| *** achanda has joined #kolla | 00:00 | |
| *** achanda has quit IRC | 00:04 | |
| *** SourabhP has quit IRC | 00:07 | |
| *** tfukushima has joined #kolla | 00:07 | |
| *** tfukushima has quit IRC | 00:12 | |
| *** achanda has joined #kolla | 00:12 | |
| *** diga has joined #kolla | 00:13 | |
| *** dims__ has quit IRC | 00:16 | |
| *** diogogmt has joined #kolla | 00:39 | |
| *** dims__ has joined #kolla | 00:51 | |
| *** zhiwei has joined #kolla | 00:55 | |
| *** dims__ has quit IRC | 00:56 | |
| *** weiyu_ has joined #kolla | 01:00 | |
| *** dayou has joined #kolla | 01:02 | |
| *** kjelly has quit IRC | 01:07 | |
| *** achanda has quit IRC | 01:09 | |
| *** diga has quit IRC | 01:20 | |
| *** erkules_ has joined #kolla | 01:21 | |
| *** erkules has quit IRC | 01:24 | |
| *** tfukushima has joined #kolla | 01:38 | |
| *** tfukushima has quit IRC | 01:42 | |
| *** diogogmt has quit IRC | 01:46 | |
| *** diogogmt has joined #kolla | 01:48 | |
| *** kjelly has joined #kolla | 01:57 | |
| *** achanda has joined #kolla | 02:04 | |
| *** achanda has quit IRC | 02:14 | |
| *** rhallisey has quit IRC | 02:15 | |
| *** achanda has joined #kolla | 02:23 | |
| *** unicell has quit IRC | 02:31 | |
| *** achanda has quit IRC | 02:49 | |
| *** achanda has joined #kolla | 03:02 | |
| *** loth has quit IRC | 03:04 | |
| *** rajalokan has joined #kolla | 03:06 | |
| *** achanda has quit IRC | 03:08 | |
| SamYaple | ubuntu gate builds in 19 minutes | 03:15 |
|---|---|---|
| SamYaple | how bout that | 03:15 |
| *** dayou has quit IRC | 03:20 | |
| *** alisonh has quit IRC | 03:20 | |
| *** dayou has joined #kolla | 03:23 | |
| *** rajalokan has quit IRC | 03:29 | |
| *** rajalokan has joined #kolla | 03:30 | |
| *** alisonh has joined #kolla | 03:38 | |
| *** tfukushima has joined #kolla | 03:38 | |
| *** tfukushima has quit IRC | 03:43 | |
| *** rajalokan has quit IRC | 03:52 | |
| *** rajalokan has joined #kolla | 03:53 | |
| *** rajalokan has quit IRC | 03:57 | |
| openstackgerrit | Kuo-tung Kao proposed stackforge/kolla: update the image-building document https://review.openstack.org/219550 | 04:00 |
| *** dims__ has joined #kolla | 04:03 | |
| *** SourabhP has joined #kolla | 04:06 | |
| *** rajalokan has joined #kolla | 04:08 | |
| *** unicell has joined #kolla | 04:10 | |
| *** unicell has quit IRC | 04:16 | |
| *** dims__ is now known as dims | 04:17 | |
| *** weiyu_ has quit IRC | 04:18 | |
| *** dims has quit IRC | 04:24 | |
| *** unicell has joined #kolla | 04:24 | |
| *** tfukushima has joined #kolla | 04:39 | |
| *** achanda has joined #kolla | 04:42 | |
| *** tfukushima has quit IRC | 04:44 | |
| openstackgerrit | Merged stackforge/kolla: Improve scripts with some check before kick rm command https://review.openstack.org/218719 | 05:00 |
| *** unicell has quit IRC | 05:07 | |
| *** unicell has joined #kolla | 05:09 | |
| *** weiyu has joined #kolla | 05:16 | |
| *** tfukushima has joined #kolla | 05:20 | |
| *** alisonh has quit IRC | 05:25 | |
| *** inc0 has joined #kolla | 05:27 | |
| inc0 | good morning | 05:27 |
| inc0 | what's the status | 05:27 |
| inc0 | ? | 05:27 |
| inc0 | of multinode | 05:27 |
| *** alisonh has joined #kolla | 05:31 | |
| *** weiyu has quit IRC | 05:34 | |
| *** weiyu has joined #kolla | 05:36 | |
| *** weiyu has quit IRC | 05:51 | |
| SamYaple | inc0: pretty sure its just sdakes gear | 05:57 |
| SamYaple | he had it plugged into 4 switches | 05:57 |
| SamYaple | moving it around helped a bit | 05:57 |
| SamYaple | thne he found an ip conflict with his vip | 05:58 |
| inc0 | I see | 05:58 |
| SamYaple | i have tested multinode all over the place, no problems | 05:58 |
| inc0 | yeah that makes stuff harder | 05:58 |
| SamYaple | definetely not merging that balance source patch now | 05:58 |
| inc0 | yeah I've just hit a wall, it seems keepalived doesn't work very well on openstack-provisioned vms | 05:58 |
| SamYaple | not unless it can be reproduced literally anywhere else | 05:58 |
| SamYaple | inc0: thats how i do most of my kolla devel | 05:59 |
| SamYaple | what are you seeing? | 05:59 |
| inc0 | vip doesn't get ping from same vlan | 05:59 |
| SamYaple | whats hosting the openstack stuff? | 05:59 |
| inc0 | but I guess that might be network conf on compute nodes, still debugging | 05:59 |
| inc0 | its our in-house OS juno, neutron, vlan | 06:00 |
| SamYaple | so you setup the appropriate allowed address pairs with neutron, yes? | 06:00 |
| SamYaple | otherwise neutron is gonna drop that vip | 06:00 |
| inc0 | you mean ip's to port? | 06:01 |
| inc0 | hmm...no I haven't | 06:01 |
| SamYaple | each port only allows a single ip/mac combo to come out of it by default (enforced by iptables) | 06:01 |
| SamYaple | yea thats the issue | 06:01 |
| SamYaple | you know how to do it? or i can point oyu in the right direction | 06:01 |
| inc0 | yeah I was meant to check iptables later on | 06:01 |
| inc0 | I'll find out, don't worry | 06:02 |
| SamYaple | yea it dropping the vip there for sure | 06:02 |
| kjelly | Hi guys. I am working on https://review.openstack.org/#/c/219550/ . Swapnil hopes me to include the method of building images like source/binary. I am not sure what's different between binary and source. Does binary mean that openstack package will be installed from apt/yum ? What does the source means ? | 06:03 |
| inc0 | I thought it might be something like that | 06:03 |
| inc0 | kjelly, exactly what it means, it will be installed from github-pulled source;) | 06:04 |
| inc0 | or rather downloaded tarball of source | 06:04 |
| inc0 | tarball will be extracted in container and pip installed | 06:04 |
| SamYaple | inc0: in case you care | 06:05 |
| SamYaple | #openstack.2014-12-10.log:03:48 < SamYaple> neutron port-update <port-id> --allowed-address-pairs type=dict list=true mac_address=<port-mac-address>,ip_address=<vip> | 06:05 |
| SamYaple | #openstack.2014-12-10.log:03:49 < SamYaple> neutron port-update 034afb8f-b62e-4058-9f38-8e510950eb35 --allowed-address-pairs type=dict list=true mac_address=ba:f5:3e:e3:67:ab,ip_address=192.168.10.10/32 | 06:05 |
| inc0 | thanks a lot Sam | 06:05 |
| kjelly | inc0: what's code version it download ? Is it possible to change the version ? | 06:06 |
| SamYaple | if you build a vpn type instance, that can also accept a range like 192.168.1.0/24 | 06:06 |
| SamYaple | kjelly: yes | 06:06 |
| inc0 | kjelly, by default it's master, you can change it | 06:06 |
| SamYaple | kjelly: check build.ini | 06:06 |
| inc0 | in build.ini | 06:06 |
| SamYaple | https://github.com/stackforge/kolla/blob/master/build.ini | 06:06 |
| SamYaple | kjelly: just change the url to anything you want | 06:06 |
| SamYaple | tarballs.openstack.org has a tarball for every tag ever created | 06:07 |
| SamYaple | damn | 06:07 |
| SamYaple | 2 years of irclogs and only consumes 87MB | 06:07 |
| SamYaple | _uncompressed_ | 06:07 |
| kjelly | SamYaple inc0: Ok. | 06:10 |
| *** diogogmt has quit IRC | 06:11 | |
| *** weiyu_ has joined #kolla | 06:17 | |
| kjelly | SamYaple inc0: I fully understood. Thanks. | 06:19 |
| inc0 | any time | 06:19 |
| *** sdake has joined #kolla | 06:24 | |
| inc0 | hmm now I guess I have some keepalived conf problem | 06:24 |
| inc0 | when I manually ip add, it works, but keepalived doesn't | 06:24 |
| SamYaple | its always something with you.... | 06:24 |
| sdake | samyaple around | 06:25 |
| inc0 | yup, I'm just a developer | 06:25 |
| sdake | samyaple i deployed aio to all 3 nodes individually, they all work fantastically well | 06:25 |
| inc0 | network is magic | 06:25 |
| sdake | i ws wondering if yo uwould try node3, then node1 in the list | 06:25 |
| sdake | this will put node 3 as master | 06:25 |
| SamYaple | :) keepalived wont assign the ip until haproxy is up an running for 20 seconds | 06:25 |
| sdake | and see if it still misbehaves in the same way -a walys node 3 | 06:25 |
| inc0 | SamYaple, it's there;) | 06:25 |
| sdake | i suspect it will not | 06:25 |
| SamYaple | sdake: sure, but im not quite following what test you want me to do | 06:25 |
| sdake | chnge ivnetory file to list node3, then node1 | 06:26 |
| SamYaple | inc0: is it faliing the health check though? | 06:26 |
| sdake | in this scenario, I think node 1 will always fail | 06:26 |
| sdake | or be "flakey" | 06:26 |
| sdake | with this mode ndoe3 will be listed first the inveentory list | 06:26 |
| SamYaple | oh i see what youre saying | 06:26 |
| inc0 | SamYaple, nope, it's ok | 06:26 |
| sdake | and will be master | 06:27 |
| SamYaple | inc0: and none of the hosts have the vip? | 06:27 |
| inc0 | no, vip is there | 06:27 |
| SamYaple | oh ok then i dont know whats happening | 06:27 |
| SamYaple | sdake: sure lets try it | 06:27 |
| *** chandankumar has joined #kolla | 06:28 | |
| openstackgerrit | Sam Yaple proposed stackforge/kolla: WIP: Ceph Ansible - READY FOR REVIEWS https://review.openstack.org/218570 | 06:29 |
| sdake | becaue a deploy to node 3 with glacne image-list in a repittion works great | 06:29 |
| SamYaple | well yea because its a single host | 06:29 |
| SamYaple | even if it was a bad host it would still work | 06:29 |
| sdake | if the ram was flakey it would implode | 06:29 |
| sdake | whether its single node or not | 06:30 |
| SamYaple | i dontthink its the ram | 06:30 |
| SamYaple | since changing to a single switch made htings noticibly better (less failures) i blame your networking | 06:30 |
| sdake | i am willing to simplify that - just woke up | 06:31 |
| SamYaple | me too | 06:31 |
| sdake | need to warm up a bit befor i go jerking around with network cables | 06:31 |
| SamYaple | sdake: youre still running diferent kernels fyi | 06:33 |
| sdake | try rebooting | 06:33 |
| sdake | node 2 had an old kernel it appears | 06:35 |
| SamYaple | yea i was making sure the grub confs were right | 06:35 |
| SamYaple | but i think you rebooted it | 06:35 |
| sdake | yup | 06:35 |
| sdake | therecorrect kernel versoin now | 06:36 |
| SamYaple | sdake: you are using the devicemapper with the loopback devices (killing performance) is this what yo uwant? | 06:37 |
| sdake | its what i want for nwo | 06:37 |
| SamYaple | cool just checking | 06:37 |
| sdake | i just am usign defaults | 06:37 |
| sdake | i just want the damn thing to work - can worry about optimization later | 06:37 |
| SamYaple | for some reason 103 still has 148 in its arp table | 06:42 |
| SamYaple | ok the hosts look good | 06:45 |
| SamYaple | which deploy would you like to test now? all 3 with 103 as the first one in the groups? | 06:45 |
| SamYaple | sdake: ^ | 06:46 |
| sdake | try just two10 | 06:46 |
| sdake | 103, 101 | 06:46 |
| SamYaple | ok | 06:46 |
| *** achanda has quit IRC | 06:47 | |
| *** dims has joined #kolla | 06:50 | |
| sdake | my speculation from this test will be tht 103 will be rock oslid nd 101 will be flakey | 06:53 |
| SamYaple | how do you purpose we test that 103 is solid? | 06:53 |
| sdake | you sid yesteray 101 was rock olid | 06:53 |
| sdake | i dont kno how you cam eto that conclusionb | 06:53 |
| sdake | ut youcn come to the same with 103 no? | 06:53 |
| SamYaple | im asking how you want to test | 06:53 |
| SamYaple | i can | 06:53 |
| SamYaple | if you dont have an opinion | 06:53 |
| sdake | whtever you did yesterdayt o determine 101 was solid | 06:53 |
| SamYaple | ok | 06:54 |
| sdake | i want to eimiinate the boxes themselves | 06:54 |
| *** dims has quit IRC | 06:54 | |
| kjelly | Should I mark which base image is recommend in document ? | 07:02 |
| SamYaple | ubuntu :) its the only one where everything builds | 07:02 |
| SamYaple | and its half the size of centos.... | 07:02 |
| SamYaple | but no, dont mark it | 07:02 |
| kjelly | SamYaple: ok | 07:03 |
| openstackgerrit | Kuo-tung Kao proposed stackforge/kolla: update the image-building document https://review.openstack.org/219550 | 07:10 |
| kjelly | I found that *.pyc is not included in gitignore. Should I send patch for it? | 07:11 |
| *** inc0 has quit IRC | 07:14 | |
| sdake | plz | 07:15 |
| SamYaple | sdake: so there is no 500 service errors at all anymore | 07:16 |
| SamYaple | the issue wis with auth | 07:16 |
| SamYaple | whats _really_ wierd is glance is trying to auth with a token that doesnt exist in teh database at all | 07:17 |
| *** coolsvap_ has joined #kolla | 07:21 | |
| SamYaple | alreighty sdake | 07:22 |
| SamYaple | all fixed up | 07:22 |
| SamYaple | we can do a three node deploy now | 07:23 |
| *** akwasnie has joined #kolla | 07:24 | |
| *** achanda has joined #kolla | 07:24 | |
| *** shardy has joined #kolla | 07:25 | |
| *** coolsvap_ has quit IRC | 07:26 | |
| SamYaple | sdake: do we have a gate blueprint somewhere? | 07:26 |
| kjelly | Do I need bug id for changing gitignore ? | 07:31 |
| SamYaple | kjelly: yea we require bugs/blueprints for everything but doc changes | 07:31 |
| SamYaple | it gets a little annoying but at least it can always be traced back to a bug for more info | 07:32 |
| kjelly | SamYaple: ok | 07:32 |
| sdake | samyaple there is no gat eblueprint | 07:34 |
| sdake | samyaple gate-froms-ource is one | 07:34 |
| sdake | but no geneeralgate blupinrt | 07:34 |
| sdake | samyaple did you sort out why my node are busted? | 07:34 |
| sdake | nodes | 07:34 |
| SamYaple | yea youre time was all jacked up | 07:34 |
| SamYaple | your* | 07:34 |
| sdake | time is th ereason? | 07:35 |
| SamYaple | its proper UTC now | 07:35 |
| SamYaple | yea | 07:35 |
| sdake | nice | 07:35 |
| sdake | well that needs to go in the rtfm! | 07:35 |
| SamYaple | all three nodes on a glance image-list while loop, never failed | 07:35 |
| sdake | NICE | 07:35 |
| sdake | see, dude, worth the effort | 07:35 |
| sdake | root caused a problem which will certinaly hit other people | 07:35 |
| sdake | how did you figure it was the time | 07:35 |
| SamYaple | which is in the docs | 07:35 |
| *** shardy_ has joined #kolla | 07:36 | |
| sdake | i must have missed that ;-) | 07:36 |
| SamYaple | keystone token didn't appear in the database and so i did a regex look for it in the database and noticed the time was way way off from the latest time | 07:36 |
| sdake | well now we can put that the sytem will fail fantastically without correct time ;-) | 07:36 |
| SamYaple | which is pretty much openstack in general | 07:36 |
| SamYaple | its all time stamped | 07:36 |
| sdake | systems should depend upon monotonic time not wallcock time | 07:37 |
| sdake | but whatever, as long as it works im happy :) | 07:37 |
| *** shardy has quit IRC | 07:37 | |
| sdake | thanks for stickign with it samyaple | 07:37 |
| sdake | ok im going to do a 3 node deploy and play around | 07:37 |
| sdake | and smoke a cigar - a big one1 | 07:37 |
| SamYaple | its deployed three node right now | 07:37 |
| *** rajalokan has quit IRC | 07:37 | |
| SamYaple | still some other issues like oyu need to crank up your db connections | 07:38 |
| SamYaple | but thats because i maxed them out with the while loops | 07:38 |
| SamYaple | see sdake, isn't fixing the problem much better than kneejerking a 'solution'? Now we dont have to have that crappy source balancing | 07:38 |
| sdake | yup | 07:39 |
| sdake | i was undrer the mistaken belief tht others were also uffering the same problem | 07:39 |
| *** shardy_ has quit IRC | 07:41 | |
| SamYaple | so am i making a blueprint for gate testing? | 07:42 |
| sdake | which machines had bad time, all 3? | 07:42 |
| *** shardy has joined #kolla | 07:42 | |
| sdake | might as well | 07:42 |
| SamYaple | sdake: i think one may have been correct, idk | 07:42 |
| SamYaple | i stomped them all to UTC | 07:42 |
| SamYaple | sdake: are you cleaning up the environment right now? | 07:43 |
| sdake | yup | 07:43 |
| SamYaple | oh | 07:43 |
| sdake | sorry were you using/ | 07:43 |
| SamYaple | that would explain why shit stopped working | 07:43 |
| SamYaple | yea just testing some stuff | 07:43 |
| SamYaple | its cool | 07:44 |
| SamYaple | im out of your servers now | 07:44 |
| *** dims has joined #kolla | 07:44 | |
| openstackgerrit | Kuo-tung Kao proposed stackforge/kolla: update gitignore for python and .testrepository https://review.openstack.org/219587 | 07:45 |
| *** SourabhP has quit IRC | 07:45 | |
| SamYaple | vbel: does git currently support checking out a tag? | 07:46 |
| SamYaple | i know it does for a branch... | 07:46 |
| *** Haomeng|2 has joined #kolla | 07:47 | |
| openstackgerrit | Kuo-tung Kao proposed stackforge/kolla: update the image-building document https://review.openstack.org/219550 | 07:48 |
| *** dims has quit IRC | 07:49 | |
| *** Haomeng has quit IRC | 07:50 | |
| sdake | nowe if we cn just get heat working in t-8 hours | 07:51 |
| sdake | we will be golden ;) | 07:51 |
| *** achanda has quit IRC | 07:51 | |
| SamYaple | heat works | 07:51 |
| SamYaple | i though your patched cnetos-binary | 07:52 |
| SamYaple | thought* | 07:52 |
| sdake | nope | 07:52 |
| SamYaple | what happened to your heat patch | 07:53 |
| sdake | i abandoned it because it hit master | 07:53 |
| SamYaple | oh | 07:53 |
| SamYaple | i thought you would keep it in for binary until delorean got its act together | 07:53 |
| sdake | delorean rebuilds master every 2-3 days | 07:54 |
| sdake | but prior was getting resource not found on centos binary | 07:54 |
| *** inc0 has joined #kolla | 08:10 | |
| *** rajalokan has joined #kolla | 08:16 | |
| openstackgerrit | Merged stackforge/kolla: update gitignore for python and .testrepository https://review.openstack.org/219587 | 08:18 |
| *** jmccarthy has quit IRC | 08:20 | |
| sdake | Kennan around? | 08:20 |
| Kennan | hi sdake: | 08:20 |
| Kennan | just back | 08:21 |
| sdake | kennan are docs are not being published | 08:21 |
| sdake | but the docs publishing job hit the infra repo | 08:21 |
| sdake | any idea why | 08:21 |
| *** jmccarthy has joined #kolla | 08:21 | |
| Kennan | sdake: how do you check doc published or not ? | 08:21 |
| sdake | docs.openstack.org/developer/kolla | 08:21 |
| sdake | = not found | 08:22 |
| Kennan | sdake: let me check in infra channel, I am not sure what others needed | 08:23 |
| *** diga has joined #kolla | 08:24 | |
| *** weiyu_ has quit IRC | 08:25 | |
| openstackgerrit | Merged stackforge/kolla: update the image-building document https://review.openstack.org/219550 | 08:25 |
| vbel | SamYaple: yes, we can checkout by tag (reference) | 08:36 |
| SamYaple | so that would accecpt the tag '2015.1.1' ? | 08:38 |
| *** dims has joined #kolla | 08:38 | |
| vbel | yes, sure, we do it locally | 08:38 |
| SamYaple | awesome! | 08:39 |
| SamYaple | good work | 08:39 |
| openstackgerrit | Vladislav Belogrudov proposed stackforge/kolla: Source based builds should get novnc from tarball (git/url) https://review.openstack.org/219204 | 08:39 |
| vbel | E.g. [swift-base] | 08:39 |
| vbel | type = git | 08:39 |
| vbel | location = git://ca-git.us.oracle.com/openstack-swift | 08:39 |
| vbel | reference = 2.3.0 | 08:39 |
| vbel | dest_filename = swift.tar | 08:39 |
| SamYaple | vbel: cool :) | 08:39 |
| vbel | SamYaple: thanks :) | 08:39 |
| SamYaple | just fyi i removed the 'dest_filename' requirement | 08:39 |
| vbel | SamYaple: yes, I saw it. So it is generated from container name, isn't it? | 08:41 |
| SamYaple | yea | 08:41 |
| SamYaple | figure thats the cleanest solution | 08:41 |
| vbel | I believe we can make a separate enhancement for the kanaka-noVNC :) and leave the review above as in first edition | 08:42 |
| *** zhiwei has quit IRC | 08:42 | |
| SamYaple | sounds good | 08:43 |
| SamYaple | it will happen with teh spice console as well | 08:43 |
| SamYaple | which reminds me i need to add that | 08:43 |
| *** dims has quit IRC | 08:43 | |
| vbel | for such sophisticated archives we need a new option like dest_folder and ln -s {{ dest_folder }}-* /usr/share/novnc | 08:44 |
| *** Haomeng has joined #kolla | 08:44 | |
| SamYaple | well i mean its just the repo name | 08:44 |
| SamYaple | i dont know that we need al lthat | 08:45 |
| *** diga has quit IRC | 08:45 | |
| vbel | the repo name is hardcoded in Dockerfile.j2. I can have vbel-noVNC or similar tarballs | 08:46 |
| openstackgerrit | Sam Yaple proposed stackforge/kolla: Setup ubuntu build gate https://review.openstack.org/219243 | 08:46 |
| SamYaple | hey guys if i can get some acks on https://review.openstack.org/#/c/219243/ | 08:46 |
| SamYaple | gate tested and working | 08:46 |
| SamYaple | and now it has a blueprint coolsvap ;) | 08:46 |
| *** Haomeng|2 has quit IRC | 08:47 | |
| vbel | SamYaple: i see, yes, we can just do something like ln -s /{{ repository_name }}-* /usr/share/novnc | 08:50 |
| vbel | and make the repository name from location url | 08:50 |
| *** vbel has quit IRC | 09:02 | |
| *** dayou has quit IRC | 09:02 | |
| *** vbel has joined #kolla | 09:03 | |
| openstackgerrit | James McCarthy proposed stackforge/kolla: Swift WIP - add some missing services, also ansible configuration https://review.openstack.org/218795 | 09:03 |
| openstackgerrit | Merged stackforge/kolla: Setup ubuntu build gate https://review.openstack.org/219243 | 09:28 |
| openstackgerrit | Merged stackforge/kolla: Source based builds should get novnc from tarball (git/url) https://review.openstack.org/219204 | 09:32 |
| *** dims has joined #kolla | 09:32 | |
| *** dims has quit IRC | 09:37 | |
| openstackgerrit | Paul Bourke proposed stackforge/kolla: Copy config to /etc/kolla rather than symlink in Vagrantfile https://review.openstack.org/219624 | 09:46 |
| *** tfukushima has quit IRC | 09:52 | |
| *** tfukushima has joined #kolla | 09:54 | |
| jmccarthy | @Sam: you about ? re: conditionals in https://review.openstack.org/#/c/218795/ | 10:07 |
| jmccarthy | So for example, maybe create three new sections in site.yml, one new bool each to group the non-server services for account, container, and object ? | 10:07 |
| jmccarthy | (I'm hesitant to break them all out individually) | 10:07 |
| pbourke | I haven't seen any docs that indicate they're optional | 10:14 |
| inc0 | SamYaple, I'm about to lose my mind. ICMP response is showing on tap device, on qbr, but not on qvb | 10:17 |
| SamYaple | inc0: thats where security group rules are applied | 10:23 |
| SamYaple | change your security group rules | 10:23 |
| SamYaple | check* | 10:24 |
| SamYaple | qbr is secgroup rules, if they get filtered you wont see it on qvb | 10:24 |
| inc0 | yeah, on secgroups it looks ok | 10:25 |
| inc0 | first place I've checked | 10:25 |
| inc0 | however...I'll do iptables -L | 10:25 |
| inc0 | ...and there are no rules for my floating ip | 10:26 |
| inc0 | do I need to do anything besides port-update? | 10:26 |
| *** dims has joined #kolla | 10:26 | |
| *** diga has joined #kolla | 10:27 | |
| SamYaple | inc0: did you include your original ip and mac in the port update? | 10:28 |
| SamYaple | i dont think it appends | 10:28 |
| SamYaple | i think it stomps | 10:28 |
| SamYaple | neutron port-show <id> | 10:28 |
| SamYaple | see if the correct ip is in the allowed-address-apirs | 10:28 |
| openstackgerrit | Mick Thomspon proposed stackforge/kolla: Add ansible support for Murano https://review.openstack.org/218894 | 10:29 |
| inc0 | http://paste.openstack.org/show/440710/ SamYaple | 10:30 |
| inc0 | 254 is floating | 10:30 |
| SamYaple | comeon dude 10.10.10.254/24 | 10:30 |
| inc0 | hmm | 10:30 |
| SamYaple | network more :P | 10:30 |
| inc0 | just noticed it | 10:30 |
| SamYaple | 10.10.10.0/24 | 10:30 |
| inc0 | shouldn't that be ip not cidr anyway? | 10:31 |
| *** dims has quit IRC | 10:31 | |
| inc0 | anyway, that was problem | 10:31 |
| inc0 | thanks | 10:31 |
| SamYaple | no that is correct syntax | 10:31 |
| SamYaple | you could do 10.10.10.254/32 | 10:31 |
| inc0 | yeah, I guess | 10:32 |
| SamYaple | but you would also need to include teh original port ip address | 10:32 |
| inc0 | it works now | 10:32 |
| inc0 | on both addresses | 10:32 |
| SamYaple | neutron port-update 034afb8f-b62e-4058-9f38-8e510950eb35 --allowed-address-pairs type=dict list=true mac_address=ba:f5:3e:e3:67:ab,ip_address=192.168.10.10/32 mac_address=ba:f5:3e:e3:67:ab,ip_address=192.168.10.11/32 | 10:32 |
| SamYaple | that would do 2 ips ^^ | 10:32 |
| inc0 | oki, thanks | 10:33 |
| SamYaple | or you can do a range (like oyu have done) | 10:33 |
| inc0 | just as long as it's correct cird | 10:33 |
| inc0 | (btw it was copy from ip a add, that's why) | 10:33 |
| SamYaple | well it just has to be the network address | 10:33 |
| *** chandankumar has quit IRC | 10:33 | |
| SamYaple | i mean you could do 192.168.1.128/23 | 10:33 |
| SamYaple | i mean you could do 192.168.1.128/25* | 10:34 |
| inc0 | brb | 10:34 |
| inc0 | yeah, I know what I did wrong | 10:34 |
| SamYaple | cool gald its working | 10:34 |
| *** gsagie has joined #kolla | 10:46 | |
| *** chandankumar has joined #kolla | 10:48 | |
| openstackgerrit | Paul Bourke proposed stackforge/kolla: Always ensure correct ownership of /var/lib/rabbitmq https://review.openstack.org/218890 | 10:51 |
| *** gsagie has left #kolla | 10:55 | |
| SamYaple | pbourke: what is not a permissions issue | 11:01 |
| SamYaple | thats a clustering issue | 11:01 |
| pbourke | SamYaple: I'll try it with one node | 11:02 |
| SamYaple | https://bugs.launchpad.net/kolla/+bug/1480496 | 11:02 |
| openstack | Launchpad bug 1480496 in kolla "When starting rabbitmq multinode with Ansible the cluster fails to form" [Critical,Triaged] | 11:02 |
| SamYaple | pbourke: ^^ | 11:02 |
| pbourke | SamYaple: ah | 11:05 |
| pbourke | SamYaple: ok I guess I'll abandon this patch | 11:05 |
| SamYaple | did one node work | 11:06 |
| pbourke | SamYaple: it seems to work when deploying to the same node using localhost (allinone). but deploying one host, but separate from the deploy node it fails | 11:09 |
| pbourke | it seems to have been introduced back when we did the move to one data image | 11:10 |
| SamYaple | how do I setup vagrant to confirm this? | 11:12 |
| pbourke | ah hang on | 11:13 |
| pbourke | :/ | 11:13 |
| pbourke | sorry it doesnt work all in one either | 11:14 |
| pbourke | you have to go from scratch to make it happen | 11:14 |
| pbourke | cd kolla/vagrant | 11:14 |
| pbourke | vagrant up operator control01 control02 | 11:14 |
| pbourke | vagrant ssh operator | 11:14 |
| pbourke | then build using the command I put in the review to push to local registry | 11:15 |
| pbourke | and hack /etc/kolla/globals.yml appropriatly | 11:15 |
| pbourke | actually given it happens in all-in-one you dont need control01 / control02 | 11:16 |
| SamYaple | pbourke: what are you running? cnetos-binary? ubuntu-source? | 11:20 |
| pbourke | centos binary | 11:20 |
| pbourke | if it works on ubuntu then Im thinking its some difference in the way rabbitmq is packaged | 11:20 |
| SamYaple | does it happen with ubuntu source (I would test but i dont have vagarant setup and im neck deep in things) | 11:20 |
| SamYaple | if you dont have time to test ill test later | 11:21 |
| pbourke | I'll try | 11:21 |
| *** dims has joined #kolla | 11:21 | |
| *** rhallisey has joined #kolla | 11:24 | |
| *** dims has quit IRC | 11:26 | |
| *** pbourke has quit IRC | 11:36 | |
| *** pbourke has joined #kolla | 11:36 | |
| *** Haomeng|2 has joined #kolla | 12:07 | |
| *** Haomeng has quit IRC | 12:10 | |
| *** mordred has quit IRC | 12:27 | |
| *** inc0_ has joined #kolla | 12:31 | |
| *** inc0 has quit IRC | 12:32 | |
| *** inc0_ has quit IRC | 12:38 | |
| pbourke | SamYaple: same issue with ubuntu-source. there's a handful of setups but this one is all-in-one ubuntu-source | 12:45 |
| pbourke | SamYaple: strange you're not seeing it | 12:45 |
| SamYaple | weird indeed since i dont have it at all | 12:45 |
| SamYaple | can you pastebin it? | 12:45 |
| SamYaple | (the errors you are seeing) | 12:46 |
| pbourke | SamYaple: http://paste.openstack.org/show/441077/ | 12:46 |
| pbourke | jmccarthy got it this morning as well as inc0 yesterday I think | 12:47 |
| pbourke | but it doesn't seem to be consistent | 12:47 |
| *** dims has joined #kolla | 12:48 | |
| SamYaple | pbourke: can you docker inspect and verify the container is using voalumes_from? | 12:49 |
| pbourke | SamYaple: yeah just looking now, check this out: http://paste.openstack.org/show/441078/ | 12:49 |
| pbourke | (snapshot is the result of docker commit on the restarting container) | 12:50 |
| SamYaple | yea thats what i thought was happening | 12:50 |
| SamYaple | it wasnt actually doing a bootstrap | 12:50 |
| SamYaple | your patch would have prevented a cluster from forming because the erlang cookie wouldnt have been the same | 12:51 |
| SamYaple | but _why_ | 12:51 |
| pbourke | yeah looking back at the output its skipping the bootstrap step | 12:51 |
| pbourke | *ansible output | 12:51 |
| SamYaple | youll need to verify the bootstrap container actually uses the volumes from rabbitmq | 12:51 |
| SamYaple | oh | 12:51 |
| SamYaple | https://github.com/stackforge/kolla/blob/master/ansible/roles/rabbitmq/tasks/bootstrap.yml#L2 | 12:51 |
| SamYaple | ;) | 12:51 |
| SamYaple | are you removing the rabbitmq data container between retires? | 12:52 |
| pbourke | yeah was just looking at the same line :) | 12:52 |
| SamYaple | or just teh rabbitmq container? | 12:52 |
| SamYaple | i can improve that check | 12:53 |
| SamYaple | i know how to fix this :) | 12:53 |
| SamYaple | give be 15 minutes | 12:53 |
| pbourke | sure - or if you're short on time give me the summary and I can try fix | 12:53 |
| *** kevsi has joined #kolla | 12:54 | |
| openstackgerrit | Sam Yaple proposed stackforge/kolla: Fix rabbitmq perms https://review.openstack.org/219681 | 12:55 |
| SamYaple | test that would you | 12:55 |
| *** diogogmt has joined #kolla | 12:56 | |
| pbourke | didnt even notice it was doing inspect | 12:56 |
| *** diogogmt has quit IRC | 12:59 | |
| *** diga has quit IRC | 13:01 | |
| pbourke | SamYaple: seems to do the trick | 13:01 |
| pbourke | SamYaple: can you tag it with bug #1490602 and I'll +2 | 13:02 |
| openstack | bug 1490602 in kolla "rabbitmq frequently fails to start due to permissions error" [Critical,In progress] https://launchpad.net/bugs/1490602 - Assigned to Paul Bourke (pauldbourke) | 13:02 |
| SamYaple | pbourke: im guessing you reran the playbooks after a bad bootstrap without cleaning up the rabbitmq_data container | 13:02 |
| SamYaple | pbourke: not at the moment, yo ucan update the commit message from gerrit though | 13:02 |
| SamYaple | (you can +2 after that) | 13:02 |
| pbourke | SamYaple: k if you dont mind | 13:02 |
| pbourke | SamYaple: thanks for help! | 13:02 |
| *** inc0 has joined #kolla | 13:03 | |
| *** diogogmt has joined #kolla | 13:04 | |
| SamYaple | yea no problem. it was a procedural difference between what we were doing | 13:05 |
| *** diogogmt has quit IRC | 13:08 | |
| SamYaple | pbourke: can you verify that the boostrap container doesnt run a second time? | 13:10 |
| pbourke | SamYaple: will do | 13:10 |
| *** dims_ has joined #kolla | 13:17 | |
| *** dims has quit IRC | 13:21 | |
| vbel | docker rm -f $(docker ps -aq) should help :) | 13:22 |
| vbel | always delete data container if bootstrap failed | 13:23 |
| kevsi | Hi folks, I have a question re the need for sudo/root in Kolla Ansible (as we will get questions on this from our security folks in the near future). | 13:23 |
| kevsi | (I know this has come up a few times before) | 13:23 |
| kevsi | My current understanding (paraphrasing previous Kolla IRC comments): | 13:23 |
| kevsi | - Ansible can be run as non-root ( e.g. $ ./tools/kolla-ansible deploy) but it must ssh into the controller/Compute/Network Nodes as "root" to allow required Host Networking (and maybe more). | 13:24 |
| SamYaple | all-in-one deploys must run ansible as root | 13:24 |
| kevsi | However, the Docker Daemon on the Controller/Compute/Network nodes will be running as the root user, so is this not enough on its own to take care of Host Networking ? | 13:24 |
| SamYaple | multinode deploys ansible itself is not run as root, but currently it is required that it ssh directly to root | 13:24 |
| kevsi | SamYaple: thanks. I am concerned with multinode only | 13:25 |
| SamYaple | really its just the work hasnt been dont to use sudo yet | 13:25 |
| SamYaple | there is no reason ansible need to ssh directly to root | 13:25 |
| SamYaple | but it does need to run some root stuff (like tweak sysctl for some things) | 13:25 |
| kevsi | so if I understand correctly: | 13:26 |
| kevsi | for multinode, Ansible would not need to be run as root, would not need to ssh-in-as-root for most things, and would not need to sudo for most things. | 13:27 |
| kevsi | (but it would need to ssh-in-as-root/sudo for some things - like sysctl changes as you noted). | 13:28 |
| kevsi | but for basic running of the Docker Containers, it does not need to ssh-in-as-root or sudo per-se ? | 13:28 |
| SamYaple | kevsi: id have to check | 13:28 |
| SamYaple | i will say this, it uses docker-py | 13:28 |
| SamYaple | so i dont know if you need root to talk to the socket via api or not | 13:29 |
| SamYaple | i would _say_ not but idk | 13:29 |
| SamYaple | id have to check | 13:29 |
| *** jtriley has joined #kolla | 13:30 | |
| openstackgerrit | Paul Bourke proposed stackforge/kolla: Fix rabbitmq perms https://review.openstack.org/219681 | 13:31 |
| kevsi | SamYaple: thanks for that. we can do some investigation here on our side. | 13:31 |
| SamYaple | kevsi: please let us know :) | 13:31 |
| kevsi | will do. | 13:31 |
| SamYaple | and if oyu want to start the work for converting to ansible sudo stuff, im all for that | 13:31 |
| SamYaple | its a time thing for me | 13:31 |
| *** inc0 has quit IRC | 13:31 | |
| kevsi | No worries. We'll have a chat here about it over the next day or so and let you know. | 13:32 |
| sdake | morning | 13:38 |
| SamYaple | morning sdake | 13:38 |
| sdake | samyaple re sduo required for sysctl, we can just do that via contiainer | 13:38 |
| sdake | docker-py doesn't require root, it requires access to the docker sockt which means the duser needs to be in the docker gruop | 13:38 |
| sdake | there is no good reason I see thatansible has to be run as root currently | 13:38 |
| sdake | other then possibly to write to /usr/share/kolla/templates | 13:38 |
| sdake | kevsi currently kolla' sansible code creates a directory /usrshare/kolla/template, this requires root | 13:39 |
| sdake | samyaple yo dude | 13:39 |
| SamYaple | sysctl needs to be permanant and i dont like what you are suggesting about using the containers root users to modify the host system | 13:39 |
| kevsi | sdake: Thanks | 13:39 |
| SamYaple | sdake: while that does require root, that is a configurable paramater kevsi sdake | 13:40 |
| SamYaple | you can put that anywhere you like | 13:40 |
| kevsi | OK. | 13:40 |
| sdake | also you oculd make tht sudo and just give mkdir opreational privilegdge to the user | 13:41 |
| sdake | passwordless sudo for mkir command only | 13:41 |
| sdake | but anyway that is the ony reason i can see for rquiring root | 13:41 |
| sdake | kevsi basically we run kolla-ansible under root atm to simplify dev, but its time before rc1 to fix that | 13:42 |
| SamYaple | we run everything everywhere as root | 13:43 |
| SamYaple | we can fix it everywhere or it doesnt really matter if we fix it in a single location | 13:43 |
| sdake | yes but we dont need to run ansible as root | 13:43 |
| SamYaple | thats required for all-in-one only | 13:43 |
| SamYaple | and that is required | 13:43 |
| sdake | may be for aio | 13:44 |
| SamYaple | we havent needed to run the ansible deploy as root for multinode in a month | 13:44 |
| sdake | isn't that what kevii asked? :) | 13:44 |
| kevsi | SamYaple: Thanks. So ye do also see the current use of having Ansible ssh-as-root/sudo as a large security concern, or ye think we are being overly-paranoid (maybe we're overlooking some existing protection mechanism) ? | 13:44 |
| SamYaple | sdake: yea that was long ago answered. thats not what he is talking about | 13:45 |
| SamYaple | kevsi: no i agree its a security problem, but we do what we can | 13:45 |
| SamYaple | i want to get it fixed for RC1 | 13:46 |
| kevsi | sdake: if I understand the situation correctly, I'm more concerned with the need to have Ansible ssh-as-root/sudo on the ControllerNode/ComputeNode...(currently we don't run the Ansible command itself as root - we use multinode) | 13:46 |
| sdake | kevsi my theory on security is there is no reson to give permisisons to a process if they are not needed | 13:46 |
| *** mickt has joined #kolla | 13:46 | |
| kevsi | Thanks folks. OK, so it looks like we are thinking along the same lines. Just a matter of time/resources :-) | 13:47 |
| kevsi | we will investigate this here then, as it will be a big issue for us (when it comes to a real-world deployment). | 13:47 |
| SamYaple | kevsi: i 100% agree with you :) | 13:48 |
| kevsi | very good :-) | 13:49 |
| sdake | kevsi are you deploying kolla multinode? | 13:49 |
| sdake | or doing an eval? | 13:50 |
| kevsi | well, we are currently testing the multinode here at Oracle. | 13:52 |
| kevsi | with the intention of deploying it within each of the Fortune 500 by end of the year :-) | 13:53 |
| rhallisey | cool | 13:54 |
| sdake | oh right oracel dude | 13:57 |
| sdake | hey rhallisey | 13:57 |
| rhallisey | sdake, hey | 13:58 |
| *** rajalokan has quit IRC | 13:59 | |
| *** inc0 has joined #kolla | 14:06 | |
| *** jtriley has quit IRC | 14:13 | |
| *** mickt has quit IRC | 14:16 | |
| inc0 | guys, anyone have ansible script to prepare ubuntu for kolla? | 14:17 |
| inc0 | install docker and such | 14:17 |
| *** sdake has quit IRC | 14:28 | |
| *** sdake has joined #kolla | 14:32 | |
| *** inc0 has quit IRC | 14:32 | |
| *** jtriley has joined #kolla | 14:33 | |
| openstackgerrit | James McCarthy proposed stackforge/kolla: Swift WIP - add some missing services, also ansible configuration https://review.openstack.org/218795 | 14:34 |
| sdake | we should probably have some seprate ansible code to bring up the machines for first time | 14:35 |
| SamYaple | sdake: yea an optional off to the side one | 14:38 |
| SamYaple | the problem is it requires pip to isntall docker-py | 14:38 |
| SamYaple | to get the newer version | 14:38 |
| sdake | right | 14:38 |
| SamYaple | that requires build-tools | 14:38 |
| SamYaple | i say problem, its just extra crap on the system | 14:38 |
| sdake | we could make repos for docker-py for ubuntu and centos | 14:38 |
| SamYaple | yea we could, but then youve got a seperate repo and i dont want to maintain that | 14:39 |
| SamYaple | its kinda out of scope but i get your point | 14:39 |
| sdake | a repo is a simple solution until the distros catch up | 14:39 |
| SamYaple | well no the simple solution is install the build tools and pip | 14:40 |
| SamYaple | a repo is not the _simple_ solution | 14:40 |
| sdake | tidy then ;) | 14:40 |
| SamYaple | agreed | 14:40 |
| sdake | re off to the side one, what would it be called? | 14:40 |
| SamYaple | bootstrap | 14:41 |
| SamYaple | host_setup | 14:41 |
| SamYaple | kickoff | 14:41 |
| SamYaple | onetime | 14:41 |
| SamYaple | idk | 14:41 |
| sdake | does it end up in the main ansible dir? | 14:41 |
| SamYaple | it can sit beside site.yml | 14:41 |
| *** diogogmt has joined #kolla | 14:42 | |
| openstackgerrit | Vladislav Belogrudov proposed stackforge/kolla: Allow arbitrary source archive name to be imported into Dockerfile https://review.openstack.org/219723 | 14:43 |
| SamYaple | vbel: does that actually extract the top level into that folder? | 14:44 |
| vbel | SamYaple: I think I have found very easy solution to import any archive / clone into any container. Yes | 14:44 |
| SamYaple | or does it have /nova-novncproxy/kanaka-noVNC- | 14:44 |
| vbel | so we could generalize this to ADD source.tar /source | 14:44 |
| vbel | RUN ls -l /source/* /keystone | 14:45 |
| SamYaple | hmm i do like the source dir name to be there for versions | 14:45 |
| SamYaple | but i like that solution too | 14:45 |
| SamYaple | let me think on it | 14:45 |
| SamYaple | i think thats a good idea | 14:45 |
| sdake | yo guys do mea favor and don't +a any patches for a few hours | 14:45 |
| sdake | thanks! | 14:45 |
| SamYaple | but ive had a long day, let me regroup tomorrow | 14:45 |
| vbel | this solves problems with kanaka-noVNC or vbel-noVNC or hack-my-vnc-and-go archives / clones | 14:47 |
| vbel | just removing hardcoded strings we expect from archives. We still could have ADD nova-novncproxy.tar /nova-novncproxy | 14:48 |
| *** diogogmt has quit IRC | 14:54 | |
| *** diogogmt has joined #kolla | 14:55 | |
| *** chandankumar has left #kolla | 14:57 | |
| *** blahRus has joined #kolla | 14:57 | |
| *** unicell1 has joined #kolla | 15:04 | |
| openstackgerrit | Merged stackforge/kolla: Fix rabbitmq perms https://review.openstack.org/219681 | 15:05 |
| *** unicell has quit IRC | 15:06 | |
| openstackgerrit | Merged stackforge/kolla: Copy config to /etc/kolla rather than symlink in Vagrantfile https://review.openstack.org/219624 | 15:10 |
| *** dims_ has quit IRC | 15:12 | |
| sdake | why is rabbitmq failing on startup | 15:18 |
| sdake | vbel pbourke any ideas? | 15:18 |
| pbourke | sdake: multinode? | 15:18 |
| sdake | yup | 15:18 |
| pbourke | there's another bug with clustering | 15:19 |
| pbourke | let me find it | 15:19 |
| vbel | what does container tell? | 15:19 |
| pbourke | sdake: https://bugs.launchpad.net/kolla/+bug/1480496 | 15:19 |
| openstack | Launchpad bug 1480496 in kolla "When starting rabbitmq multinode with Ansible the cluster fails to form" [Critical,Triaged] | 15:19 |
| sdake | well now in confirmed state | 15:20 |
| * sdake groans | 15:20 | |
| pbourke | :( | 15:20 |
| vbel | hmm, I could not get this issue. But our rabbit is much newer :) | 15:21 |
| sdake | ERROR (InternalServerError): An unexpected error prevented the server from fulfilling your request: (OperationalError) (1040, 'Too many connections') None None (Disable debug mode to suppress these details.) (HTTP 500) (Request-ID: req-3d2c7558-c3b1-41c4-8cc6-76444a1d9dfc) | 15:21 |
| sdake | anyone seen that problem prior? | 15:21 |
| vbel | is it because of haproxy probes? | 15:21 |
| sdake | no idea what causes it | 15:22 |
| sdake | i deplyed a 3 node | 15:22 |
| sdake | rabbitmq was crashed so i started it fresh | 15:22 |
| sdake | then i ran init-runocne and created 5 nova instances | 15:22 |
| pbourke | vbel: everyone should use oraclelinux, only answer ;) | 15:22 |
| sdake | the 2nd nova instance failed | 15:23 |
| SamYaple | sdake: yea youll always hit that | 15:23 |
| SamYaple | just start teh rabbit nodes again | 15:23 |
| SamYaple | i can throw up a 50cents fix for it | 15:23 |
| SamYaple | its ugly as dirt | 15:23 |
| sdake | yes lets get something temporary | 15:23 |
| sdake | but what about the operational error | 15:23 |
| sdake | this was after retarting rabbitmq | 15:23 |
| SamYaple | that you can control via config options | 15:24 |
| SamYaple | aka how ot into openstack | 15:24 |
| sdake | this is the first time i've ever had multinode orking - looks like our defdaults need tuning :) | 15:24 |
| SamYaple | well these are openstack defaults | 15:25 |
| sdake | SamYaple if you could get a quick fixx up for the rabbitmq crash that would be great - so i can tag the releae | 15:25 |
| SamYaple | that would be for the operator to do the tuning | 15:25 |
| sdake | what is the specific default | 15:25 |
| SamYaple | max_pool_size = 10 | 15:25 |
| SamYaple | in the [database] section | 15:25 |
| SamYaple | if you change 'max_pool_size = 200' in your /etc/kolla/config/database.conf | 15:26 |
| SamYaple | (in the [database] section) | 15:26 |
| sdake | idea is out of the box defaults for kolla should result in working cluster | 15:26 |
| sdake | i wouldn't define operational error as working ;) | 15:26 |
| SamYaple | sure, but were not hard coding that value in the config | 15:26 |
| SamYaple | if you want to start dumping it in the database.conf, id be ok with that i suppose | 15:27 |
| sdake | what does that max_pool_size option do? | 15:27 |
| SamYaple | im still at work on the phone, cant talk about that stuff right now | 15:28 |
| SamYaple | i would suggest googling | 15:28 |
| SamYaple | thats the option you want to tweak | 15:28 |
| sdake | ok try to get the fix up for rabbitmq today plz | 15:28 |
| SamYaple | im warning oyu, its awful | 15:30 |
| SamYaple | its the reason i havent submitted it as a stopgap | 15:30 |
| pbourke | is it making oraclelinux the default? :p | 15:30 |
| SamYaple | how did you know?! | 15:31 |
| *** absubram has joined #kolla | 15:31 | |
| vbel | :)))) | 15:32 |
| openstackgerrit | Sam Yaple proposed stackforge/kolla: Fix rabbitmq part 2 https://review.openstack.org/219746 | 15:33 |
| SamYaple | thats all i have | 15:33 |
| SamYaple | that needs testing | 15:33 |
| SamYaple | people can submit a patch on top of it, you should see my train of thought | 15:34 |
| SamYaple | if it works merge, if not submit on top | 15:34 |
| openstackgerrit | Jeff Peeler proposed stackforge/kolla: WIP: Add Ansible support for Ironic https://review.openstack.org/219747 | 15:34 |
| *** britthouser has quit IRC | 15:34 | |
| sdake | wtb resolution for https://bugs.launchpad.net/kolla/+bug/1491477' | 15:35 |
| openstack | Launchpad bug 1491477 in kolla "default max pool size does not render a functional cluster" [High,New] | 15:35 |
| vbel | ^^^ mariadb? | 15:37 |
| sdake | sam says that should fix it | 15:38 |
| sdake | although i tried with those evalues with no good result | 15:38 |
| sdake | i am starting a fresh | 15:38 |
| vbel | https://bugs.launchpad.net/mos/+bug/1475433 - sounds similar? | 15:40 |
| openstack | Launchpad bug 1475433 in Mirantis OpenStack 7.0.x "MySQL connection limit should be increased" [High,Fix released] - Assigned to Denis Egorenko (degorenko) | 15:40 |
| vbel | 200 nodes :) | 15:41 |
| sdake | similar but 4096 is the default there | 15:48 |
| sdake | wow samyaple that patch is dirty like zebra! | 15:50 |
| *** dwalsh has joined #kolla | 15:50 | |
| *** dims has joined #kolla | 15:51 | |
| SamYaple | i warned you | 15:52 |
| pbourke | sdake: Im seeing it I can test it with a newer rabbitmq | 15:52 |
| sdake | pbourke that patch will likely work on a newer if it works on an older | 15:53 |
| sdake | it would be nie if wecould get to a solution that doesnt involve upgrading rabbitmq ;-) | 15:53 |
| pbourke | sdake: no I think that patch should not be needed | 15:53 |
| sdake | pbourke how do you intend to get rabbitmq to work on distros that ship older versions then? | 16:01 |
| sdake | keepign in mind something needs to be done today so i can tag | 16:02 |
| sdake | so we can get on with our rc1 development | 16:02 |
| pbourke | sdake: I was hoping to find an external rpm source that provided new packages | 16:02 |
| *** unicell1 has quit IRC | 16:02 | |
| *** inc0 has joined #kolla | 16:02 | |
| *** dims has quit IRC | 16:02 | |
| pbourke | sdake: but I guess times a short for that today | 16:02 |
| *** vinkman has joined #kolla | 16:03 | |
| *** dims has joined #kolla | 16:03 | |
| sdake | i dont ike the third party import repos we have atm | 16:03 |
| sdake | i really dont want to add to the problem | 16:03 |
| *** dims_ has joined #kolla | 16:04 | |
| inc0 | what's the issue? | 16:04 |
| sdake | rabbitmq crashes on startup | 16:04 |
| sdake | apparently because its too old | 16:04 |
| inc0 | well...it didn't for me | 16:05 |
| inc0 | today | 16:05 |
| inc0 | today's build | 16:05 |
| sdake | i just built 1 hour ago fresh centos | 16:05 |
| inc0 | sdake, too old kernel problem? | 16:05 |
| inc0 | I had to upgrade kernel to 3.19 for it to not fail - ubuntu on ubuntu | 16:06 |
| sdake | no rabbitmq and its illion dependencies | 16:06 |
| inc0 | ah ok | 16:06 |
| *** dims has quit IRC | 16:07 | |
| inc0 | this release is problematic one | 16:08 |
| sdake | we are doing a bunch of new things all at once | 16:09 |
| sdake | likely to be problems with that scenario | 16:09 |
| inc0 | yeah | 16:11 |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Replace config-external with a JSON file https://review.openstack.org/216874 | 16:12 |
| vbel | sometimes I think progress is actually bad thing :) | 16:14 |
| inc0 | well, use this until someone actually deploy kolla on prod | 16:14 |
| inc0 | changes after that will be much harder | 16:14 |
| *** devlaps has joined #kolla | 16:16 | |
| openstackgerrit | James McCarthy proposed stackforge/kolla: Swift WIP - add some missing services, also ansible configuration https://review.openstack.org/218795 | 16:20 |
| sdake | meeting in t-10 minutes | 16:22 |
| sdake | well we can launch 1 nova instance before everything falls to shit | 16:25 |
| sdake | shp it ;) | 16:25 |
| sdake | i have heard from multiple sources nova-scheduler may not be run in ha mode | 16:26 |
| sdake | as in it can only be one on one controller | 16:26 |
| vbel | I thought it was state-less (running via rabbit so it is ha without haproxy) | 16:28 |
| sdake | can a cr please ack https://review.openstack.org/219746 | 16:28 |
| sdake | i have verified it | 16:29 |
| sdake | it is unclear if it works because nova launch fails still | 16:29 |
| vbel | 'shpt' is a good release name :) | 16:29 |
| sdake | meeting time folks :) | 16:31 |
| SamYaple | pbourke: rabbitmq actually provides a repo | 16:33 |
| SamYaple | fyi | 16:33 |
| *** akwasnie has quit IRC | 16:39 | |
| *** absubram has quit IRC | 16:43 | |
| *** unicell has joined #kolla | 16:47 | |
| *** akwasnie has joined #kolla | 16:47 | |
| openstackgerrit | Merged stackforge/kolla: Fix rabbitmq part 2 https://review.openstack.org/219746 | 16:53 |
| *** clayton has joined #kolla | 16:56 | |
| clayton | just started looking at Kolla since I didn't realize the option was added for building from source. Is there or is there planned to be any way to provide a requirements.txt file when building the container? | 16:56 |
| sdake | clayton we are in a team meeting atm we will be back to you ok ? :) | 16:57 |
| clayton | np, thanks | 16:57 |
| *** pbourke-mobile has joined #kolla | 17:04 | |
| *** rajalokan has joined #kolla | 17:10 | |
| *** pbourke-mobile_ has joined #kolla | 17:11 | |
| *** pbourke-mobile has quit IRC | 17:12 | |
| *** pbourke-mobile_ has quit IRC | 17:19 | |
| jpeeler | inc0: did you see my ironic ansible review? | 17:19 |
| inc0 | hold on | 17:19 |
| sdake | clayton why would you want to specify a custom requirements.txt file? | 17:19 |
| jpeeler | i think the only thing left is to create a separate nova container, but it's pretty close | 17:19 |
| jpeeler | thought you were really interested in it | 17:20 |
| inc0 | yeah I was | 17:20 |
| inc0 | and I still am | 17:20 |
| clayton | few reasons, one is for example, we had a designate sink we wrote that we want to include inside the container so it can be used | 17:20 |
| inc0 | I'll give it a try tomorrow maybe, thanks for heads up! | 17:20 |
| clayton | but also because we generally freeze all the requirements when building the container | 17:20 |
| clayton | that way if we rebuild it later we'll have reproducible results | 17:20 |
| inc0 | I'm off guys, see ya | 17:20 |
| sdake | what is a designate sink, some kind of plugin? | 17:20 |
| sdake | later inc0 | 17:21 |
| clayton | designate sink is the designate service that allows you take actions based on neutron or nova events | 17:21 |
| sdake | clayton we do have some ideas for dealing with plugins for neutron, cinder, and horizon | 17:21 |
| clayton | we also have a custom keystone plugin for auth | 17:21 |
| coolsvap | designate sink is a service in designate | 17:21 |
| sdake | we have a prebuild sink service | 17:21 |
| clayton | right, but anyone using sink would write a plugin, what comes with it is intended to be an example, nothing more | 17:21 |
| sdake | ok so plugin, yes we have plans to handle that | 17:22 |
| sdake | and the plugins would yes install the requierments.txt | 17:22 |
| sdake | it will not happen for liberty i fear | 17:22 |
| sdake | unless pbourke pulls off a miracle ;-) | 17:22 |
| clayton | it looks like it'd be fairly easy to add a requirements option to build.ini and copy that over the one in the project if provided when generating the tarball | 17:22 |
| sdake | we ar ejust going to use the rquirements.txt in the tarball clayton | 17:23 |
| sdake | no reason to specify a unique one | 17:23 |
| clayton | I disagree. For example, aside from the plugin issue, we want to specify frozen requirements so that if we rebuild the container in the future we get exactly the same results | 17:23 |
| sdake | you can do that by psecifying a frozen tarball | 17:24 |
| sdake | the tarball containing the rquirement.ttxt that are frozen | 17:24 |
| clayton | I was hoping not to build and maintain tarballs when kolla can already do that :) | 17:24 |
| *** achanda has joined #kolla | 17:24 | |
| sdake | you ahve to maintain that file somewhere | 17:24 |
| sdake | if its in a git repo, kolla can pull from git as well | 17:24 |
| clayton | maintaining a requirement.txt and maintaining a mechanism to turn that into a tarball isnt' the same thing | 17:25 |
| clayton | nod, thought about that, but that requires us to maintain a fork of the repo and overwrite the upsteam requirements.txt | 17:25 |
| sdake | you maintain the requirements.txt alongside the code it is meant to install | 17:25 |
| *** inc0 has quit IRC | 17:25 | |
| sdake | clayton hate to be argumentantive,but ins't htat preisely what you want kolla to do :) | 17:25 |
| clayton | but it already exists in kolla :) | 17:26 |
| sdake | clayton there is a blueprint on this | 17:26 |
| sdake | please add your ideas to it | 17:26 |
| clayton | ah, ok. link? | 17:26 |
| clayton | we're maintaining our own containers right now, but we'd prefer to use kolla if we can fit it into how we do things. I'm coming at this from the standpoint of how we already do things, so this is a bit foreign | 17:27 |
| sdake | clayton https://blueprints.launchpad.net/kolla/+spec/neutron-third-party-plugins | 17:27 |
| sdake | fwiw I am not keen on adding options or other cruft to build.ini or build.py without a good good reason | 17:28 |
| clayton | sdake: the way we handle plugins now is that we just add them as a git+https source in the requirements.txt | 17:28 |
| sdake | plugins are a good reason | 17:29 |
| sdake | i htinkk itremains to be seen what the best way to tackle plugins is | 17:29 |
| sdake | whaty ou propose, a unique requirements.txt file sounds clunky - where do those requirements.txt files come from? | 17:30 |
| sdake | some git repo? | 17:30 |
| sdake | pbourke is your man to talk to clayton | 17:30 |
| clayton | well, if I were going off and solve this problem in a vacuum, I'd just make it so that build.ini took a path to a file to copy into the tarball on a per project basis | 17:31 |
| sdake | yes but you can see how that doesn't work well if someone doesnt have that requirements.txt file :) | 17:31 |
| sdake | what your asking for is a rquirements.txt override | 17:31 |
| sdake | again i dont seee value of freezing requirements | 17:31 |
| clayton | I assumed build.ini was intended to be customized locally | 17:31 |
| sdake | customized locally if desired but should work out of the box | 17:32 |
| clayton | sdake: if I rebuild a container, I don't want to suddently get newer versions of a half dozen libraries if I'm making a trivial change. I want to upgrade libraries regularly, but intentionally. | 17:32 |
| *** absubram has joined #kolla | 17:32 | |
| clayton | sdake: nod, I'd expect the out of the box build.ini not to have any value for the new option I was proposing | 17:32 |
| *** dims_ has quit IRC | 17:33 | |
| clayton | my expecation would be that most people building from source would customize build.ini | 17:33 |
| sdake | yes what you are talking about has nothign to do with plugins and has to do with stable build requirements | 17:33 |
| sdake | i can almost see that use cae | 17:34 |
| sdake | but our plate is full for liberty-3 ;) | 17:34 |
| sdake | so unlikely to be addressed | 17:35 |
| clayton | sdake: I may be conflating them since we've used the same approach to solve both problems | 17:35 |
| clayton | since half the reason we're using containers is to get away from upstream packages anyway, I tend to think of it in terms of building from source. | 17:35 |
| clayton | not saying that's valid, just explaining :) | 17:35 |
| *** dims has joined #kolla | 17:35 | |
| sdake | clayton i think what would make sense is for you to file a blueprint | 17:36 |
| sdake | the descriptoin of the blueprint might be "allow for override of requirements.txt per build source" | 17:36 |
| sdake | the text field of which would describe your proposed solution | 17:37 |
| sdake | and i'll put in discussion phsae and we can discuss during mitaka | 17:37 |
| sdake | clayton make sense? | 17:37 |
| clayton | yeap | 17:38 |
| sdake | the argument to be made in he blueprint is that you want consistent library dependencies pinned at a specific version | 17:38 |
| *** rajalokan has quit IRC | 17:45 | |
| *** tfukushima has quit IRC | 17:45 | |
| clayton | thanks | 17:49 |
| *** loth has joined #kolla | 17:57 | |
| *** chandankumar has joined #kolla | 18:04 | |
| *** jtriley has quit IRC | 18:06 | |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Replace config-external with a JSON file https://review.openstack.org/216874 | 18:09 |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Replace config-external with a JSON file for Galera https://review.openstack.org/217880 | 18:09 |
| *** sdake_ has joined #kolla | 18:14 | |
| *** sdake has quit IRC | 18:17 | |
| *** sdake has joined #kolla | 18:21 | |
| *** dwalsh has quit IRC | 18:23 | |
| jpeeler | sdake: i may not know what i'm talking about, but it's looking like there's not a way to differential requests to the proper scheduler based on the necessary compute driver | 18:24 |
| jpeeler | so i'm not sure if trying to spin up nova services configured for ironic alongside nova configured for libvirt will work | 18:24 |
| sdake | jpeeler cant' you just run a different api server on a different port? | 18:24 |
| sdake | try joining the ironic channel and asking :) | 18:25 |
| *** sdake_ has quit IRC | 18:25 | |
| jpeeler | i'm reading that the containers have to be separated by availability zones, but i guess i can try that | 18:25 |
| jpeeler | the primary use case for using ironic though, would be undercloud i would think | 18:25 |
| sdake | ya maybe ironic needs to be depdloyed first on an aio deploy | 18:26 |
| sdake | and second the real cloud gets deployed | 18:26 |
| sdake | sorting out how to integrate ironic and ansible in this way woud lbe pretty helpful ;) | 18:26 |
| jpeeler | i can ask if my understanding is correct though | 18:26 |
| jpeeler | if it is, you'd just toggle the enable_ironic flag | 18:27 |
| jpeeler | sdake: is the plan to keep this channel name the same after openstack namespace change? | 18:30 |
| sdake | yes | 18:30 |
| *** jasonsb has quit IRC | 18:32 | |
| jpeeler | only a handful that doesn't have that prefix, adding kolla to the list here https://wiki.openstack.org/wiki/IRC | 18:34 |
| *** sdake_ has joined #kolla | 18:41 | |
| sdake_ | jpeeler ya blame rhallisey for that one ;) | 18:42 |
| sdake_ | btw, release tagged, email sent | 18:42 |
| sdake_ | nice work folks | 18:43 |
| rhallisey | ya | 18:43 |
| sdake_ | now for rc1 ;) | 18:43 |
| *** sdake has quit IRC | 18:45 | |
| sdake_ | so folks looks like we get 3 fishbowl 5 workroom sessions and a half day meetup on friday | 18:51 |
| sdake_ | although not yet set in stone | 18:52 |
| sdake_ | typically meeetup is 2 hours of stuff that didn't get finished during the regular workroom and fishbowls plus 2 hours of drinking on the ptl whoever that may be ;) | 18:53 |
| *** coolsvap_ has joined #kolla | 19:09 | |
| *** chandankumar has quit IRC | 19:39 | |
| *** jasonsb_ has joined #kolla | 19:41 | |
| *** coolsvap_ has quit IRC | 19:45 | |
| *** absubram has quit IRC | 19:54 | |
| *** shardy has quit IRC | 20:00 | |
| *** sdake_ is now known as sdake | 20:15 | |
| sdake | rhallisey ar you the person that added the tripleo roadmap? | 20:18 |
| sdake | or was that inc0 | 20:18 |
| *** dwalsh has joined #kolla | 20:24 | |
| *** absubram has joined #kolla | 20:33 | |
| rhallisey | sdake, inc0 | 20:39 |
| sdake | ok well i am thinking this will end in the half day contributor meetup sessions | 20:40 |
| sdake | probably first on the agenda to work well with getting tripleo cats in the room | 20:40 |
| rhallisey | sure | 20:40 |
| rhallisey | wfm | 20:40 |
| *** achanda has quit IRC | 21:08 | |
| *** unicell has quit IRC | 21:14 | |
| *** unicell has joined #kolla | 21:14 | |
| *** dwalsh has quit IRC | 21:17 | |
| *** dims has quit IRC | 21:19 | |
| *** dims has joined #kolla | 21:19 | |
| *** dims has quit IRC | 21:24 | |
| *** achanda has joined #kolla | 21:37 | |
| *** absubram has quit IRC | 21:44 | |
| *** achanda has quit IRC | 21:47 | |
| *** achanda has joined #kolla | 21:54 | |
| *** banix has joined #kolla | 22:03 | |
| *** diogogmt has quit IRC | 22:19 | |
| *** achanda has quit IRC | 22:24 | |
| *** achanda_ has joined #kolla | 22:24 | |
| *** dwalsh has joined #kolla | 22:35 | |
| *** diogogmt has joined #kolla | 22:40 | |
| *** unicell has quit IRC | 22:43 | |
| *** unicell has joined #kolla | 22:43 | |
| *** unicell has quit IRC | 22:44 | |
| *** unicell has joined #kolla | 22:44 | |
| *** diogogmt has quit IRC | 22:55 | |
| *** dwalsh has quit IRC | 23:02 | |
| *** banix has quit IRC | 23:08 | |
| *** banix has joined #kolla | 23:12 | |
| *** achanda_ has quit IRC | 23:12 | |
| *** sdake has quit IRC | 23:15 | |
| *** akwasnie has quit IRC | 23:19 | |
| *** achanda has joined #kolla | 23:21 | |
| *** sdake has joined #kolla | 23:27 | |
| *** blahRus has quit IRC | 23:29 | |
| *** absubram has joined #kolla | 23:50 | |
| *** achanda has quit IRC | 23:52 | |
| *** absubram has quit IRC | 23:57 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!