | perhaps nova-objectstore doesn't have the proper flagfile | 09:14 |
| *** hadrian has joined #openstack | 09:14 | |
| diegoparrilla | I found this https://bugs.launchpad.net/nova/+bug/724853 | 09:14 |
| *** MarkAtwood has joined #openstack | 09:14 | |
| uvirtbot | Launchpad bug 724853 in nova "after uploading images to nova-objectstore they are only decrypting " [Undecided,Invalid] | 09:14 |
| vishy | is nova-objectstore on the same host as nova-api? | 09:15 |
| diegoparrilla | yes | 09:15 |
| diegoparrilla | nova-api, nova-objectstore and nova-scheduler are in the same box | 09:15 |
| vishy | and they all have the proper flags telling them where to find the CA folder? | 09:16 |
| vishy | and they all can read the CA folder? | 09:16 |
| diegoparrilla | I'm using the default properties plus use_ca_project | 09:16 |
| diegoparrilla | yes, they can for sure | 09:16 |
| diegoparrilla | use_ca_project=true I mean | 09:16 |
| vishy | interesting perhaps use_ca_project is broken | 09:16 |
| diegoparrilla | vishy: I found out that the default use_ca_project parameter is false, but use_ca_project=true was needed | 09:17 |
| diegoparrilla | vishy: mmm... I'm going to turn to 'false' the use_ca_project parameter and try again | 09:18 |
| vishy | it isn't really needed | 09:18 |
| vishy | unless you are using vpns | 09:18 |
| diegoparrilla | vishy: correct. I was dealing with VlanManager and vpn | 09:18 |
| diegoparrilla | vishy: but I gave up and move to FlatDHCPMAnager mode :-( | 09:19 |
| vishy | hmm i believe we're using use_project_ca successfully | 09:20 |
| diegoparrilla | vishy: thanks vishy, I'm going to test some more stuff | 09:20 |
| vishy | it is a build slightly behind trunk | 09:20 |
| diegoparrilla | it's bexar 2011,1 | 09:20 |
| diegoparrilla | I don't work with the trunk | 09:20 |
| *** rchavik has joined #openstack | 09:20 | |
| vishy | but i don't know if we have tested actually generating new cas recently | 09:20 |
| vishy | we may be testing only with existing projects | 09:21 |
| diegoparrilla | vishy: no worries, I'm going to play a little bit with the image.py code and see what happens | 09:24 |
| *** rds__ has joined #openstack | 09:28 | |
| vishy | diegoparrilla: FWIW that code will all be moved by cactus, but if you find an issue please let me know | 09:32 |
| *** adjohn has quit IRC | 09:33 | |
| diegoparrilla | vishy: interesting. I will report the issue (if any). BTW,are you in the west coast? it's late there! | 09:33 |
| vishy | i know, up late making slides | 09:33 |
| vishy | :( | 09:33 |
| *** adjohn has joined #openstack | 09:33 | |
| *** adjohn has quit IRC | 09:39 | |
| HugoKuo | hi guys, in flat mode , metadata should be handled by gateway. is this correct ? | 09:45 |
| *** z0_ has joined #openstack | 09:48 | |
| *** fayce has quit IRC | 09:51 | |
| *** fayce has joined #openstack | 09:52 | |
| *** MarkAtwood has quit IRC | 09:59 | |
| *** guigui1 has quit IRC | 10:03 | |
| *** guigui1 has joined #openstack | 10:03 | |
| *** drico_ has quit IRC | 10:04 | |
| *** guigui has joined #openstack | 10:05 | |
| *** guigui1 has quit IRC | 10:08 | |
| *** kashyap has quit IRC | 10:17 | |
| ttx | HugoKuo: no, the gateway should forward metadata queries to the API server | 10:22 |
| HugoKuo | ttx: thanks that's useful info for me | 10:26 |
| ttx | HugoKuo: it's done for you in VlanManager mode, but you have to set up networking rules yourself in the other modes. | 10:27 |
| *** perestrelka has quit IRC | 10:30 | |
| *** shoaib has quit IRC | 10:32 | |
| *** diegoparrilla has quit IRC | 10:34 | |
| *** kashyap has joined #openstack | 10:35 | |
| *** miclorb_ has joined #openstack | 10:39 | |
| *** ddumitriu has joined #openstack | 10:47 | |
| *** z0_ has quit IRC | 10:50 | |
| *** fayce has quit IRC | 10:57 | |
| *** justinsb has quit IRC | 11:03 | |
| *** justinsb has joined #openstack | 11:03 | |
| *** guigui has quit IRC | 11:04 | |
| *** miclorb_ has quit IRC | 11:07 | |
| *** zenmatt has joined #openstack | 11:08 | |
| *** h0cin has joined #openstack | 11:08 | |
| *** h0cin has joined #openstack | 11:08 | |
| *** zenmatt has quit IRC | 11:10 | |
| *** zenmatt has joined #openstack | 11:10 | |
| *** z0_ has joined #openstack | 11:14 | |
| *** miclorb_ has joined #openstack | 11:15 | |
| *** jaysun has quit IRC | 11:21 | |
| *** zenmatt has joined #openstack | 11:22 | |
| *** jaysun_ has joined #openstack | 11:23 | |
| *** fayce has joined #openstack | 11:32 | |
| *** miclorb_ has quit IRC | 11:32 | |
| *** zenmatt has quit IRC | 11:34 | |
| *** zenmatt has joined #openstack | 11:34 | |
| *** rchavik has quit IRC | 11:36 | |
| *** zenmatt has quit IRC | 11:36 | |
| *** zenmatt has joined #openstack | 11:37 | |
| *** smaresca has quit IRC | 11:38 | |
| *** adjohn has joined #openstack | 11:41 | |
| *** z0_ has quit IRC | 11:48 | |
| *** fayce has quit IRC | 11:49 | |
| *** sebastianstadil has joined #openstack | 11:49 | |
| sebastianstadil | Hey guys, this is Sebastian from Scalr (aka openstack-platform-php on launchpad). | 11:50 |
| sebastianstadil | Is there any public Nova cluster we can get access to? | 11:50 |
| *** smaresca has joined #openstack | 11:51 | |
| *** MarcMorata has quit IRC | 11:54 | |
| *** Nacx has quit IRC | 11:54 | |
| *** MarcMorata has joined #openstack | 11:55 | |
| *** Nacx has joined #openstack | 11:55 | |
| *** thatsdone has joined #openstack | 12:01 | |
| *** ramkrsna_ has joined #openstack | 12:06 | |
| *** ramkrsna has quit IRC | 12:09 | |
| *** z0_ has joined #openstack | 12:16 | |
| *** [EGC-i]ThinIce has joined #openstack | 12:19 | |
| *** thatsdon_ has joined #openstack | 12:21 | |
| *** thatsdone has quit IRC | 12:22 | |
| *** ctennis has quit IRC | 12:27 | |
| HugoKuo | hello sebastian :> | 12:32 |
| *** ppetraki has joined #openstack | 12:35 | |
| *** ksteward has quit IRC | 12:38 | |
| *** ksteward has joined #openstack | 12:39 | |
| *** fayce has joined #openstack | 12:39 | |
| *** dillon-w has joined #openstack | 12:47 | |
| *** Failbait1 is now known as Failbaitr | 12:48 | |
| dillon-w | the object-servers in my swift cluster are doing unkown writing, is that normal? | 12:49 |
| *** aryan_ has quit IRC | 12:50 | |
| dillon-w | and object server is writing to /srv/node/sdX1/tmp directories. | 12:50 |
| *** aryan_ has joined #openstack | 12:51 | |
| dillon-w | some of the files are as large as 25GB. what are they? | 12:51 |
| dillon-w | according to 'file', these are "tmpDHMMu4: ASCII text, with very long lines, with no line terminators" | 12:53 |
| *** guigui1 has joined #openstack | 12:53 | |
| *** littleidea has joined #openstack | 12:57 | |
| dillon-w | notmyname, creiht, gholt, any idea? | 13:01 |
| *** guigui1 has quit IRC | 13:02 | |
| dsockwell | dillon-w: they're storage servers, i think it's safe to assume that those files are the storage | 13:11 |
| *** freeflying has quit IRC | 13:11 | |
| dillon-w | dsockwell: well, I hope so. but 1st of all, I haven't even created any account/container yet. | 13:11 |
| dsockwell | well that's a little disturbing | 13:12 |
| dillon-w | dsockwell: 2nd, these are text files, with only one line, all text was "system" | 13:12 |
| dsockwell | i'm afraid i'm not too familiar with swift | 13:12 |
| dsockwell | so the content of the file is 'systemsystemsystem... ' ? | 13:13 |
| dillon-w | how can one text file be 25GB big with only the word "system" in it? | 13:13 |
| dillon-w | yup | 13:13 |
| dsockwell | or 'system' + gibberish | 13:13 |
| dillon-w | it's just 'systemsystemsystem...' | 13:13 |
| dsockwell | that sounds like a Bug to me | 13:13 |
| *** freeflying has joined #openstack | 13:13 | |
| *** ctennis has joined #openstack | 13:14 | |
| dsockwell | and a big Bug, at that | 13:14 |
| dsockwell | probably worthy of an entry on that Launchpad doodad | 13:15 |
| dillon-w | well, if any swift core dev could help me confirm or identify that, i'll submit a bug | 13:16 |
| dsockwell | well good luck, they should be around eventually | 13:18 |
| *** z0_ has quit IRC | 13:18 | |
| dsockwell | whoever they are | 13:18 |
| dillon-w | thanks. r u in US too? it's early there, i'll have to wait for while | 13:19 |
| dsockwell | yes, i'm in the US. I would wait about 4 hours for everyone to wake up. | 13:20 |
| dsockwell | of course, some will wake up earlier than others | 13:21 |
| dsockwell | we might see east coast people as soon as 1 hour from now | 13:22 |
| dillon-w | so you are in west coast. | 13:23 |
| dillon-w | ? | 13:24 |
| dsockwell | I'm irresponsible, I stayed up too late drinking and now it's morning :( | 13:25 |
| dillon-w | :) another all-niter | 13:25 |
| *** hggdh has quit IRC | 13:25 | |
| *** sebastianstadil has quit IRC | 13:26 | |
| dsockwell | it's funny how that seems to happen | 13:26 |
| *** hggdh has joined #openstack | 13:27 | |
| *** hggdh has joined #openstack | 13:27 | |
| *** nerens has left #openstack | 13:28 | |
| *** sebastianstadil has joined #openstack | 13:30 | |
| dillon-w | it's half past 9 at night here. hopefully some one will show up before i have to leave for bed. | 13:34 |
| *** m_3 has joined #openstack | 13:35 | |
| dsockwell | with a time differential like that you're probably better off filing a bug report and having it rejected | 13:36 |
| jaypipes-afk | sandywalsh: so.... can we approve zones2 yet? | 13:36 |
| dsockwell | something like that 'systemsystemsystem...' can't be right | 13:36 |
| dsockwell | especially not 3 billion times over | 13:37 |
| dsockwell | not that I think it will be rejected, I apologize if i'm not as clear as I'd like to be | 13:38 |
| sandywalsh | jaypipes-afk, ttx is waiting to get novaclient in hudson so we can merge | 13:39 |
| sandywalsh | jaypipes-afk, ... the long and winding road | 13:39 |
| *** jaypipes-afk is now known as jaypipes | 13:40 | |
| jaypipes | sandywalsh: is soren around to install it on the build machine? | 13:40 |
| dsockwell | dillon-w: well it seems like better help than me has arrived, I'm going to duck out now. Good luck. | 13:41 |
| sandywalsh | jaypipes, not sure, ttx was looking for soren and mtaylor earlier | 13:42 |
| sebastianstadil | HugoKuo: Hi! | 13:42 |
| *** winston-d has quit IRC | 13:42 | |
| *** sebastianstadil has quit IRC | 13:43 | |
| ttx | jaypipes, sandywalsh: I have packages uploaded to https://launchpad.net/~ttx/+archive/ppa | 13:46 |
| *** ramkrsna_ has quit IRC | 13:46 | |
| ttx | jaypipes, sandywalsh; they need to be copied to the nova-core PPA and pushed to Hudson | 13:46 |
| ttx | soren: if you're around: ^ | 13:46 |
| dsockwell | ttx: do you know who dillon-w would talk to about strange behavior from swift? | 13:46 |
| dsockwell | he says it wrote a text file with 'system' repeated 4E9 times | 13:47 |
| ttx | dsockwell: the swift devs should be able to help him, they should be around in 1 or 2 hours | 13:47 |
| jaypipes | dsockwell: oh, that's Swift's new "ghost in the machine" mode. :P | 13:48 |
| dsockwell | so it's known? | 13:48 |
| jaypipes | dsockwell: sorry, I was joking :) | 13:48 |
| dsockwell | oh ok | 13:48 |
| ttx | sandywalsh: getting punished for adding an unpackaged build dep :P | 13:48 |
| jaypipes | gholt, redbo, notmyname, creiht: any idea about dillon-w's problems? | 13:48 |
| sandywalsh | ttx, sorry, what's the problem? | 13:49 |
| *** bkkrw has joined #openstack | 13:49 | |
| dillon-w | dsockwell: thanks. :) I'll stay up for another hour | 13:50 |
| ttx | sandywalsh: was joking about the long road. | 13:52 |
| *** dprince has joined #openstack | 13:53 | |
| *** reldan has joined #openstack | 13:54 | |
| sandywalsh | ttx, heh, yeah the long road had nothing to do with the packaging. Licensing/naming was a bigger PITA. | 13:54 |
| ttx | true :) | 13:55 |
| *** cascone has quit IRC | 14:00 | |
| *** dprince has quit IRC | 14:09 | |
| *** aliguori has joined #openstack | 14:09 | |
| *** dprince has joined #openstack | 14:10 | |
| *** lamar has quit IRC | 14:12 | |
| Code_Bleu | can openstack run Windows images, or is it strictly *nix based? | 14:13 |
| *** ericrw has joined #openstack | 14:15 | |
| soren | ttx: I can do it today, but not right now. Sorry. | 14:18 |
| soren | Any nova-core member can do it, though. | 14:21 |
| dendrobates | ttx what do you need? | 14:22 |
| *** naehring has quit IRC | 14:29 | |
| *** adjohn has quit IRC | 14:32 | |
| ttx | dendrobates: copy python-novaclient packages from ppa:ttx/ppa to ppa:nova-core/trunk, then install it on Hudson | 14:33 |
| ttx | soren: not everyone can install the package on Hudson though, which is what is blocking the branch atm | 14:33 |
| *** jero has joined #openstack | 14:35 | |
| dillon-w | Code_Bleu: I think Windows guest is not a problem as long as your hypervisor/hw supports them | 14:39 |
| dendrobates | ttx: I don't have access to hudson | 14:40 |
| ttx | dendrobates: yep, that's why I need soren or mtaylor :) But I think "today" is OK. | 14:43 |
| *** mray has joined #openstack | 14:45 | |
| *** thatsdon_ has quit IRC | 14:46 | |
| *** f4m8 is now known as f4m8_ | 14:47 | |
| *** z0_ has joined #openstack | 14:52 | |
| *** bcwaldon has joined #openstack | 14:54 | |
| *** aliguori has quit IRC | 14:56 | |
| *** dkocher has joined #openstack | 15:00 | |
| *** blakeyeager has joined #openstack | 15:00 | |
| *** aliguori has joined #openstack | 15:02 | |
| *** zenmatt has joined #openstack | 15:05 | |
| *** oldbam has joined #openstack | 15:07 | |
| dillon-w | any swift core dev around? for nwo? | 15:08 |
| dillon-w | for now? | 15:08 |
| _0x44 | soren: Hey, I'm in meetings all day today and have swapped review days with jk0 | 15:08 |
| oldbam | swauth_add_user results in error: "Could not retrieve account id value: /v1/AUTH_.auth/<account_name>". Does smth special need to be done for setting swauth except modifying proxy-server.conf? | 15:11 |
| gholt | dillon-w: Still there? | 15:12 |
| gholt | oldbam: Did you just switch to swauth from devauth, or just setting up swauth from scratch, or? | 15:13 |
| *** blakeyeager has quit IRC | 15:14 | |
| oldbam | i switched from devauth | 15:14 |
| *** rds__ has quit IRC | 15:15 | |
| dillon-w | gholt: hi there | 15:15 |
| gholt | Ah, okay, there's no amazing "make the switch" doc yet. :) But first, you'll need to run swauth-prep | 15:15 |
| gholt | oldbam: And then there's a tool swauth-auth-to-swauth that should help with transitioning the users over. | 15:16 |
| dillon-w | gholt: i'm still stucking at swauth-add-user | 15:16 |
| dillon-w | same error as yesterday | 15:17 |
| dillon-w | gholt: and do you know why object-server keeps writing to /srv/node/sdX1/tmp? | 15:18 |
| gholt | Ah, notmyname just told me you are the same as another irc nick | 15:18 |
| gholt | I have trouble remembering things, and that wasn't helping. :) | 15:19 |
| *** Ryan_Lane has joined #openstack | 15:19 | |
| *** jero has quit IRC | 15:19 | |
| gholt | tmp is used when uploading files, replicating, etc. pretty much most things really | 15:19 |
| *** blakeyeager has joined #openstack | 15:20 | |
| dillon-w | gholt: now 'tmp' in all storage nodes are bigger than 30GB. but I don't even successfully craeted a user yet. | 15:20 |
| gholt | Ah, well something's misconfigured then that's for sure. :) | 15:21 |
| gholt | My guess would be rsync, but just a guess. | 15:21 |
| gholt | If you shut down all the object replicators, does the i/o stop? | 15:21 |
| dillon-w | let me try that | 15:22 |
| *** MarkAtwood has joined #openstack | 15:22 | |
| dillon-w | by the way, all tmp files are text files, with only one word 'system' in it | 15:22 |
| *** reldan has quit IRC | 15:23 | |
| gholt | Hmm. That doesn't ring any bells, but interesting. | 15:23 |
| *** jero has joined #openstack | 15:25 | |
| dillon-w | object-server process is still writing to the disk, reported by 'iotop' | 15:25 |
| dillon-w | anyway, do you have any idea why 'swauth-add-user' failed? It seems that object-server didn't received PUT from proxy server. | 15:26 |
| * gholt wishes he had access to poke around. :) | 15:27 | |
| gholt | It could be ring errors, but it seems your object servers are getting accessed by something. | 15:27 |
| gholt | How many object servers do you have? | 15:27 |
| dillon-w | i have 4. | 15:28 |
| *** reldan has joined #openstack | 15:28 | |
| *** hub_cap has joined #openstack | 15:28 | |
| gholt | Okay, that's not too many. When you tail the logs on those nodes, are there any access log lines? | 15:29 |
| dillon-w | for object-server? | 15:29 |
| gholt | Ah yes, sorry, specifically for that right now | 15:29 |
| dillon-w | account/container replicator logs are the main part. | 15:30 |
| dillon-w | some object-auditor activity too | 15:30 |
| dillon-w | i used 'swift-ring-builder' to check object ring, it seems fine | 15:31 |
| gholt | Okay, I'm fuzzy on what we ended with yesterday, it was the proxy stating 503s from the object servers but you don't see any lines on the object servers? | 15:32 |
| *** imsplitbit has joined #openstack | 15:33 | |
| dillon-w | exactly. | 15:33 |
| *** dkocher has quit IRC | 15:33 | |
| gholt | In the proxy logs, they usually indicate the ip and port they were trying to connect to. Can you telnet to there from the proxy? | 15:34 |
| *** aliguori has quit IRC | 15:35 | |
| dillon-w | i tired. all object-server port can be access via telnet. | 15:36 |
| *** aliguori has joined #openstack | 15:36 | |
| gholt | Okay, well can you post the proxy logs again? I'm not sure why you'd be able to connect and the proxy says it can't. Maybe I'm remembering the error line incorrectly. | 15:36 |
| oldbam | gholt: after running swauth-prep, swauth-add-user results in a weird KeyError 'x-container-meta-account-id' | 15:37 |
| dillon-w | http://paste.openstack.org/show/819/ | 15:38 |
| gholt | dillon-w: Ah, timeout on final status, not connect. I did misremember. | 15:39 |
| gholt | oldbam: Can you paste that error? | 15:39 |
| gholt | dillon-w: If you see that 'final status' error on the proxy, and you see nothing on the corresponding server it's talking about, I'm not sure what I can do without access to poke around randomly myself. :/ | 15:40 |
| dillon-w | but object-server didn't seem to have received the request from proxy server, at least there was nothing about that in object server log | 15:40 |
| bcwaldon | I'd really appreciate it if I could get some eyes on this merge prop today: | 15:40 |
| bcwaldon | https://code.launchpad.net/~bcwaldon/nova/osapi-content-type-negotiation/+merge/52241 | 15:41 |
| *** littleidea has quit IRC | 15:41 | |
| dillon-w | gholt: well, unfortunately, my servers are behind company VPN. so... | 15:41 |
| gholt | Oh, I definitely understand. | 15:42 |
| gholt | The proxy believes it connected to *something* at 192.168.4.57:6000 and sent an http request for which it never got a response. If you see nothing on the other end... | 15:43 |
| *** troytoman-away is now known as troytoman | 15:43 | |
| dillon-w | gholt: hmm... that means maybe some other services were listening to 6000 port? I met this issue before, because VNC server use 6000 port. But if that happened, object-server won't be able to start. Now, i'm sure object-server has started, and is running. and it has writen tons of 'system' to tmp... | 15:45 |
| gholt | Well, you could always shutdown all object servers and make sure netstat -l shows nothing on 6000, hehe | 15:46 |
| oldbam | gholt: http://paste.openstack.org/show/835/ (error when running swauth-add-user) | 15:46 |
| dillon-w | good point | 15:46 |
| dillon-w | after shuting down object-server, 6000 is gone in netstat output. | 15:47 |
| *** z0_ has quit IRC | 15:48 | |
| dillon-w | also, swauth-add-user would result proxy-server complaining it cannot connect to object-server: Connection refused if i shutdown object-server. | 15:49 |
| gholt | oldbam: The first error is "Could not create account on the Swift cluster: /v1/AUTH_5029269d-7806-4f31-bb35-77ed0b4ce09c 401 Unauthorized" | 15:50 |
| gholt | dillon-w: "would result" or you just ran that it did actually result in that? | 15:51 |
| dillon-w | i ran it and it did result that | 15:52 |
| redbo | gholt is so untrusting | 15:52 |
| *** daveiw has quit IRC | 15:52 | |
| gholt | Heh, just being sure; I've misinterpreted so many things. | 15:52 |
| gholt | I wonder if whatever it is that sending systemblah-files to your object servers is keeping them so busy they aren't processing any other requests... | 15:53 |
| gholt | dillon-w: You might trying shutting down all swift services, servers, proxies, updaters, replicators, w/e and then just starting the object servers and seeing if they're still writing a bunch to tmp. | 15:54 |
| redbo | did netstat show anything connected to the object server? | 15:54 |
| *** dendrobates is now known as dendro-afk | 15:54 | |
| gholt | oldbam: I wonder if there's a problem because of the double-slash? | 15:55 |
| gholt | oldbam: https://10.0.0.2:8080//v1/ | 15:55 |
| oldbam | gholt: in the "Could not create account on the Swift cluster ... 401" which credentials are checked? super_admin key? | 15:57 |
| *** DigitalFlux has joined #openstack | 16:00 | |
| dillon-w | redbo: no, if stop all service and restart only object-server | 16:00 |
| oldbam | gholt: i wonder where does this double-slash come from. in the proxy-server.conf for default_swift_cluster i have only one slash. i also run the swauth-add-user with "-A https://10.0.0.2:8080/auth" | 16:01 |
| redbo | is the tmp directory still growing? | 16:01 |
| gholt | oldbam: Swauth uses a temporary token. Swauth assumes the proxy server being asked to create the account is configured to talk back to Swauth to validate the token. (most the time, they're the same machine and even server daemon) | 16:01 |
| gholt | oldbam: Ah, you're right, false alarm on that, it must be the log call code itself. | 16:02 |
| *** rnirmal has joined #openstack | 16:02 | |
| dillon-w | gholt: interesting, i stopped all server. and start only object-server, then do 'swauth-add-user', proxy-server complained it cannot connect to accout server. then i start account-server & container-server. then do 'swauth-add-user' again, it still failed, (saying no final status of PUT from object-server), BUT once the 'swauth-add-user' executed, object-server started to write tmp! | 16:03 |
| dillon-w | redbo: please see my answer to gholt. | 16:03 |
| gholt | dillon-w: And it keeps on a writing? | 16:03 |
| dillon-w | gholt: yes. | 16:03 |
| redbo | huh... sounds like the wsgi server or read-loop in object-server getting stuck in a loop. | 16:04 |
| gholt | dillon-w: Can you post one of your object-server.conf? | 16:04 |
| dillon-w | for today, it already consumed > 300GB space | 16:05 |
| dillon-w | gholt: sure. | 16:05 |
| gholt | dillon-w: I don't expect to find anything there, but ya never know. :/ | 16:05 |
| dillon-w | here it is: http://paste.openstack.org/show/836/ | 16:06 |
| *** dragondm has joined #openstack | 16:07 | |
| gholt | Well durn, nothing. Pristine. | 16:07 |
| creiht | dillon-w: what platform are you running on? | 16:08 |
| gholt | What version of eventlet? | 16:08 |
| gholt | RHEL creiht | 16:08 |
| creiht | oh | 16:08 |
| creiht | well there is his issue :P | 16:08 |
| gholt | Heheh | 16:08 |
| redbo | seems like request.body_file.read() might be stuck returning 'system' over and over. | 16:09 |
| creiht | dillon-w: and what version of python? | 16:09 |
| dillon-w | creiht: python-2.6.5-3.el6.x86_64 | 16:10 |
| dillon-w | gholt: eventlet-0.9.14-py2.6.egg | 16:11 |
| gholt | That's pretty dang new. Newer than mine. :) | 16:11 |
| gholt | Uhm, I guess check the webob version too? | 16:12 |
| dillon-w | well, i have to leave for now. it's already mid-nite. my wife complained a lot... :( | 16:12 |
| gholt | Heh, I can imagine. | 16:12 |
| redbo | (he really can't) | 16:12 |
| creiht | lol | 16:13 |
| gholt | I know, I've seen it before, really. | 16:13 |
| dillon-w | hopefully, tomorrow morning (my time) some of you are still around. creiht used to be. thanks for the help, all of you. :) bye for now | 16:13 |
| creiht | hah | 16:13 |
| *** iammartian has joined #openstack | 16:14 | |
| creiht | dillon-w: my wife complains when I do that :) | 16:14 |
| dillon-w | creiht: then we have a deadlock. :D | 16:14 |
| *** dillon-w has quit IRC | 16:15 | |
| *** mahadev has joined #openstack | 16:16 | |
| *** blakeyeager_ has joined #openstack | 16:16 | |
| redbo | it's like dining philosophers, but with creiht instead of a philosopher | 16:17 |
| creiht | aww | 16:17 |
| *** DigitalFlux has quit IRC | 16:17 | |
| *** blakeyeager has quit IRC | 16:18 | |
| *** blakeyeager_ is now known as blakeyeager | 16:18 | |
| *** dendro-afk is now known as dendrobates | 16:21 | |
| *** ccustine has joined #openstack | 16:26 | |
| *** kashyap has quit IRC | 16:32 | |
| *** blueadept has joined #openstack | 16:33 | |
| oldbam | gholt: can the problem with swauth-add-user be a result of self-generated certificate over https connection? | 16:35 |
| *** bcwaldon has quit IRC | 16:37 | |
| *** enigma has joined #openstack | 16:38 | |
| gholt | oldbam: No, that should be no problem at all. The 401 itself is the proxy saying it can't validate the token it got as good. | 16:39 |
| *** rds__ has joined #openstack | 16:40 | |
| gholt | oldbam: What's your configuration? 1 proxy with swauth, or swauth with an external proxy, or? | 16:40 |
| *** letterj has left #openstack | 16:41 | |
| *** bcwaldon has joined #openstack | 16:45 | |
| *** bcwaldon has quit IRC | 16:47 | |
| *** bcwaldon has joined #openstack | 16:47 | |
| oldbam | one proxy with swauth | 16:49 |
| oldbam | gholt: one proxy with swauth | 16:49 |
| gholt | Very odd, [trying imagine why it can't authenticate with it's own durn self] | 16:50 |
| *** zul has quit IRC | 16:50 | |
| *** letterj has joined #openstack | 16:50 | |
| *** ChanServ sets mode: +v letterj | 16:50 | |
| gholt | oldbam: Can you paste your proxy-server.conf [masking the super_admin_key of course] | 16:50 |
| *** kashyap has joined #openstack | 16:51 | |
| oldbam | gholt: http://paste.openstack.org/show/837/ (knowing my super_admin_key is harmless :) | 16:53 |
| gholt | Hehe | 16:53 |
| gholt | Okay, and 10.0.0.2 is definitely that same box right? Sorry, for the probably dumb questions | 16:54 |
| gholt | token_life = 0, that's probably it | 16:55 |
| oldbam | 10.0.0.2 - is the box for proxy and auth | 16:55 |
| gholt | I'm pretty sure that means "don't let tokens have a life" | 16:55 |
| gholt | token_life should equal the number of seconds you'd like tokens to be valid for. | 16:55 |
| oldbam | token = 0 worked in devauth, but maybe for the temporary token it does not work | 16:56 |
| *** msassak has joined #openstack | 16:56 | |
| gholt | Hmm, could be. It's a new/different codebase, so probably just doesn't have the exact same translations. | 16:56 |
| gholt | If you want, you can just remove the line completely and it'll default to 1 day, 86400 seconds. | 16:58 |
| *** paltman has joined #openstack | 16:59 | |
| oldbam | gholt: you saved my evening! thanks a lot ;) | 16:59 |
| *** z0_ has joined #openstack | 17:00 | |
| *** nelson has quit IRC | 17:00 | |
| gholt | Hehe, working okay now? [makes a note to put error checking around that setting for the next swift version] | 17:00 |
| *** mahadev has quit IRC | 17:01 | |
| *** eikke has quit IRC | 17:02 | |
| *** MarkAtwood has quit IRC | 17:03 | |
| oldbam | gholt: yeah, works. btw, a doc update to run swauth-prep is also required when switching from devauth :) another thing i noticed was that swauth-add-user hangs if -A is not specified and default url does not work. or maybe i was too impatient to wait till timeout | 17:07 |
| jk0 | soren: looks like hudson had a problem merging your cross-binary-sync merge (was unable to load the lockfile module) FYI | 17:12 |
| *** littleidea has joined #openstack | 17:13 | |
| jarrod | if i set my kvm instances to use FlatDHCPManager, will the bridge interface get assigned anything my local DHCP server gives it? even a public IP? | 17:13 |
| *** blakeyeager has quit IRC | 17:14 | |
| *** zul has joined #openstack | 17:18 | |
| *** MarkAtwood has joined #openstack | 17:20 | |
| *** Nacx has quit IRC | 17:20 | |
| jaypipes | justinsb: feel free to set https://code.launchpad.net/~justin-fathomdb/nova/bug724623/+merge/52505 back to Work In Progress while you fix the merge conflict thingie. I tried to explain why that .THIS file got into the branch in a comment on the merge prop. | 17:22 |
| justinsb | Thanks jay - just did that :-) | 17:23 |
| jaypipes | justinsb: cool, thx mate | 17:23 |
| justinsb | jaypipes: I'm not using git-bzr, and I don't think I tried going under the covers on a resolve, but who knows what I did... | 17:24 |
| jaypipes | justinsb: it's my official "review day" today, so I'll be pushing through the backlog... might be bugging you with lots of questions, so be prepared ;) | 17:24 |
| jaypipes | justinsb: k, those were the two reasons I could think of. Usually, it's when someone deletes the .BASE or .OTHER files not understanding what they are... | 17:24 |
| jaypipes | mtaylor: ping | 17:25 |
| *** mahadev has joined #openstack | 17:26 | |
| *** mahadev_ has joined #openstack | 17:30 | |
| *** mahadev has quit IRC | 17:31 | |
| jaypipes | dubs: if you have a free minute today, a review on this would be great. https://code.launchpad.net/~jaypipes/glance/checksum/+merge/52569 thx in advance! | 17:31 |
| dubs | sure thing | 17:32 |
| dubs | gonna grab some lunch and i'll check it out right afterwards | 17:33 |
| justinsb | jaypipes: Can't remember - are you the Hudson expert? I want to know if it's OK to use RabbitMQ in the unit tests or do I have to use fake_rabbit instead? | 17:33 |
| jaypipes | dubs: cheers | 17:34 |
| jaypipes | justinsb: not in the unit tests. at least not now I think. | 17:34 |
| jaypipes | justinsb: there's been some efforts to put some more functional/integration tests into the Hudson setup. soren has some, vishy and termie have some, tr3buchet's got some, but nobody's integrated them into the OpenStack Hudson builders yet, AFAIK. | 17:35 |
| justinsb | jaypipes: OK... I'll see if I can fix it. Might take another MP off your plate for the day :-) | 17:35 |
| *** zul has quit IRC | 17:35 | |
| jaypipes | hehe | 17:35 |
| *** z0_ has quit IRC | 17:37 | |
| *** dendrobates is now known as dendro-afk | 17:38 | |
| jaypipes | jk0, dabo: secondary reviews on https://code.launchpad.net/~justin-fathomdb/nova/bug724623/+merge/52505 would be great. feel free to set merge prop to approved if both of you are in agreement with the patch. Ignore the comment of merge conflicts at the top. LP just hasn't "caught up" with the most recent commit on the branch that removed the conflicts... | 17:39 |
| * jk0 looks | 17:39 | |
| *** z0_ has joined #openstack | 17:39 | |
| *** rlucio has joined #openstack | 17:39 | |
| *** zul has joined #openstack | 17:40 | |
| *** joearnold has joined #openstack | 17:41 | |
| jaypipes | dragondm: I like this: 8+AccountCommands = ProjectCommands | 17:46 |
| jaypipes | dragondm: that's the definition of "punt" :)( | 17:47 |
| jaypipes | dragondm: overall, reviewing your multi-tenant branch. looks very good. nice work! | 17:47 |
| dragondm | heh yah. | 17:48 |
| dragondm | thx | 17:48 |
| *** aliguori has quit IRC | 17:53 | |
| *** eikke has joined #openstack | 17:56 | |
| *** bkkrw has quit IRC | 17:57 | |
| jaypipes | dragondm: question... why did you add a **kw to most of the methods in the OS controllers? | 17:59 |
| dragondm | ah, ya, prolly should have pulled that too... It was a relic of the account change that I had pulled before. | 18:00 |
| *** aliguori has joined #openstack | 18:00 | |
| dragondm | previously the mergeprop was following ziad & co.'s spec and added the account to the server-management url like cloudfiled does. | 18:01 |
| dragondm | s/cloudfiled/cloudservers/ | 18:01 |
| jaypipes | dragondm: ah, I see. | 18:02 |
| dragondm | that proved controversial, and given the changes w/ the forthcoming authn/authz stuff that was pulled. | 18:02 |
| jaypipes | dragondm: I'm going to ask you to remove those in the review, since the **kw isn't used in any of the methods. ok? | 18:02 |
| dragondm | sure | 18:02 |
| jaypipes | dragondm: k. cheers | 18:02 |
| dragondm | ya, I just forgot those :> | 18:03 |
| *** oldbam has quit IRC | 18:10 | |
| *** Ryan_Lane is now known as Ryan_Lane|food | 18:12 | |
| *** dendro-afk is now known as dendrobates | 18:13 | |
| dragondm | jaypipes: removed those **kw's and re-pushed | 18:15 |
| jaypipes | dragondm: thx mate | 18:15 |
| *** CloudChris has left #openstack | 18:15 | |
| *** rcc has quit IRC | 18:17 | |
| *** zul has quit IRC | 18:18 | |
| *** zul has joined #openstack | 18:19 | |
| *** mahadev_ has quit IRC | 18:20 | |
| *** mahadev has joined #openstack | 18:21 | |
| *** rlucio has quit IRC | 18:21 | |
| *** eikke has quit IRC | 18:25 | |
| _cerberus_ | vishy: question for you -> the instance_type attribute on instance is string. Is that a foreign key to the name attribute on the InstanceType model? | 18:25 |
| _cerberus_ | I ask because you own the line that adds it to the Instances model :-) | 18:26 |
| justinsb | jaypipes: I fixed up my testing-volumes & keys branch . Right now Vish is the only reviewer on there (I think because I resubmitted it before I knew better). Should I re-add nova-core? | 18:30 |
| mtaylor | jaypipes: pong | 18:30 |
| jaypipes | justinsb: I'm currently reviewing it. | 18:30 |
| jaypipes | mtaylor: hey, so did you fix Tarmac to only return the failed test output? I saw this: https://code.launchpad.net/~soren/nova/cross-binary-sync/+merge/51533 and thought you might have made that change | 18:31 |
| mtaylor | jaypipes: I did not - but it's possible that upstream did | 18:32 |
| *** MarcMorata has quit IRC | 18:34 | |
| *** diegoparrilla has joined #openstack | 18:42 | |
| kpepple | _cereberus, vishy : that column was added after instance_types table (instance_types was added in revno757) ... so no. i would be concerned about adding it as a foreign key as you can delete instance_types (with --purge) | 18:44 |
| kpepple | _cerberus_: sorry ... see above ^^^^^^ | 18:44 |
| jaypipes | justinsb: man, you're not going to like my review... I really hate to do this. | 18:46 |
| jaypipes | justinsb: I feel like I'm the big bad wolf stomping on just your merge props. :( | 18:47 |
| justinsb | jaypipes: Well I think (most) reviews make the code better... yours definitely do! | 18:47 |
| openstackhudson | Project nova build #609: SUCCESS in 1 min 53 sec: http://hudson.openstack.org/job/nova/609/ | 18:48 |
| openstackhudson | Tarmac: replaced ConnectionFailed with Exception in tools/euca-get-ajax-console | 18:48 |
| openstackhudson | was not working for me with euca2tools 1.2 (version 2007-10-10, release 31337) | 18:48 |
| justinsb | jaypipes: I think we need testing - 9 months into the project we still don't have good tests... | 18:49 |
| jaypipes | justinsb: I'm trying my best to be flexible. but this string of dependent merge props is not making me feel good. I feel bad saying that, because I like *some* of the code, but the web of interdependency between all the branches as well as the inclusion of new code features that aren't in scope for Cactus make it very difficult. | 18:49 |
| justinsb | jaypipes: I need the keys because otherwise I can't launch a server | 18:50 |
| justinsb | jaypipes: If I pull out the volume code though, can be let it in? | 18:50 |
| jaypipes | justinsb: you can't launch a server without the keys stuff? how so? | 18:50 |
| *** jaysun_ has quit IRC | 18:50 | |
| justinsb | jaypipes: I can't launch a server without a key registered - it throws an exception | 18:51 |
| jaypipes | justinsb: so have the test register the key... that's what I mean by introducing the new keys.API functionality... it wasn't necessary to test the code paths. | 18:51 |
| justinsb | jaypipes: The problem with the existing tests is that they keep cheating by doing that and going behind the scenes. So all the tests pass, but then when someone tries to do it for real, it fails hard | 18:52 |
| justinsb | jaypipes: So like my metadata patch, and it worked originally, but it got broken in the review process; the tests always passed but it simply didn't work. | 18:53 |
| jaypipes | justinsb: which is what the smoketests are for... when doing real functional testing, you need to perform these kind of things *outside* of the process that the servers are running in... | 18:53 |
| justinsb | jaypipes: That was pretty embarrassing to me, and I wanted to fix it for real | 18:53 |
| jaypipes | justinsb: I totally understand your motivation for these patches. I do. But I am expressing my opinion that the code destabilizes more than it adds in testing. | 18:54 |
| justinsb | jaypipes: It does move things around (API keys). I see that | 18:55 |
| justinsb | jaypipes: I guess ot | 18:55 |
| justinsb | jaypipes: whoops | 18:55 |
| justinsb | jaypipes: I guess I think we have to make internal code changes to get us stability, but it sounds like you're interpreting it as fewer internal changes | 18:56 |
| justinsb | jaypipes: So "code stability" (i.e. change as little code as possible) vs "stability of nova" (i.e. something that works) | 18:56 |
| *** dkocher has joined #openstack | 18:57 | |
| jaypipes | justinsb: we need to stabilize the existing code base, and adding *any* new functionality or changing existing code paths serves to destabilize the code base. Not saying those code changes aren't good or that they won't be wanted post-Cactus, but for now, I need to see tests of existing code paths before I approve code that adds new code paths. Does that make sense? | 18:57 |
| justinsb | jaypipes: I missed the summit (NFI) - which was the definition of stability we're aiming for? | 18:57 |
| *** jaysun has joined #openstack | 18:58 | |
| justinsb | jaypipes: I understand where you're coming from, but I feel like we're kicking an increasingly embarrassing can down the road | 18:58 |
| dprince | dabo: there? | 18:59 |
| Vek | looking at LP722979 (convert calls to session.execute("update <blah>") to SQLAlchemy code), and I have a quick question. In places like trunk/nova/db/sqlalchemy/api.py:704, using the session.execute() has the effect of not automatically changing the 'updated_at' field. The naive approach to changing this code to use SQLAlchemy code will cause updated_at to be modified. Is this a problem? | 19:00 |
| justinsb | jaypipes: I'm going to put my "two definitions of stability" stuff on the MP and we'll see what others think. I see where you're coming from, but I think we should be willing to make sacrifices for stability definition #2 | 19:01 |
| jaypipes | justinsb: is this branch supposed to be testing the OS API or the EC2 API? Because the OS API does not have keys OR volumes... | 19:02 |
| justinsb | jaypipes: It's supposed to test the OS API, creating a server really, but that's in the next MP. I need keys for that. I definitely agree that it's unfortunate that volumes got mixed in. | 19:04 |
| jaypipes | justinsb: OK, this is what is so confusing... why do you need Keys to test the OS API? OS API does not use keys... | 19:04 |
| justinsb | jaypipes: http://bazaar.launchpad.net/~hudson-openstack/nova/trunk/view/head:/nova/api/openstack/servers.py | 19:06 |
| justinsb | jaypipes: Check out line 152 | 19:06 |
| justinsb | jaypipes: I'm slightly worried by the idea that there are not supposed to be keypairs in OS, but that's a whole different topic :-) | 19:07 |
| *** eikke has joined #openstack | 19:08 | |
| *** gregp76 has joined #openstack | 19:09 | |
| *** MarkAtwood has quit IRC | 19:10 | |
| jaypipes | justinsb: I don't see why you can't just stubout auth_manager.AuthManager.get_key_pairs to return a pre-determined keypair... Your patch adds a /keys API resource to the OS API when there is none at all. The code you just pointed to above is *internal* code to the /servers API endpoint controller, and has nothing to do with public-facing testing.. | 19:11 |
| *** Ryan_Lane|food is now known as Ryan_Lane | 19:11 | |
| justinsb | jaypipes: I wanted real tests which don't stub things out, because otherwise the tests have a nasty habit of passing when things are actually broken. | 19:13 |
| justinsb | jaypipes: I wanted tests that I could trust | 19:13 |
| justinsb | jaypipes: I don't get the internal code point... Users have to create a key to create a server, so my test does that too | 19:14 |
| jaypipes | justinsb: but it creates a key through a method that doesn't exist in the OS API. That's my point. | 19:14 |
| sirp_ | jaypipes: i'm looking into https://bugs.launchpad.net/glance/+bug/731684 ; the assumption here is that requiring 'swift' to run the unit-tests is unacceptable. do you agree with that assumption? | 19:14 |
| uvirtbot | Launchpad bug 731684 in glance "Glance lazy loading issue -- still coupled with Swift" [High,In progress] | 19:15 |
| justinsb | justinsb: Sure, it uses a private API like the admin API, until key support is added | 19:15 |
| justinsb | jaypipes: Ooops - talking to myself there. At least I agree with me :-) | 19:15 |
| jaypipes | justinsb: let's see what other reviewers have to say, ok? we're just going round in circles... | 19:15 |
| justinsb | jaypipes: OK! | 19:15 |
| jaypipes | justinsb: :) | 19:16 |
| *** littleidea has quit IRC | 19:16 | |
| jaypipes | sirp_: I disagree with that assumption, yes. | 19:16 |
| jaypipes | sirp_: meaning I think requiring swift is perfectly fine. | 19:17 |
| sirp_ | jaypipes: so, for every backend we write, will we required the modules are installed, or are we going to consider 'swift' a privileged backend? | 19:17 |
| jaypipes | sirp_: and it's not "just to run the unittests". anything that imports swift.store and calls get_from_backend() will hit this issue. | 19:18 |
| sirp_ | jaypipe: not if we use a lazy-loading pattern, which the bug seeks to add | 19:18 |
| sirp_ | only if the code path actually uses swift would be import the swift module | 19:19 |
| sirp_ | *would we | 19:19 |
| jaypipes | sirp_: get_from_backend() just imports glance.store.swift, which imports swift.common.client... | 19:19 |
| sirp_ | right, my branch changes that | 19:19 |
| sirp_ | swift isn't import module-wide in my branch, only when it's actually used | 19:20 |
| jaypipes | sirp_: ok, cool with me. as long as it doesn't make the code incredibly difficult to follow, I'm fine with the lazy-loading pattern. | 19:20 |
| sirp_ | jaypipes: i ask, because i'm going to have to swift the current swift tests to use fakes rather than stubout (since the swift module to stubout won't be required) | 19:21 |
| jaypipes | sirp_: I was *really* trying to avoid that. | 19:21 |
| sirp_ | jaypipes: yeah, i noticed that while reading the code, that's why i'm pinging you :) | 19:21 |
| jaypipes | sirp_: the reason is because the Fakes completely get out-of-sync with the real thing and end up masking way too much and causing subtle bugs. | 19:22 |
| jaypipes | sirp_: for the tests, I'd rather stick with the existing test and just use unittest2 or try to figure out programmatically how to use nosetests's skipper. | 19:22 |
| *** gholt_ has joined #openstack | 19:23 | |
| sirp_ | jaypipes: one option would be to try to import swift, on ImportError, log a warning, and then skip the swift-stubout tests... | 19:24 |
| *** dkocher has quit IRC | 19:24 | |
| jaypipes | sirp_: uhm, that's what was in there before I was asked to remove that and use absolute_import instead... | 19:25 |
| jaypipes | sirp_: remember all the if SWIFT_INSTALLED: blocks? | 19:25 |
| sirp_ | jaypipes: not really, i'd need to go back and look | 19:25 |
| sirp_ | jaypipes: if that's the case, wanna just write that off as a bad decision, and go back to that pattern? (seems like the best compromise in this case) | 19:26 |
| jaypipes | sirp_: sure, sounds good. | 19:27 |
| sirp_ | jaypipes: cool | 19:27 |
| jaypipes | sirp_: you can put the entire test case in the try: except ImportError block, which would remove the need to use the if SWIFT_INSTALLED: blocks. Looks ugly, but it works. | 19:28 |
| jaypipes | sirp_: you'd have to go and re-trim all the lines back to 80 chars wide, too ;) | 19:28 |
| *** tahoe has joined #openstack | 19:28 | |
| jaypipes | sirp_: alternately, we could require unittest2 and not need the if SWIFT_INSTALLED blocks and just use the @skipIf decorator. | 19:29 |
| jaypipes | sirp_: I tried adding unittest2 as a Glance depends before, but was told it wasn't good to have it as a dependency. | 19:29 |
| jaypipes | sirp_: which is kinda funny, since Sqlalchemy installs unittest2 anyway ;)( | 19:29 |
| *** burris has quit IRC | 19:29 | |
| sirp_ | jaypipes: heh, i'll play with a couple of those options you mentioned and see which looks cleanest | 19:30 |
| jaypipes | sirp_: cheers | 19:30 |
| *** mgoldmann has quit IRC | 19:30 | |
| *** paltman has quit IRC | 19:33 | |
| *** diegoparrilla has quit IRC | 19:36 | |
| uvirtbot | New bug: #732204 in nova "Make keypairs optional in the Openstack API" [Undecided,New] https://launchpad.net/bugs/732204 | 19:36 |
| *** photron has joined #openstack | 19:38 | |
| *** tahoe has quit IRC | 19:38 | |
| ttx | mtaylor: maybe you can help, we need some PPA package copy (copy python-novaclient from ppa:ttx/ppa to ppa:nova-core/trunk) and install python-novaclient on hudson, to unblock the zones2 branch | 19:39 |
| mtaylor | ttx: I can always help | 19:39 |
| ttx | mtaylor: asked soren but he is a bit busy so might not tackle it today | 19:39 |
| *** tahoe has joined #openstack | 19:39 | |
| ericrw | execvp patch finished. merge request submitted. | 19:39 |
| ttx | ericrw: cool! wll definitely have a look. | 19:40 |
| *** mahadev_ has joined #openstack | 19:41 | |
| mtaylor | ttx: copied - will install them once they build in the new ppa | 19:41 |
| ttx | mtaylor: sounds good. Will push to natty proper once we are happy with it. | 19:41 |
| mtaylor | ttx: btw - we're only a few days away from having the jenkins plugin that will let us spin up new cloud servers for tests - which will make the "install deps" thing be handled much better | 19:42 |
| ttx | mtaylor: cool, who coded that ? | 19:42 |
| mtaylor | ttx: I started it, ccustine finished it ... he's doing some final testing on it currently | 19:42 |
| *** littleidea has joined #openstack | 19:43 | |
| ttx | ok | 19:43 |
| *** dinu has quit IRC | 19:43 | |
| *** mahadev has quit IRC | 19:43 | |
| mtaylor | ttx: it will honestly make our testing stuff WAY the hell cleaner | 19:44 |
| *** reldan has quit IRC | 19:44 | |
| dubs | jaypipes: just finished review of checksum branch | 19:45 |
| jaypipes | dubs: cheers cory | 19:45 |
| jaypipes | ironcamel: around? | 19:46 |
| *** brd_from_italy has joined #openstack | 19:48 | |
| jaypipes | sirp_: I'm beginning to hate sqlalchemy-migrate. | 19:48 |
| jaypipes | sirp_: scratch that. I hate sqlalchemy-migrate. | 19:48 |
| *** deepheat has quit IRC | 19:48 | |
| sirp_ | jaypipes: welcome to the club, it consists of everybody, and we meet at the bar :) | 19:48 |
| *** deepheat has joined #openstack | 19:49 | |
| ironcamel | jaypipes: hey | 19:50 |
| jaypipes | ironcamel: hey, sending your http-error-codes off to Hudson. should be in trunk shortly if nothing fails in the automated tests. | 19:51 |
| ironcamel | sweet | 19:51 |
| jaypipes | ironcamel: sorry for the delay on it... backlog of reviews. | 19:51 |
| ironcamel | thank you | 19:51 |
| *** jheiss has quit IRC | 19:52 | |
| justinsb | jaypipes: Tearing through those MPs! If you actually clear the backlog, there won't be anything left for whoever is up tomorrow :-) | 19:52 |
| *** jheiss has joined #openstack | 19:52 | |
| jaypipes | justinsb: heh. I highly doubt that. :) | 19:53 |
| jaypipes | sandywalsh: I know I must have asked you this already, but are we ready to set zones2 to approved? :) | 19:54 |
| *** MarkAtwood has joined #openstack | 19:54 | |
| sandywalsh | jaypipes, if novaclient has been added to hudson ... yes | 19:55 |
| jaypipes | mtaylor: ? ^^ | 19:57 |
| *** bkkrw has joined #openstack | 19:58 | |
| openstackhudson | Project nova build #610: SUCCESS in 1 min 48 sec: http://hudson.openstack.org/job/nova/610/ | 19:58 |
| openstackhudson | Tarmac: Fixes bug #729400. Invalid values for offset and limit params in http requests now return a 400 response with a useful message in the body. Also added and updated tests. | 19:58 |
| uvirtbot | Launchpad bug 729400 in nova "http query params limit and offset are ignored when they are malformed" [Undecided,In progress] https://launchpad.net/bugs/729400 | 19:58 |
| jaypipes | ironcamel: merged. | 19:58 |
| mtaylor | jaypipes: yes. waiting for it to hit the ppa | 19:58 |
| jaypipes | mtaylor: is this a manual or automated process? something we can speed up? | 19:58 |
| sandywalsh | mtaylor, jaypipes I think ttx already did that | 19:59 |
| mtaylor | jaypipes: automated. patience young jedi | 19:59 |
| *** MarkAtwood has quit IRC | 20:00 | |
| *** MarkAtwood has joined #openstack | 20:00 | |
| *** paltman has joined #openstack | 20:01 | |
| *** retr0h has joined #openstack | 20:03 | |
| *** retr0h has joined #openstack | 20:03 | |
| *** h0cin has quit IRC | 20:03 | |
| *** maplebed has joined #openstack | 20:04 | |
| *** perestrelka has joined #openstack | 20:05 | |
| uvirtbot | New bug: #732220 in openstack-dashboard "credentials email subject template not stripping whitespace" [High,In progress] https://launchpad.net/bugs/732220 | 20:05 |
| *** DigitalFlux has joined #openstack | 20:09 | |
| openstackhudson | Project nova build #611: SUCCESS in 1 min 48 sec: http://hudson.openstack.org/job/nova/611/ | 20:13 |
| openstackhudson | Tarmac: Add a decorator that lets you synchronise actions across multiple binaries. Like, say, ensuring that only one worker manipulates iptables at a time. | 20:13 |
| soren | \o/ | 20:15 |
| *** daveiw has joined #openstack | 20:17 | |
| * ericrw has some cleanup to do. | 20:20 | |
| *** littleidea has quit IRC | 20:21 | |
| *** reldan has joined #openstack | 20:23 | |
| *** joearnold has quit IRC | 20:25 | |
| *** clauden_ has joined #openstack | 20:27 | |
| *** MarcMorata has joined #openstack | 20:27 | |
| *** bcwaldon has quit IRC | 20:30 | |
| ericrw | pep8 clean :) | 20:32 |
| *** ppetraki has quit IRC | 20:32 | |
| jarrod | How does rackspace have public ip addresses assigned to their instances? | 20:33 |
| mtaylor | jaypipes: ok. installed | 20:34 |
| jarrod | here is a better question -- does rackspace actually use openstack? | 20:34 |
| Ryan_Lane | jarrod: I don't believe they currently do | 20:34 |
| jarrod | well that makes sense | 20:35 |
| mtaylor | jarrod: I believe they use it for cloud files but not yet for cloud servers | 20:35 |
| Ryan_Lane | but you can use openstack nova and have public ips | 20:35 |
| jarrod | how | 20:35 |
| Ryan_Lane | jarrod: if you use flat dhcp mode, and use floating ips | 20:35 |
| Ryan_Lane | or vpn mode | 20:35 |
| *** rds__ has quit IRC | 20:35 | |
| Ryan_Lane | or just use a public range | 20:36 |
| soren | Or, if you're using VLAN mode, just assign a public subnet to the instances. | 20:36 |
| Ryan_Lane | err | 20:36 |
| jarrod | soren that doesnt work | 20:36 |
| Ryan_Lane | vlan mode, that's what I meant :) | 20:36 |
| uvirtbot | New bug: #732236 in nova "bzr clone lp:nova returns Errno 110" [Undecided,New] https://launchpad.net/bugs/732236 | 20:36 |
| soren | jarrod: Could have fooled me. | 20:36 |
| jarrod | because nova network jacks something up | 20:36 |
| soren | jarrod: How doesn't it work? | 20:36 |
| jaypipes | mtaylor: cheers, and thx | 20:36 |
| jaypipes | vishy, xtoddx: what is the purpose of flags.DECLARE? | 20:36 |
| xtoddx | jaypipes: so you can find flags that are defined in other modules | 20:37 |
| Ryan_Lane | jarrod: I believe you need to modify the security groups to allow access (if you haven't done that) | 20:37 |
| Ryan_Lane | I don't know vlan mode very well though. | 20:37 |
| xtoddx | jaypipes: i think we use it in testing because not everything is imported that has the flags we reference in some tests | 20:38 |
| jaypipes | xtoddx: instead of just importing the module? | 20:38 |
| xtoddx | jaypipes: right. i don't know the pros and cons of declare vs import the module though. | 20:38 |
| jaypipes | xtoddx: seems wonky to me. | 20:38 |
| soren | declare wouldn't cause circular imports, I suppose. | 20:39 |
| xtoddx | jaypipes: the gflags docstring might explain it better | 20:39 |
| eday | jaypipes: whats the state of conf in glance? anything worth putting into openstack-common for burrow to use? | 20:39 |
| eday | jaypipes: I was looking at it yesterday, wasn't clear how it's being used | 20:39 |
| jaypipes | eday: we ripped out gflags entirely. glance is almost identical to swift as far as options and conf processing. | 20:40 |
| jaypipes | eday: in other words, we use the standard optparse and ConfigParser modules, along with paste.deploy. | 20:40 |
| eday | jaypipes: hmm, ok. I need something less wsgi specific, as wsgi is just one frontend for burrow | 20:42 |
| jaypipes | eday: uhm, ConfigParser? | 20:42 |
| jaypipes | eday: all paste.deploy uses is ConfigParser. that's a stdlib module. | 20:42 |
| eday | jaypipes: I was referring to paste.deploy, but yeah, configparser is fine | 20:43 |
| jaypipes | eday: paste.deploy uses ConfigParser. | 20:43 |
| eday | jaypipes: yup | 20:43 |
| jaypipes | death to flagfiles. | 20:43 |
| eday | jaypipes: I was asking more about the optparse/configparser stuff you were working on for openstack-common | 20:43 |
| jaypipes | eday: I gave up on it. gflags folks aren't budging, and swift wanted nothing to do with module-level defines of options, and don't think have CLI options for everything is a good idea. And I happen to agree with them. | 20:44 |
| eday | jaypipes: ahh, ok | 20:45 |
| *** ppetraki has joined #openstack | 20:46 | |
| *** ahmedelgamil has joined #openstack | 20:48 | |
| eday | jaypipes: I still want module-level defines for burrow, trying to figure out the best way to do this (and figure out how to have optparse make multiple runs, ignoring some options, if possible) | 20:48 |
| jaypipes | eday: easy. just have the module have a function called add_options(parser) and pass an optparse.OptionParser to it. Add the module-level options as an OptionGroup or not... | 20:49 |
| jaypipes | eday: though I almost guarantee you'll find in the long run that putting options in config files with limited CLI options for a few things is a better strategy. | 20:50 |
| *** ppetraki has quit IRC | 20:50 | |
| *** daveiw has quit IRC | 20:51 | |
| eday | jaypipes: it's a chicken/egg problem though. don't know what module to load (to add options) until you parse, but parsing unknown options errors | 20:52 |
| eday | jaypipes: and I'm still a fan of conf options should == cli :) | 20:52 |
| jaypipes | eday: I'm not, sorry... | 20:52 |
| *** gholt_ has quit IRC | 20:52 | |
| *** DigitalFlux has quit IRC | 20:52 | |
| eday | jaypipes: you used to be, I lost you! | 20:53 |
| jaypipes | eday: yeah, after actually using Glance, it was easier to use Swift's usage pattern (bin/swift-init start <server> conf_file... | 20:53 |
| jaypipes | eday: CLI options aren't useful for deploying much at all. | 20:54 |
| eday | jaypipes: sure, but are useful for dev and some users | 20:54 |
| jaypipes | eday: only a few options are useful in testing (like --debug, --verbose, etc). The rest is just easier to put into a standard conf file and supply the conf file when testing. | 20:54 |
| eday | jaypipes: I'm actually thinking of doing that since optparse isn't quite what I need, argparse is too new, and I don't want to write my own | 20:55 |
| jaypipes | eday: yes, please don't write your own. | 20:56 |
| *** mahadev_ has quit IRC | 20:56 | |
| eday | jaypipes: which is why I was hoping you did and put it in openstack-common :) | 20:56 |
| jaypipes | eday: after having used all these different strategies, I have to say I think gflags adds way more bugs and opportunities for misuse than any benefit it might have (I don't feel there are many, but that's just my opinion) | 20:57 |
| *** mahadev has joined #openstack | 20:57 | |
| *** bcwaldon has joined #openstack | 20:57 | |
| jaypipes | eday: what I wrote for openstack-common was pretty much what I described to you above: use optparse and if you want to have module-defined options, have a function in the module that gets supplied a optparse.OptionParser. The openstack-common approach was just some convenience wrappers around that general concept. | 20:58 |
| eday | jaypipes: I'm def not considering gflags, no worries there :) | 20:58 |
| *** MarcMorata has quit IRC | 20:58 | |
| *** gholt has quit IRC | 20:58 | |
| *** laurensell has joined #openstack | 20:58 | |
| *** ddumitriu has quit IRC | 21:00 | |
| *** bcwaldon_ has joined #openstack | 21:00 | |
| *** gholt has joined #openstack | 21:00 | |
| * blamar_ cheers | 21:01 | |
| *** mahadev_ has joined #openstack | 21:01 | |
| *** imsplitbit has quit IRC | 21:02 | |
| *** bcwaldon has quit IRC | 21:03 | |
| *** mahadev has quit IRC | 21:04 | |
| *** ppetraki has joined #openstack | 21:07 | |
| *** ahmedelgamil has quit IRC | 21:09 | |
| *** allsystemsarego has quit IRC | 21:14 | |
| * soren grumbles and puts on his debugging hat | 21:14 | |
| *** MarcMorata has joined #openstack | 21:19 | |
| soren | It's an actual hat, by the way: http://blog.linux2go.dk/debugging_hat.jpg | 21:20 |
| westmaas_ | haha | 21:21 |
| *** littleidea has joined #openstack | 21:21 | |
| dabo | cute | 21:22 |
| *** bcwaldon has joined #openstack | 21:26 | |
| openstackhudson | Project nova build #612: SUCCESS in 1 min 52 sec: http://hudson.openstack.org/job/nova/612/ | 21:28 |
| openstackhudson | Tarmac: Update the create server call in the Openstack API so that it generates an 'adminPass' and calls set_admin_password in the compute API. This gets us closer to parity with the Cloud Servers v1.0 spec. | 21:28 |
| *** bcwaldon_ has quit IRC | 21:28 | |
| jarrod | is openstack compute ready for production yet? | 21:30 |
| annegentle | jarrod: it's in production at NASA | 21:31 |
| jarrod | bexar? | 21:32 |
| *** ddumitriu has joined #openstack | 21:32 | |
| openstackhudson | Project nova build #613: SUCCESS in 1 min 51 sec: http://hudson.openstack.org/job/nova/613/ | 21:33 |
| openstackhudson | Tarmac: Hi guys | 21:33 |
| openstackhudson | This branch fixes lp722982 (ability to delete networks with nova-manage) and lp715427 (nova-manage does not check if a network exists before creating it) . | 21:33 |
| openstackhudson | Regards | 21:33 |
| *** lamar has joined #openstack | 21:34 | |
| annegentle | jarrod: I'm not sure if they run a packaged release | 21:34 |
| jarrod | i wonder which network mode they use | 21:35 |
| *** letterj has quit IRC | 21:35 | |
| annegentle | jarrod: pretty sure it's vlan, they have pretty strict network reqs at NASA | 21:36 |
| *** bcwaldon has quit IRC | 21:36 | |
| xtoddx | jarrod: we run VlanManager at NASA | 21:38 |
| xtoddx | and we build our own packages from the ~anso/deploy branch | 21:38 |
| *** letterj has joined #openstack | 21:38 | |
| *** ChanServ sets mode: +v letterj | 21:38 | |
| annegentle | xtoddx: ah, thanks, that makes sense | 21:38 |
| jarrod | thanks man | 21:40 |
| jaypipes | phew...live-migrations review was a beast. | 21:42 |
| jaypipes | soren: love the debugging hat. you look like a character from World of Warcraft. | 21:43 |
| soren | I think I'm offended by that comment :) | 21:44 |
| *** joearnold has joined #openstack | 21:44 | |
| *** bcwaldon has joined #openstack | 21:44 | |
| eday | soren: you should be | 21:46 |
| comstud | agree | 21:47 |
| jaypipes | soren: just expressing my mind ;P | 21:50 |
| *** miclorb_ has joined #openstack | 21:50 | |
| creiht | I was thinking that picture belongs on a pokemon card | 21:51 |
| soren | jaypipes: Yeah, well.. I don't love you anymore now. | 21:51 |
| soren | jaypipes: :) | 21:51 |
| jaypipes | lol :) | 21:51 |
| jk0 | jaypipes: http://fuckyeahnouns.com/jaypipes | 21:51 |
| soren | \o/ | 21:52 |
| jk0 | http://fuckyeahnouns.com/soren | 21:52 |
| jk0 | :D | 21:52 |
| jk0 | this site is fun | 21:52 |
| soren | jk0: You just put that there, didn't you. | 21:52 |
| jk0 | nah, I think it searches for the word and tries to match a pic with it | 21:53 |
| soren | http://fuckyeahnouns.com/jk0 | 21:53 |
| jk0 | mine sucks :( | 21:53 |
| soren | Oh, that's how it works. | 21:54 |
| *** [EGC-i]ThinIce has quit IRC | 21:55 | |
| *** dprince has quit IRC | 21:55 | |
| kpepple | eday: time for a quick questions ? | 21:55 |
| jaypipes | jk0: wow, that's an oldie. :) where'd you find that one? | 21:56 |
| jk0 | the site searches for them somewhere | 21:56 |
| jk0 | type in something into that box and it will try to match | 21:56 |
| jaypipes | jk0: ah :) | 21:56 |
| * jk0 was typing in openstack names | 21:56 | |
| jk0 | yours was the best | 21:56 |
| jk0 | haha | 21:56 |
| jaypipes | jk0: well, it was an actual picture of me... in my transformers t-shirt that I no longer fit in ;) | 21:57 |
| *** patri0t has quit IRC | 21:57 | |
| jk0 | I feel ya | 21:57 |
| jaypipes | hehe | 21:57 |
| *** patri0t has joined #openstack | 21:58 | |
| kpepple | eday: i'm trying to retire lp721254 which I believe is being caused by nova/compute/api.py:168 ... or the lack of a line like 'state': '0x00' in that block. Is there a reason that we are using ''state_description': 'scheduling'" in the base_options ? this state_description is not part of the EC2 standard (http://docs.amazonwebservices.com/AWSEC2/latest/APIReference/index.html?ApiReference-query-DescribeInstances.html) and I believe it should be 'pendin | 21:59 |
| *** dirakx has quit IRC | 22:01 | |
| openstackhudson | Project nova build #614: FAILURE in 24 sec: http://hudson.openstack.org/job/nova/614/ | 22:01 |
| openstackhudson | Tarmac: Fixes bug 726359. Passes unit tests. | 22:01 |
| uvirtbot | Launchpad bug 726359 in nova "utils.execute uses potentially insecure shell=True" [High,Fix committed] https://launchpad.net/bugs/726359 | 22:01 |
| openstackhudson | Changes parameters of utils.execute to utils.execute(*cmd, **kwargs). The kwargs themselves have not changed, other than changing check_exit_code to default to 0. The exit code of the process is now checked against this variable, unless None. | 22:01 |
| uvirtbot | New bug: #732274 in nova "Should be a way to avoid having a server password generated" [Undecided,New] https://launchpad.net/bugs/732274 | 22:01 |
| ericrw | hmm | 22:02 |
| ericrw | oh hudson... | 22:03 |
| eday | kpepple: yeah, we probably need a default state in there | 22:03 |
| *** ke4qqq has quit IRC | 22:03 | |
| soren | kpepple: nova/compute/api.py should be agnostic to what the EC2 API says. If this is believed to be a problem, it needs to be fixed in the EC2 frontend to map from internal state to EC2-like states. | 22:03 |
| soren | ericrw: hm? | 22:03 |
| ericrw | soren: that was my patch that failed. | 22:04 |
| eday | soren: it may not be ec2 specific, but agree if it is nova.compute is not the place to fix | 22:04 |
| soren | ericrw: Ah, no. I don't think it failed. | 22:04 |
| soren | ericrw: Let me check. I changed something in that general area yesterday which may have caused this. | 22:05 |
| kpepple | soren: while i agree with you in principal, i believe we still need to be setting a power state there if we are going to set a power state description ... although i am unsure why we need both power state code and a power state description in the db (since one is supposed to map to the other) | 22:06 |
| soren | kpepple: Yeah, that's always been a bit of a mystery to mee, too. | 22:06 |
| tr3buchet | does anyone know if there is a way to get the id of the current zone in nova? | 22:06 |
| *** cw_ has quit IRC | 22:07 | |
| uvirtbot | New bug: #732277 in nova "Password Generation (in)security" [Undecided,New] https://launchpad.net/bugs/732277 | 22:07 |
| *** cw has joined #openstack | 22:07 | |
| eday | kpepple: I would set state: 0 as the fix, seems resonable to me | 22:08 |
| *** brd_from_italy has quit IRC | 22:08 | |
| *** MarcMorata has quit IRC | 22:08 | |
| kpepple | eday , soren : my fix is to put a "0x00" state code into that section (basically initializing a state code along with the proper state description) to get us to cactus and fix the broken api and then put in a blueprint / bug to get harmonize power states between EC2 / Openstack API. sound reasonable ? | 22:08 |
| *** ke4qqq has joined #openstack | 22:09 | |
| kpepple | eday , soren : btw bobya (who reported the bug) has been using this fix for the last week with no problems | 22:09 |
| soren | Yeah, that's probably fine. | 22:10 |
| openstackhudson | Project nova build #615: STILL FAILING in 14 sec: http://hudson.openstack.org/job/nova/615/ | 22:11 |
| openstackhudson | Tarmac: - Content-Type and Accept headers handled properly | 22:11 |
| openstackhudson | - Content-Type added to responses | 22:11 |
| openstackhudson | - Query extensions no long cause computeFaults | 22:11 |
| openstackhudson | - adding wsgi.Request object | 22:11 |
| openstackhudson | - removing request-specific code from wsgi.Serializer | 22:11 |
| kpepple | eday , soren : thx. pushing up fix. | 22:11 |
| jaypipes | vishy: looks like a quick merge and resolve is needed for the kill-objectstore branch due to some branches that recently hit trunk... | 22:12 |
| *** photron has quit IRC | 22:13 | |
| *** ctennis has quit IRC | 22:13 | |
| vishy | okes | 22:13 |
| soren | ericrw: Ah, yes, it broke the docs build. | 22:14 |
| soren | ericrw: [pid 27938] execve("cd /home/soren/src/openstack/nova/nova/doc && ./generate_autodoc_index.sh", ["cd /home/soren/src/openstack/nov"...], [/* 51 vars */]) = -1 ENOENT (No such file or directory) | 22:14 |
| ericrw | pfft.. docs ;-) | 22:15 |
| soren | Well, it means tarballs don't get built, nor do packages :) | 22:15 |
| soren | Cascading failure. | 22:16 |
| *** maplebed has quit IRC | 22:16 | |
| ericrw | I'll check it out | 22:16 |
| soren | Awesomesauce. | 22:16 |
| soren | ./doc/ext/nova_autodoc.py | 22:16 |
| soren | is the culprit. | 22:16 |
| ericrw | pretty clear, looking at it | 22:17 |
| *** blueadept has quit IRC | 22:19 | |
| *** littleidea has quit IRC | 22:22 | |
| * annegentle cries | 22:24 | |
| ericrw | soren: do I submit a new bug report and attach the fix to that? | 22:24 |
| *** littleidea has joined #openstack | 22:24 | |
| *** burris has joined #openstack | 22:25 | |
| *** ctennis has joined #openstack | 22:26 | |
| *** ctennis has joined #openstack | 22:26 | |
| soren | ericrw: Technically, yes. | 22:29 |
| *** ppetraki has quit IRC | 22:29 | |
| soren | ericrw: ...but I don't particularly care. The problem and solution are both obvious. | 22:29 |
| *** mahadev_ has quit IRC | 22:29 | |
| ericrw | I simply submitted a proposal for merge | 22:29 |
| soren | ericrw: link? | 22:29 |
| soren | Ah, I see it. | 22:30 |
| ericrw | https://code.launchpad.net/~ewindisch/nova/execvp/+merge/52775 | 22:30 |
| *** mahadev has joined #openstack | 22:30 | |
| soren | Another review on that branch would be lovely. tarballs and packages are not being built until it's in. | 22:30 |
| * soren hugs jk0 | 22:32 | |
| jk0 | :) | 22:32 |
| *** ddumitriu has quit IRC | 22:36 | |
| *** azneita has joined #openstack | 22:36 | |
| ericrw | now, before Hudson comes back, I'm going to disappear and pretend my migraine doesn't exist. | 22:37 |
| *** ddumitriu has joined #openstack | 22:37 | |
| *** londo has quit IRC | 22:39 | |
| *** reldan has quit IRC | 22:40 | |
| *** jheiss has quit IRC | 22:40 | |
| soren | ericrw: Almost there. | 22:41 |
| *** romain_lenglet_ has joined #openstack | 22:41 | |
| *** reldan has joined #openstack | 22:41 | |
| ericrw | still fails, or Hudson is almost going to try it? | 22:42 |
| soren | Almost done trying it. | 22:42 |
| *** maplebed has joined #openstack | 22:42 | |
| soren | Wait for it.... | 22:42 |
| openstackhudson | Yippie, build fixed! | 22:43 |
| openstackhudson | Project nova build #616: FIXED in 1 min 49 sec: http://hudson.openstack.org/job/nova/616/ | 22:43 |
| openstackhudson | Tarmac: Fixes doc build after execvp patch. | 22:43 |
| soren | Bam. | 22:43 |
| ericrw | sweet | 22:43 |
| soren | ericrw: Thanks! For doing all the work to begin with and for fixing it now. | 22:43 |
| soren | ericrw: I've been meaning to fix it for ages, but never worked myself up to diving in to those millions of execute calls. good job! | 22:43 |
| ericrw | I figured what else to do on a PHL-SFO flight? | 22:44 |
| soren | I've never been on a PHL-SFO where I didn't write a patch just like that one. That's all I can say. | 22:44 |
| *** londo has joined #openstack | 22:48 | |
| *** adjohn has joined #openstack | 22:51 | |
| *** lamar has quit IRC | 22:56 | |
| *** hub_cap has quit IRC | 22:57 | |
| *** adjohn has quit IRC | 22:57 | |
| *** ddumitriu has quit IRC | 23:04 | |
| *** matiu has joined #openstack | 23:05 | |
| *** msassak_ has joined #openstack | 23:12 | |
| *** msassak has quit IRC | 23:14 | |
| *** clauden_ has quit IRC | 23:19 | |
| *** eikke has quit IRC | 23:19 | |
| *** clauden has joined #openstack | 23:21 | |
| *** msassak_ has quit IRC | 23:24 | |
| *** troytoman is now known as troytoman-away | 23:27 | |
| *** sirp_ has left #openstack | 23:29 | |
| *** sirp_ has joined #openstack | 23:30 | |
| *** hggdh has quit IRC | 23:32 | |
| *** hggdh has joined #openstack | 23:35 | |
| *** hggdh has joined #openstack | 23:35 | |
| *** mray has quit IRC | 23:39 | |
| *** bcwaldon has quit IRC | 23:45 | |
| *** romain_lenglet_ has quit IRC | 23:46 | |
| *** clauden___ has joined #openstack | 23:46 | |
| *** clauden has quit IRC | 23:46 | |
| *** clauden___ is now known as clauden | 23:46 | |
| *** dendrobates is now known as dendro-afk | 23:48 | |
| openstackhudson | Project nova build #617: SUCCESS in 1 min 51 sec: http://hudson.openstack.org/job/nova/617/ | 23:48 |
| openstackhudson | Tarmac: Fixes a race condition where multiple greenthreads were attempting to resize a file at the same time. Adds tests to verify that the image caching call will run concurrently for different files, but will block other greenthreads trying to cache the same file. | 23:48 |
| *** dendro-afk is now known as dendrobates | 23:53 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
