| *** topshare has quit IRC | 00:02 | |
| *** topshare has joined #openstack-trove | 00:11 | |
| *** mingdang1 has joined #openstack-trove | 00:11 | |
| *** chlong has joined #openstack-trove | 00:11 | |
| *** topshare has quit IRC | 00:15 | |
| *** saurabhs has joined #openstack-trove | 00:15 | |
| *** mayurig has quit IRC | 00:23 | |
| *** mingdang_ has joined #openstack-trove | 00:25 | |
| *** mingdang1 has quit IRC | 00:29 | |
| *** esmute__ has quit IRC | 00:40 | |
| *** esmute__ has joined #openstack-trove | 00:41 | |
| *** saurabhs has quit IRC | 00:46 | |
| *** ishi_ has joined #openstack-trove | 00:51 | |
| *** esmute__ has quit IRC | 00:52 | |
| *** topshare has joined #openstack-trove | 00:57 | |
| *** Longgeek has joined #openstack-trove | 01:16 | |
| *** Longgeek has quit IRC | 01:20 | |
| *** erkules_ has joined #openstack-trove | 01:23 | |
| *** erkules has quit IRC | 01:26 | |
| *** mingdang_ has quit IRC | 01:27 | |
| *** ishi_ has quit IRC | 01:28 | |
| *** hogepodge has quit IRC | 01:29 | |
| *** jaredroh_ has joined #openstack-trove | 01:34 | |
| *** hogepodge has joined #openstack-trove | 01:41 | |
| *** tianqing has joined #openstack-trove | 01:56 | |
| *** jaredroh_ has quit IRC | 01:56 | |
| *** jaredroh_ has joined #openstack-trove | 01:56 | |
| *** jaredroh_ has quit IRC | 01:56 | |
| *** jaredrohe_ has joined #openstack-trove | 02:04 | |
| *** jaredrohe has quit IRC | 02:09 | |
| *** jaredrohe_ has quit IRC | 02:11 | |
| *** jaredrohe has joined #openstack-trove | 02:12 | |
| tianqing | why we use reload mysql rather than restart it to make configuration overrides active? | 02:21 |
|---|---|---|
| tianqing | we right now use restart to make mysql configuration overrides active...i want to ask we may use reload instead of restart the mysql... | 02:22 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/trove: Updated from global requirements https://review.openstack.org/199377 | 02:31 |
| SlickNik | tianqing: There are service variable in MySQL that are not dynamic that require a service restart to take effect. They don't take effect with just a service reload. | 02:32 |
| SlickNik | tianqing: I believe SIGHUP for mysql only flushes privileges, and doesn't actually reload any non-dynamic variables (for which you need a restart). | 02:34 |
| *** bhunter71 has quit IRC | 02:35 | |
| openstackgerrit | Masaki Matsushita proposed openstack/python-troveclient: Make subcommands accept flavor name and cluster name https://review.openstack.org/189573 | 02:37 |
| *** rnowling has joined #openstack-trove | 02:44 | |
| tianqing | when the user attach/detach a configuration, we may probably restart mysql. | 02:53 |
| tianqing | the logic to detach a configuration is to check whether there are some parameter not in default configuration. If there are we will restart the mysql. | 02:54 |
| tianqing | But there is a hypothesis, if the override parameter in default configuration, it must be a dynamic configuration. Here should we check for that? | 03:02 |
| *** csoukup has joined #openstack-trove | 03:10 | |
| *** Longgeek has joined #openstack-trove | 03:16 | |
| *** Longgeek has quit IRC | 03:22 | |
| *** nexusz99 has joined #openstack-trove | 03:41 | |
| *** annashen has joined #openstack-trove | 04:01 | |
| *** c_soukup has joined #openstack-trove | 04:14 | |
| *** csoukup has quit IRC | 04:16 | |
| *** david-ly_ has joined #openstack-trove | 04:19 | |
| *** sgotliv has joined #openstack-trove | 04:21 | |
| *** david-lyle has quit IRC | 04:22 | |
| *** annashen has quit IRC | 04:28 | |
| *** chlong has quit IRC | 04:35 | |
| *** chlong has joined #openstack-trove | 04:43 | |
| *** rnowling has quit IRC | 04:45 | |
| *** chlong has quit IRC | 04:51 | |
| *** nirupam has quit IRC | 04:55 | |
| *** madhavan has joined #openstack-trove | 05:01 | |
| *** madhavan has quit IRC | 05:02 | |
| *** Longgeek has joined #openstack-trove | 05:16 | |
| *** chlong has joined #openstack-trove | 05:19 | |
| *** ishi_ has joined #openstack-trove | 05:23 | |
| *** c_soukup has quit IRC | 05:34 | |
| *** ishi_ has quit IRC | 05:36 | |
| *** ig0r__ has joined #openstack-trove | 05:51 | |
| *** sgotliv has quit IRC | 05:52 | |
| *** sgotliv has joined #openstack-trove | 05:52 | |
| *** ig0r_ has quit IRC | 05:52 | |
| *** tobe has joined #openstack-trove | 06:19 | |
| *** mingdang1 has joined #openstack-trove | 06:21 | |
| *** dmgweb has joined #openstack-trove | 06:21 | |
| dmgweb | hello | 06:22 |
| dmgweb | ERROR trove.instance.models [-] An error occurred communicating with the guest: any idea? | 06:22 |
| openstackgerrit | Craig Vyvial proposed openstack/trove-specs: Added Percona cluster spec https://review.openstack.org/132132 | 06:29 |
| tobe | Hi @dmgweb, did your guestagent start? | 06:30 |
| dmgweb | in controller node? | 06:30 |
| *** coolsvap|away is now known as coolsvap | 06:32 | |
| abhis | how to check whether trove is installed properly and services are running | 06:34 |
| tobe | @dmgweb in compute node | 06:36 |
| tobe | the guest agent runs in the virtual machine | 06:36 |
| tobe | @abhis You may check the log in /var/log/trove/api.log and try "trove list" | 06:36 |
| dmgweb | but the archive trove-guestagent.conf it has in controller node | 06:37 |
| dmgweb | ? | 06:37 |
| dmgweb | the image virtual machine is tesora | 06:37 |
| dmgweb | srry for my english | 06:37 |
| tobe | The trove-guestagent.conf is used by guest agent which runs in the vm and you can create it by running "trove create" | 06:38 |
| dmgweb | Then I would need to enter the mv and edit that file, right? | 06:39 |
| dmgweb | It is not simply use a picture of trove of tesora | 06:40 |
| tobe | No, the file will pass to vm from the controller. You don't need to add it in the image manually | 06:40 |
| tobe | I'm not familiar with the tesora service. You may contact their developers :) | 06:41 |
| dmgweb | ah okay, then where you can be the fault? | 06:41 |
| dmgweb | I installed without tesora trove and I have met with the same error. | 06:42 |
| dmgweb | I installed without tesora trove and I have met with the same error. | 06:42 |
| dmgweb | so I think I'm missing some concept. | 06:42 |
| dmgweb | service trove-guestagent it must be installed on controller? | 06:43 |
| dmgweb | or only the archive | 06:43 |
| *** mingdang1 has quit IRC | 06:45 | |
| openstackgerrit | Merged openstack/trove-integration: MongoDB smallfiles/bind_ip settings not working properly https://review.openstack.org/195264 | 06:45 |
| openstackgerrit | Craig Vyvial proposed openstack/trove-specs: Added Percona cluster spec https://review.openstack.org/132132 | 06:48 |
| openstackgerrit | Craig Vyvial proposed openstack/trove-specs: Added Percona cluster spec https://review.openstack.org/132132 | 06:53 |
| *** k4n0 has joined #openstack-trove | 06:59 | |
| openstackgerrit | Craig Vyvial proposed openstack/trove-specs: Added Percona cluster spec https://review.openstack.org/132132 | 07:01 |
| openstackgerrit | Craig Vyvial proposed openstack/trove-specs: Added Percona cluster spec https://review.openstack.org/132132 | 07:06 |
| abhis | @dmgweb while running glance image create command, it is giving error "it is unable to connect to http://controller /v1/image" connection refused | 07:14 |
| abhis | i am not able to remove the error | 07:14 |
| abhis | i am getting error finding address | 07:15 |
| dmgweb | @abhis can you ping yo controller? | 07:22 |
| *** exploreshaifali has joined #openstack-trove | 07:38 | |
| openstackgerrit | Sushil Kumar proposed openstack/trove-specs: Datastore Registration API spec https://review.openstack.org/188072 | 07:46 |
| *** tobe has quit IRC | 07:46 | |
| *** tobe has joined #openstack-trove | 08:16 | |
| *** tosky has joined #openstack-trove | 08:19 | |
| *** chlong has quit IRC | 08:21 | |
| *** tobe has quit IRC | 08:25 | |
| *** tobe has joined #openstack-trove | 08:25 | |
| *** tianqing has quit IRC | 08:28 | |
| abhis | yes i am able to ping | 08:40 |
| *** tobe has quit IRC | 08:40 | |
| openstackgerrit | Merged openstack/trove: Updated from global requirements https://review.openstack.org/199377 | 09:19 |
| *** tobe has joined #openstack-trove | 09:20 | |
| dmgweb | @tobe not working... :( | 09:21 |
| *** mingdang1 has joined #openstack-trove | 09:26 | |
| *** mingdang_ has joined #openstack-trove | 09:29 | |
| *** mingdang1 has quit IRC | 09:32 | |
| *** IanGovett has joined #openstack-trove | 09:39 | |
| *** tianqing has joined #openstack-trove | 09:54 | |
| *** mingdang_ has quit IRC | 09:55 | |
| openstackgerrit | Merged openstack/trove: Removed the non-existent method call https://review.openstack.org/196762 | 10:05 |
| *** nexusz99 has quit IRC | 10:06 | |
| *** dmgweb has quit IRC | 10:08 | |
| tianqing | hello.. | 10:21 |
| tianqing | where is the info about master stored in slave? | 10:21 |
| *** tobe has quit IRC | 10:29 | |
| *** coolsvap is now known as coolsvap|away | 10:33 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/trove: Updated from global requirements https://review.openstack.org/199482 | 10:36 |
| dougshelley66 | tianqing for mysql you can connect using mysql client and issue "SHOW SLAVE STATUS" | 10:43 |
| tianqing | i know | 10:45 |
| tianqing | the info is persistent. | 10:45 |
| tianqing | and it is stored in master.info | 10:45 |
| dougshelley66 | ok | 10:45 |
| tianqing | so if the slave is restart, it can get master's info | 10:46 |
| tianqing | i want to check that... | 10:46 |
| tianqing | those is what i get... | 10:46 |
| tianqing | thank you all the same... | 10:46 |
| *** tianqing has quit IRC | 10:50 | |
| *** mingdang1 has joined #openstack-trove | 10:56 | |
| *** mingdang1 has quit IRC | 11:00 | |
| *** topshare has quit IRC | 11:02 | |
| *** tosky has quit IRC | 11:16 | |
| *** tosky has joined #openstack-trove | 11:16 | |
| *** pmackinn has joined #openstack-trove | 11:24 | |
| pmackinn | morning | 11:24 |
| *** topshare has joined #openstack-trove | 11:26 | |
| *** mingdang1 has joined #openstack-trove | 11:29 | |
| *** topshare has quit IRC | 11:30 | |
| *** Longgeek has quit IRC | 11:41 | |
| *** Longgeek has joined #openstack-trove | 11:42 | |
| *** Longgeek has quit IRC | 11:42 | |
| *** Longgeek has joined #openstack-trove | 11:42 | |
| *** Longgeek has quit IRC | 11:43 | |
| *** tianqing has joined #openstack-trove | 11:47 | |
| *** exploreshaifali has quit IRC | 11:51 | |
| *** topshare has joined #openstack-trove | 11:54 | |
| tianqing | how to view the old message about this channel? | 11:57 |
| pmackinn | tianqing, http://eavesdrop.openstack.org/irclogs/%23openstack-trove/ | 11:59 |
| tianqing | oh thanks a lot... | 12:00 |
| tianqing | where do you find that, i search on google, but do not have meaningful result... | 12:00 |
| tianqing | the key words for search is freenode old messages... | 12:01 |
| tianqing | may be i use wrong key words... | 12:01 |
| pmackinn | openstack irc | 12:01 |
| *** tosky_ has joined #openstack-trove | 12:02 | |
| tianqing | thanks a lot @pmackinn | 12:03 |
| *** tosky has quit IRC | 12:04 | |
| pmackinn | np | 12:04 |
| *** Longgeek has joined #openstack-trove | 12:25 | |
| *** exploreshaifali has joined #openstack-trove | 12:28 | |
| *** chlong has joined #openstack-trove | 12:30 | |
| *** amrith is now known as _amrith_ | 12:33 | |
| *** erkules_ is now known as erkules | 12:39 | |
| *** erkules has joined #openstack-trove | 12:39 | |
| *** mingdang1 has quit IRC | 12:43 | |
| *** _crobertsrh is now known as crobertsrh | 12:50 | |
| *** coolsvap|away is now known as coolsvap | 12:52 | |
| *** tomblank has joined #openstack-trove | 13:04 | |
| *** pmackinn has quit IRC | 13:05 | |
| *** rnowling has joined #openstack-trove | 13:29 | |
| *** radez_g0n3 is now known as radez | 13:37 | |
| *** pmackinn has joined #openstack-trove | 13:49 | |
| *** csoukup has joined #openstack-trove | 13:58 | |
| *** _amrith_ is now known as amrith | 14:07 | |
| *** bhunter71 has joined #openstack-trove | 14:07 | |
| *** tianqing has quit IRC | 14:11 | |
| *** flwang has quit IRC | 14:40 | |
| *** sriram_tesora has joined #openstack-trove | 14:53 | |
| *** csoukup has quit IRC | 15:01 | |
| *** coolsvap is now known as coolsvap|away | 15:04 | |
| *** coolsvap|away is now known as coolsvap | 15:12 | |
| *** jgriffith has quit IRC | 15:20 | |
| *** topshare_ has joined #openstack-trove | 15:25 | |
| *** topshare has quit IRC | 15:26 | |
| *** nexusz99 has joined #openstack-trove | 15:41 | |
| *** david-ly_ is now known as david-lyle | 15:42 | |
| *** exploreshaifali has quit IRC | 15:42 | |
| *** coolsvap is now known as coolsvap|away | 15:47 | |
| *** coolsvap|away is now known as coolsvap | 15:48 | |
| *** amrith is now known as _amrith_ | 16:02 | |
| *** tomblank has quit IRC | 16:04 | |
| *** ishi_ has joined #openstack-trove | 16:04 | |
| *** tomblank has joined #openstack-trove | 16:08 | |
| *** dontalton has joined #openstack-trove | 16:10 | |
| openstackgerrit | Sushil Kumar proposed openstack/trove: Fixes the detection of user with extra access https://review.openstack.org/199635 | 16:10 |
| *** coolsvap is now known as coolsvap|away | 16:11 | |
| *** cweid has joined #openstack-trove | 16:18 | |
| *** flwang has joined #openstack-trove | 16:18 | |
| *** johnma has joined #openstack-trove | 16:20 | |
| *** ishi_ has quit IRC | 16:22 | |
| *** mayurig has joined #openstack-trove | 16:26 | |
| *** tomblank has quit IRC | 16:28 | |
| *** tomblank has joined #openstack-trove | 16:39 | |
| *** dontalton is now known as bitblt | 16:41 | |
| *** bitblt has quit IRC | 16:41 | |
| *** bitblt has joined #openstack-trove | 16:41 | |
| *** bitblt has quit IRC | 16:42 | |
| *** tosky_ has quit IRC | 16:43 | |
| *** tosky_ has joined #openstack-trove | 16:43 | |
| *** tomblank has quit IRC | 16:44 | |
| sushilkm | i wanted reviewers attention towards https://review.openstack.org/188072 | 16:45 |
| *** ashleighfarnham has joined #openstack-trove | 16:46 | |
| *** _amrith_ is now known as amrith | 17:00 | |
| *** topshare_ has quit IRC | 17:02 | |
| *** tomblank has joined #openstack-trove | 17:02 | |
| *** tomblank has quit IRC | 17:09 | |
| *** tosky_ has quit IRC | 17:12 | |
| *** mingdang1 has joined #openstack-trove | 17:18 | |
| *** mingdang1 has quit IRC | 17:22 | |
| *** cweid has quit IRC | 17:28 | |
| *** tomblank has joined #openstack-trove | 17:29 | |
| *** exploreshaifali has joined #openstack-trove | 17:48 | |
| *** mayurig has quit IRC | 17:48 | |
| *** mayurig has joined #openstack-trove | 17:49 | |
| *** mayurig has quit IRC | 17:49 | |
| *** mayurig has joined #openstack-trove | 17:50 | |
| *** annashen has joined #openstack-trove | 17:53 | |
| *** Longgeek has quit IRC | 17:57 | |
| *** tosky has joined #openstack-trove | 17:57 | |
| *** Longgeek has joined #openstack-trove | 17:57 | |
| *** Longgeek has quit IRC | 17:57 | |
| *** Longgeek has joined #openstack-trove | 17:59 | |
| *** pmackinn has quit IRC | 18:27 | |
| *** saurabhs has joined #openstack-trove | 18:33 | |
| *** saurabhs has left #openstack-trove | 18:33 | |
| *** exploreshaifali has quit IRC | 18:36 | |
| openstackgerrit | Sharika Pongubala proposed openstack/trove-integration: Change DATASTORE_PKG_LOCATION for vertica https://review.openstack.org/181186 | 18:38 |
| *** saurabhs has joined #openstack-trove | 18:56 | |
| amrith | so ... who's coming to mid-cycle raise your hands now | 19:01 |
| amrith | ./ | 19:01 |
| cp16net | ./ | 19:02 |
| dougshelley66 | o/ | 19:02 |
| amrith | 3 people, not much of a mid-cycle. | 19:02 |
| dougshelley66 | but it is the party crowd | 19:02 |
| cp16net | lol | 19:02 |
| mvandijk_ | ~o~ | 19:02 |
| SlickNik | amrith: I've gotten a lot more registrations that that. | 19:02 |
| * SlickNik heads off to look it up | 19:02 | |
| cp16net | i havnt been confirmed to go yet that i know of tho | 19:03 |
| cp16net | but i plan on it :-P | 19:03 |
| SlickNik | lol @ cp16net | 19:03 |
| * amrith goes away to recycle a whole mornings coffee | 19:03 | |
| SlickNik | Why haven't you registered yet? | 19:03 |
| SlickNik | amrith: I've got 10 so far. | 19:03 |
| SlickNik | (And that's not counting about at least 3-4 more HP folks that I know are going) | 19:04 |
| *** nshah has left #openstack-trove | 19:04 | |
| cp16net | i thought i did... | 19:05 |
| cp16net | yeah i did | 19:06 |
| amrith | SlickNik, I thought I registered ... would you share the evite/eventbrite link again please | 19:07 |
| mvandijk_ | https://www.eventbrite.com/e/trove-mid-cycle-sprint-liberty-tickets-17600134476 | 19:07 |
| mvandijk_ | that? | 19:07 |
| SlickNik | Hah, I thought you said you didn't — but I see you in eventbrite. | 19:07 |
| SlickNik | mvandijk_: yes | 19:07 |
| SlickNik | mvandijk_: Thanks! | 19:08 |
| SlickNik | amrith: I don't see your registration yet | 19:08 |
| amrith | SlickNik, I just registered. | 19:08 |
| SlickNik | Ah, cool | 19:08 |
| * SlickNik hits F5 | 19:08 | |
| SlickNik | There you are | 19:08 |
| amrith | and then there were 11 | 19:08 |
| cp16net | midcycle goes to 11 | 19:09 |
| amrith | oh, I have a question about the agenda for mid-cycle. | 19:09 |
| * cp16net waits in anticipation for the question | 19:10 | |
| *** bhunter71 has quit IRC | 19:11 | |
| amrith | cp16net, I just updated the agenda on the wiki ;) | 19:11 |
| amrith | https://wiki.openstack.org/wiki/Sprints/TroveLibertySprint#Schedule | 19:11 |
| *** bitblt has joined #openstack-trove | 19:13 | |
| amrith | cp16net, see it? | 19:14 |
| *** coolsvap|away is now known as coolsvap | 19:14 | |
| SlickNik | OH YEA! | 19:15 |
| amrith | SlickNik, I'm going to work on that new thing ... | 19:15 |
| SlickNik | Thanks for updating the agenda, amrith. | 19:16 |
| amrith | you get to participate | 19:16 |
| amrith | you can help me organize for the rest ;) | 19:16 |
| amrith | on the condition that I can play too ... | 19:16 |
| SlickNik | Sounds like a plan | 19:16 |
| amrith | I will post (to ml) a note about this. I would like to see if I can make it part of other projects as well. | 19:16 |
| cp16net | oh i had to look at the history | 19:18 |
| cp16net | i see it | 19:18 |
| cp16net | i put a map | 19:21 |
| SlickNik | amrith: Let me know when you have a moment — jodah wanted to have a quick chat about https://review.openstack.org/#/c/172161 | 19:22 |
| amrith | SlickNik, I have a moment | 19:27 |
| cp16net | hmm how does enabling root on a cluster differ from enabling root on an instance in a cluster? | 19:28 |
| amrith | SlickNik, I have several moments | 19:28 |
| *** pmackinn has joined #openstack-trove | 19:31 | |
| *** csoukup has joined #openstack-trove | 19:32 | |
| SlickNik | jodah: around? | 19:36 |
| amrith | SlickNik, I'm here (some conditions apply, read the fine print) | 19:39 |
| SlickNik | amrith: So I've been talking to jodah about the best way forward with https://review.openstack.org/#/c/172161/ | 19:39 |
| SlickNik | In order for this to not be a breaking change, we'll have to take the "support the new guest API and the with old guest API for one release" approach. | 19:40 |
| SlickNik | and the old* | 19:40 |
| amrith | may I back up the truck a couple of yards | 19:40 |
| amrith | why provide the password? | 19:40 |
| amrith | I was told (days of yore) that the reason it was done this way | 19:41 |
| SlickNik | The primary reason that jodah and team want to do this is to provide a better UX for root create through horizon. | 19:41 |
| amrith | was that trove didn't know what the password rules were for a database | 19:41 |
| amrith | but the datastore did | 19:41 |
| amrith | so easiest for the datastore/guest to generate the password | 19:41 |
| amrith | more secure | 19:41 |
| amrith | blah blah blah | 19:41 |
| amrith | I bought it. | 19:41 |
| amrith | the current UX is (I believe) that when you click enable root, the response says success or failure and shows you the password. | 19:42 |
| SlickNik | But how do you do that on instance create? | 19:42 |
| amrith | you would have to provide a response to the user no matter what; whether the user provides the password or not. | 19:42 |
| amrith | instance was created successfully, the password is 'shazzam' | 19:42 |
| amrith | you get the password in the response to the POST | 19:42 |
| amrith | there's no *requirement* that you must provide the password in a popup | 19:43 |
| amrith | I don't think. | 19:43 |
| amrith | But I'm not a UX expert. | 19:43 |
| amrith | if they want to provide password, that's fine. | 19:44 |
| SlickNik | Our instance create API for users already does this (i.e. put in a password when you create users / schemas) for mysql | 19:44 |
| amrith | About how they do it. | 19:44 |
| amrith | currently for instance creation, the way you do this is ... | 19:44 |
| amrith | in the POST for a create instance you set root-enable | 19:44 |
| amrith | and you get a password on response. | 19:44 |
| SlickNik | And this would achieve consistency between the UX for user creation and root creation. | 19:44 |
| SlickNik | Consistent UX is a good thing. | 19:44 |
| amrith | later if you want to enable root, you post to something. | 19:44 |
| amrith | so, could we do this in a way that wouldn't break existing guest images; agents. | 19:45 |
| amrith | i.e. make the new API in addition to the existing one? | 19:45 |
| amrith | so if the user doesn't provide a password, then make one up and send it back. | 19:45 |
| SlickNik | Pop-ups with guids that you have to remember (and cannot change) are a bad UX :) | 19:45 |
| amrith | if the user provides a password, use it. | 19:45 |
| amrith | that should solve the problem with no compat issues? | 19:45 |
| SlickNik | I think that's the proposal. | 19:46 |
| SlickNik | If no password is given, make one up | 19:46 |
| SlickNik | So you're still backwards compatible. | 19:46 |
| amrith | sorry, I missed that part. | 19:46 |
| amrith | if that's the case, I'm wrong in my -1 | 19:46 |
| SlickNik | (exactly like it is today) | 19:46 |
| cp16net | only compat issue i see is the rpc method call changes when provided | 19:46 |
| amrith | line 106 "OPTIONAL" password | 19:47 |
| cp16net | but that should be fine if there is a default set | 19:47 |
| amrith | is that what I'm looking for? | 19:47 |
| amrith | for consistency, will we make all current guest agents that support root also accept a password if one is provided? | 19:47 |
| SlickNik | Yeah, line 28 as well. | 19:47 |
| cp16net | yeah i think that makes sense | 19:48 |
| SlickNik | But it would be good for the spec to call this out explicitly. | 19:48 |
| SlickNik | I know that's jodah's plan because I had the _same_ discussion with him. | 19:48 |
| amrith | SlickNik, call what out? that existing guests will also be updated to do this? | 19:48 |
| amrith | existing guests for single instance for example? | 19:48 |
| cp16net | q. does rpc support method overrides? | 19:49 |
| SlickNik | Specifically call out that making an API call without a password will still result in exactly the same behavior as it exists today. | 19:49 |
| cp16net | err not overrides... overloads | 19:49 |
| SlickNik | (i.e. will generate a password, and return it as part of the response.) | 19:49 |
| amrith | and about existing (say) MySQL agent? | 19:50 |
| *** rbrooker has joined #openstack-trove | 19:50 | |
| SlickNik | amrith: We'd have to create a new RPC API for the passworded scenario to ensure backwards compat. | 19:52 |
| amrith | SlickNik, that I don't understand | 19:52 |
| cp16net | if we overloaded the enable_root function then we wouldnt have add a new RPC api call | 19:52 |
| amrith | wouldn't the code on the guest just check if a password is provided and use it if it is and create one if not? | 19:52 |
| cp16net | assuming that works | 19:52 |
| amrith | cp16net, not sure you have to go taht far | 19:52 |
| amrith | it's a matter of just checking if a password was provided. | 19:52 |
| amrith | or am I missing something? | 19:53 |
| * cp16net looks at the guest api calls... | 19:53 | |
| amrith | we could make the same change in the guest(s). | 19:53 |
| SlickNik | amrith: You'd be changing the function signature of the RPC call otherwise, and that will result in a failed RPC call. | 19:53 |
| amrith | what, wait ... | 19:53 |
| amrith | why? | 19:53 |
| jodah | the simplest thing would be to simply add a password parameter with a default value. but the potential breakage of doing that, i'm unsure of | 19:53 |
| SlickNik | That's the way messaging in oslo works (or at least used to, in the oslo-incubator days) | 19:54 |
| SlickNik | I don't think oslo.messaging has done anything to fix this. | 19:54 |
| SlickNik | (but I could be wrong) | 19:54 |
| amrith | I believe the method already has a password parameter. | 19:55 |
| * amrith waits for cp16net to confirm | 19:55 | |
| jodah | not on the guestagent API | 19:55 |
| SlickNik | really? | 19:56 |
| * SlickNik heads off to check if the guest API already has a password parameter | 19:56 | |
| jodah | https://github.com/openstack/trove/blob/master/trove/guestagent/api.py#L179 | 19:56 |
| cp16net | looks like most of the service calls accept a root_password=None param | 19:56 |
| SlickNik | thanks jodah | 19:57 |
| cp16net | but the RPC call doesnt send it | 19:57 |
| amrith | so ... we have enable root | 19:57 |
| amrith | at /v1.0/{tenant-id}/instances/{instance}/root | 19:57 |
| amrith | which currently doesn't take a password | 19:57 |
| amrith | but it is merely a matter of the handler for this (which we know) going and looking to see if there's a parameter | 19:57 |
| *** coolsvap is now known as coolsvap|away | 19:58 | |
| cp16net | yeah doesnt seem like that will break anything to me | 19:58 |
| amrith | that won't break anything | 19:58 |
| SlickNik | That's good so far. | 19:58 |
| SlickNik | However, if there _is_ a password field in the request. | 19:58 |
| SlickNik | It must now send this to the guest through the RPC call | 19:58 |
| jodah | we propose { "password":"foo"} | 19:58 |
| cp16net | and most* of the datastores already ahve a root_passowrd=None | 19:59 |
| amrith | the root controller has to just handle it | 19:59 |
| cp16net | postgresql/couchbase/mysql in their guestagent | 19:59 |
| amrith | looking at that | 19:59 |
| jodah | yea, i'm not sure when that snuck in https://github.com/openstack/trove/blob/master/trove/guestagent/datastore/mysql/service.py#L416 | 20:00 |
| amrith | bbiab | 20:00 |
| SlickNik | So this means either 1. changing the RPC call signature on the guest. (not backwards compatible), or 2. creating a new RPC call with a new signature calling this (backwards compatible) | 20:01 |
| SlickNik | and calling this* | 20:01 |
| jodah | i assume #2 wins. is there any precedent for this sort of API change? | 20:04 |
| SlickNik | jodah: Yes, we had to do this once before — although I can't remember what the specific scenario that required that was. | 20:05 |
| *** exploreshaifali has joined #openstack-trove | 20:05 | |
| SlickNik | jodah: just heard that amrith had to duck out due to a different commitment. I think that with some care (option #2) this should be achievable. | 20:06 |
| jodah | ok. will revise spec accordingly | 20:06 |
| SlickNik | amrith should be back in an hour or so — let's close on the conversation so that we can move the BP forward. | 20:06 |
| SlickNik | jodah: ++ Thanks! | 20:06 |
| jodah | ...after then :) | 20:06 |
| jodah | so update now or wait for amrith? | 20:07 |
| *** bhunter71 has joined #openstack-trove | 20:08 | |
| SlickNik | Let's wait for an hour — if we don't have the conversation in about an hour (for whatever reason) we can update the spec and drive it forward through the gerrit review. | 20:08 |
| SlickNik | sound good? | 20:08 |
| SlickNik | (If you want to eagerly update the spec, I wouldn't be opposed to that either :)) | 20:09 |
| *** coolsvap|away is now known as coolsvap | 20:09 | |
| *** openstackgerrit has quit IRC | 20:10 | |
| *** amrith is now known as _amrith_ | 20:10 | |
| *** openstackgerrit has joined #openstack-trove | 20:10 | |
| jodah | k | 20:11 |
| *** rbrooker has quit IRC | 20:11 | |
| openstackgerrit | Mayuri Ganguly proposed openstack/python-troveclient: Accepting network and availability zone for instances in cluster https://review.openstack.org/184349 | 20:13 |
| *** sgotliv has quit IRC | 20:26 | |
| *** rbrooker has joined #openstack-trove | 20:37 | |
| *** _amrith_ is now known as amrith | 20:38 | |
| amrith | SlickNik, I couldn't stay away so I'm back! | 20:39 |
| amrith | jodah, SlickNik, cp16net ... ping | 20:41 |
| cp16net | pong | 20:41 |
| amrith | so what did y'all decide? | 20:41 |
| cp16net | i think it was on #2 although i dont *think* its nessesary | 20:41 |
| SlickNik | amrith: pong | 20:41 |
| amrith | what does "changing the RPC signature" mean? | 20:42 |
| SlickNik | Changing the call here: https://github.com/openstack/trove/blob/master/trove/guestagent/api.py#L179 | 20:43 |
| SlickNik | to something like: def enable_root(self, password=None): | 20:44 |
| SlickNik | ^ amrith | 20:45 |
| * amrith looks | 20:45 | |
| amrith | ok, and then, you'd still have to change the self._call to "enable_root", ..., password=password | 20:46 |
| amrith | yes? | 20:46 |
| amrith | the rest would just be magic on the guest agent. | 20:47 |
| amrith | sounds good. | 20:47 |
| amrith | why not just do that? | 20:47 |
| amrith | why #2? | 20:47 |
| openstackgerrit | Sharika Pongubala proposed openstack/trove-integration: Change DATASTORE_PKG_LOCATION for vertica https://review.openstack.org/181186 | 20:48 |
| cp16net | i think #2 was a fool proof way if it breaks the api | 20:48 |
| SlickNik | Because #1 breaks the RPC contract | 20:49 |
| * amrith ponders ... | 20:49 | |
| SlickNik | If I have a TM that's calling enable_root(self), and a guest that responds only to enable_root(self, password=None) the RPC messaging layer fails. | 20:50 |
| SlickNik | (or at least would fail when I checked last about 6 months ago) | 20:51 |
| cp16net | yeah i'm curious about that | 20:51 |
| amrith | mumble mumble | 20:51 |
| amrith | OK, sounds good. | 20:52 |
| amrith | let's do that | 20:52 |
| amrith | but then what happens | 20:52 |
| amrith | you cannot call self._call("enable_root", ...) for the same reasons. | 20:53 |
| amrith | you'll have to self._call("enable_root2", ...) or some such thing. | 20:53 |
| SlickNik | So for one release the guest supports both enable_root(self) and enable_root_with_password(self, password=None) | 20:53 |
| SlickNik | (at least for one release*) | 20:53 |
| amrith | so are we saying that the guest agent of release N-1 should work with a controller of release N? | 20:54 |
| *** radez is now known as radez_g0n3 | 20:55 | |
| SlickNik | The rule we've always tried to follow is that the guest release N+1 works with a controller of release N — this enables having a rolling upgrading of the guests before we upgrade the control plane (and minimize downtime). | 20:57 |
| amrith | the question I have with this approach is that I could, in theory, have a TM that has code that wants to send down a password; how does it know whether the guest will support it or not? | 20:58 |
| SlickNik | currently the TM doesn't support that because the RPC API isn't currently versioned. | 20:58 |
| amrith | what you just said, guest N+1 and controller N (which I call guest N and controller N-1) is the opposite of what I believe #2 provides for. | 20:58 |
| amrith | iiuc #2 would do this ... | 20:59 |
| amrith | change https://github.com/openstack/trove/blob/master/trove/guestagent/api.py#L179 to | 20:59 |
| amrith | def enable_root(self, password=None) | 20:59 |
| amrith | and then make it call | 20:59 |
| SlickNik | If you had a versioned RPC API, you'd be able to change the call based on the supported RPC API version. | 20:59 |
| amrith | return self._call("enable_root_with_password", ..., password=password) | 20:59 |
| *** tomblank has quit IRC | 20:59 | |
| amrith | and on the guest we would have a new method called enable_with_password() | 21:00 |
| amrith | yes? | 21:00 |
| cp16net | how do you restart the guestagent on an instance? | 21:00 |
| cp16net | i dont see a /etc/init.d/ script | 21:01 |
| amrith | https://github.com/openstack/trove/blob/master/trove/guestagent/datastore/mysql/manager.py#L99 | 21:01 |
| amrith | would remain unchanged | 21:01 |
| SlickNik | No, #2 is proposing you'd leave https://github.com/openstack/trove/blob/master/trove/guestagent/api.py#L179 the way it is, and add a "new def enable_root(self, password=None)" | 21:01 |
| amrith | and right below that for example you would have enable_root_with_password ... | 21:01 |
| amrith | and that would call MySQLAdmin.enable_root(password=password) | 21:02 |
| SlickNik | Sorry add a new "def enable_root_with_password(self, password=None)" | 21:02 |
| amrith | is that correct? | 21:02 |
| amrith | cp16net, there's currently no API that will do that | 21:03 |
| amrith | <sales-pitch> if you use our images (tesora) you can do a service restart ... </sales-pitch> | 21:03 |
| amrith | SlickNik, I don't follow the need to change api.py | 21:04 |
| cp16net | but manually? | 21:04 |
| cp16net | i'm in the container i guess i will poke around it | 21:04 |
| amrith | I believe that api.py is entirely a control plane thing, and entirely in one python process (task manager) | 21:04 |
| amrith | in a container, service commands don't work | 21:04 |
| amrith | that's the big issue with containers | 21:05 |
| SlickNik | Sorry my mistake — we wouldn't need to change api.py — just manager.py — api.py could stay the same. | 21:05 |
| SlickNik | I copy/pasta'd the URI from above, without realizing it was api.py and not manager.py | 21:05 |
| * amrith thinks ... pasta, yum ;) | 21:05 | |
| SlickNik | All of what I was saying applies to manager.py, _not_ api.py | 21:05 |
| SlickNik | sorry for the confusion. | 21:06 |
| * amrith thinks about that (manager, api.py comment) | 21:06 | |
| amrith | cp16net, I was wrong | 21:07 |
| amrith | dougshelley66, and abramley just pointed me to this. | 21:07 |
| amrith | trove-integration/scripts/files/trove-guest.upstart.conf | 21:08 |
| amrith | these are the things that launch guest agent on first boot as well | 21:08 |
| *** crobertsrh is now known as _crobertsrh | 21:08 | |
| *** tosky has quit IRC | 21:13 | |
| *** tosky has joined #openstack-trove | 21:14 | |
| *** rnowling has quit IRC | 21:16 | |
| amrith | jodah, I understand what SlickNik, cp16net, and you are proposing ... I've removed my -1. https://review.openstack.org/#/c/172161/17 | 21:19 |
| jodah | making sure i understand it too.... :) | 21:19 |
| SlickNik | amrith: Thanks! | 21:21 |
| SlickNik | jodah: let's update the spec with the proposed approach. | 21:21 |
| jodah | pushing now..... | 21:22 |
| *** coolsvap is now known as coolsvap|away | 21:22 | |
| jodah | hmm, as soon as git review unhangs itself. | 21:23 |
| openstackgerrit | Jonathan Halterman proposed openstack/trove-specs: Add cluster root-enable. https://review.openstack.org/172161 | 21:25 |
| *** fifieldt has quit IRC | 21:30 | |
| openstackgerrit | Sharika Pongubala proposed openstack/trove-integration: Change DATASTORE_PKG_LOCATION for vertica https://review.openstack.org/181186 | 21:33 |
| *** fifieldt has joined #openstack-trove | 21:43 | |
| *** my24thautumn has joined #openstack-trove | 21:43 | |
| *** my24thautumn_ has joined #openstack-trove | 21:51 | |
| my24thautumn_ | /msg NickServ VERIFY REGISTER my24thautumn acbnrnhsfspa | 21:51 |
| openstackgerrit | Jonathan Halterman proposed openstack/trove-specs: Add cluster root-enable. https://review.openstack.org/172161 | 21:52 |
| *** my24thautumn has quit IRC | 21:53 | |
| *** amrith is now known as _amrith_ | 21:55 | |
| *** my24thautumn_ has left #openstack-trove | 22:02 | |
| *** mingdang1 has joined #openstack-trove | 22:04 | |
| *** mayurig has quit IRC | 22:05 | |
| *** mayurig has joined #openstack-trove | 22:09 | |
| *** mingdang1 has quit IRC | 22:22 | |
| *** IanGovett1 has joined #openstack-trove | 22:26 | |
| *** IanGovett has quit IRC | 22:28 | |
| *** sriram_tesora has quit IRC | 22:48 | |
| *** csoukup has quit IRC | 22:57 | |
| *** annashen has quit IRC | 23:12 | |
| *** exploreshaifali has quit IRC | 23:21 | |
| *** IanGovett1 has quit IRC | 23:29 | |
| *** tianqing has joined #openstack-trove | 23:35 | |
| *** topshare has joined #openstack-trove | 23:41 | |
| *** tosky has quit IRC | 23:47 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!