| *** VW has quit IRC | 00:10 | |
| *** Syed__ has quit IRC | 01:25 | |
| *** VW has joined #craton | 01:38 | |
| *** valw has joined #craton | 02:22 | |
| *** VW has quit IRC | 02:29 | |
| *** VW has joined #craton | 02:53 | |
| tojuvone | jimbaker: Let's see if can catch some meeting next week. | 03:42 |
|---|---|---|
| *** valw has quit IRC | 03:51 | |
| *** valw has joined #craton | 03:52 | |
| *** valw has quit IRC | 04:13 | |
| *** valw has joined #craton | 04:28 | |
| *** VW has quit IRC | 05:02 | |
| *** valw has quit IRC | 05:08 | |
| *** valw has joined #craton | 06:53 | |
| *** valw has quit IRC | 06:58 | |
| *** valw has joined #craton | 08:52 | |
| *** valw has quit IRC | 08:57 | |
| *** johnthetubaguy has quit IRC | 09:16 | |
| *** johnthetubaguy has joined #craton | 09:19 | |
| *** tojuvone has quit IRC | 09:36 | |
| *** tojuvone has joined #craton | 09:36 | |
| *** rainya_ has quit IRC | 10:00 | |
| *** rainya has joined #craton | 10:01 | |
| *** valw has joined #craton | 12:55 | |
| *** valw has quit IRC | 13:14 | |
| *** valw has joined #craton | 14:40 | |
| *** valw has quit IRC | 14:42 | |
| *** valw has joined #craton | 14:46 | |
| *** valw has quit IRC | 14:56 | |
| *** valw has joined #craton | 15:00 | |
| *** VW has joined #craton | 15:07 | |
| *** Syed__ has joined #craton | 15:14 | |
| *** jovon has joined #craton | 15:15 | |
| *** VW has quit IRC | 15:20 | |
| *** VW has joined #craton | 15:21 | |
| *** valw_ has joined #craton | 15:26 | |
| *** rainya_ has joined #craton | 15:27 | |
| *** valw has quit IRC | 15:28 | |
| *** rainya has quit IRC | 15:28 | |
| *** valw_ has quit IRC | 15:35 | |
| *** valw has joined #craton | 15:36 | |
| *** valw has quit IRC | 15:38 | |
| jimbaker | tojuvone, sounds good! | 15:46 |
| *** VW_ has joined #craton | 16:01 | |
| *** VW_ has quit IRC | 16:02 | |
| *** VW_ has joined #craton | 16:03 | |
| *** VW_ has quit IRC | 16:03 | |
| *** rainya_ has quit IRC | 16:04 | |
| *** VW has quit IRC | 16:04 | |
| *** VW has joined #craton | 16:04 | |
| tojuvone | Hi, catch some evening snack, brb | 16:48 |
| tojuvone | rye bread. It was just nominated our national food yesterday. | 16:56 |
| *** VW has quit IRC | 16:58 | |
| *** VW has joined #craton | 16:58 | |
| jimbaker | tojuvone, nice. i do enjoy rye bread myself. rye also makes for a great beer :) | 17:02 |
| tojuvone | :) | 17:03 |
| *** valw has joined #craton | 18:08 | |
| *** jovon has quit IRC | 18:10 | |
| sigmavirus | In order of preference: pumpernickel rye, rye, pumpernickel | 18:11 |
| sigmavirus | other breads are just too dang sweet for me | 18:11 |
| jimbaker | sigmavirus, yeah, no sugar please in breads. unless it's a quick bread or something similar | 18:17 |
| jimbaker | (been trying to cut out added sugars in my diet in 2017. so far, working out pretty well) | 18:17 |
| tojuvone | I still eat everything I like and try to do sport | 18:18 |
| tojuvone | seems not workign that good anymore as getting age :D | 18:18 |
| jimbaker | tojuvone, hah | 18:18 |
| jimbaker | me, i'm probably doing about the same. but i'm just trying to avoid certain unnecessary things. most breads traditionally are not made with sugar. but these days in the US, they are, in part because sugar is a preservative | 18:20 |
| sigmavirus | and in part because sugar is added to all the foods in the US these days | 18:20 |
| sigmavirus | which is so unsettling | 18:20 |
| sigmavirus | I'm trying to follow an Atkins like diet (and have done so to great effect in the past) but it's hard | 18:21 |
| jimbaker | so one could go to an extreme: limited carbs. or another. eat more traditionally. i'm trying the latter approach. it takes a careful eye to sort this stuff out | 18:21 |
| sigmavirus | Yeah | 18:21 |
| tojuvone | tell me about it. In Austin it was so annying in hotel when all I could get for breakfast was sugar | 18:21 |
| sigmavirus | This diet doesn't work for my fiancée for example, she needs grains | 18:21 |
| sigmavirus | tojuvone: it's also cheaper when it has sugar in it, oddly enough | 18:21 |
| tojuvone | we have tax for sugar :) | 18:21 |
| jimbaker | sure, but a traditional muesli has no added sugar | 18:21 |
| jimbaker | so one can get complex grains. and if it has some dry fruit like raisins or dates, that's fine too. anyway, i'm trying "traditional diet" for 2017. any world tradition is fine. like to mix it up! | 18:23 |
| jimbaker | tojuvone, +1 about the sugar tax | 18:23 |
| tojuvone | ok, but back to Craton as short of time | 18:23 |
| jimbaker | yes indeed :) | 18:24 |
| tojuvone | while this was highly important discussion :D | 18:24 |
| jimbaker | (it's only our health!) so yesterday we were talking about auditing, which i think is close to what you need for notification support | 18:24 |
| jimbaker | https://blueprints.launchpad.net/craton/+spec/craton-notifications | 18:25 |
| jimbaker | we need to expand this blueprint out; come up with a spec; and part of that spec should include user stories from what is needed for NFV | 18:26 |
| tojuvone | ok | 18:26 |
| jimbaker | so user stories can include the discussion you had with us back in november re health checks; or performance constraints | 18:27 |
| *** VW has quit IRC | 18:28 | |
| tojuvone | what I would fast see is notification about host maintenance "state". I think that is important for everybody | 18:28 |
| tojuvone | I mean the part that some alarms should not be triggered from onitoring when host is actualyl in maintenance | 18:29 |
| *** VW has joined #craton | 18:29 | |
| tojuvone | wow, my typing is soo good | 18:29 |
| jimbaker | so we have some very early modeling about this in the note and active columns for devices | 18:30 |
| jimbaker | tojuvone, ;) | 18:30 |
| tojuvone | then that part is more "telco" that user should know about coming / ongoing maintenance by alarm | 18:30 |
| jimbaker | tojuvone, i always saw this modeling as sort of inadequate, but a start for such governance aspects. consider it a placeholder | 18:30 |
| tojuvone | but that is also if there is notification from Craton, "monitoring" could catch and send alarm here too | 18:31 |
| jimbaker | right. so i think the important thing here is to come up with some sort of state description that can be used in such automation | 18:32 |
| tojuvone | Do not know Monasca that much, but like Vitrage does RCA for different issues in cloud and raises alarm on everything related | 18:33 |
| jimbaker | i know the basic problem space for root cause analysis, although no detailed knowledge of vitrage; i have worked with monasca which is different | 18:34 |
| tojuvone | so maintenance could be addition to receiving "raw faults", anylysing that and sending alarms (I mean at least in vitrage) | 18:34 |
| jimbaker | so presumably we can share information with vitrage for RCA | 18:35 |
| tojuvone | Yes, Vitrage is quite new | 18:35 |
| tojuvone | anyhow still any monitoring might want to ignore some alarming when maintenance if not otherwise informs about it. I am not aware of billing here. I am still far from ops myself :) | 18:36 |
| jimbaker | tojuvone, we should have enough to make a start here | 18:38 |
| jimbaker | 1. notification events to indicate something is of interest. this will build on auditing | 18:39 |
| tojuvone | then I know there is other workflow engines. I do not know if somebody wants to use them. Like catching notification and do workflow somewhere else | 18:40 |
| jimbaker | 2. refine how we express our states. active/note are not sufficient for automated descriptions of maintenance | 18:40 |
| jimbaker | with that then | 18:40 |
| jimbaker | 1. capture of such events. may be into a workflow engine. (we are not building a workflow engine for process stuff.) or could be a pulling script for now | 18:41 |
| jimbaker | 2. use craton inventory to help with the RCA graph | 18:42 |
| jimbaker | 3. profit ;) | 18:42 |
| jimbaker | (with some sort of computation of course) | 18:42 |
| tojuvone | I was in some Mistral workflow session in Barcelona, but the demo failed. Then again I do not like tiding tons of projects doing different parts, gets more complicated and more room for error | 18:45 |
| jimbaker | tojuvone, yeah, it's a great question. i had assumed this was the place for mistral to do such workflows. maybe vitrage could do them instead, depending on how pluggable it is? | 18:47 |
| jimbaker | but that's why i suggested doing it in a script | 18:47 |
| jimbaker | because if it cannot be done in a cronjob, likely not going to work in some overly complicated workflow engine ;) | 18:48 |
| tojuvone | well, I think Vitrage does some simple API calls and not whole flows of actions | 18:48 |
| tojuvone | simple thing is when notice host fault and call nova (compute) force down | 18:48 |
| *** VW has quit IRC | 18:49 | |
| tojuvone | some real workflow should do different operation to "move" VMs depending what "user" wants to do with them | 18:49 |
| tojuvone | surely first step is that admin might just offer one static flow that does predifined actions always | 18:50 |
| tojuvone | to "move VMS" | 18:50 |
| tojuvone | more advanced case is that user get to interact (configure) way or the other. And still he wouldn't have rights, but admin would | 18:51 |
| jimbaker | tojuvone, right, so kicking off movement is something we can isolate from, do the movement (which is what sulo is working on) | 18:51 |
| jimbaker | tojuvone, if in craton, we will have pretty good support for that with the forthcoming rbac impl | 18:52 |
| jimbaker | pretty much can lock down anything that is a resource in craton | 18:52 |
| tojuvone | yes, I took a quick glance on that | 18:52 |
| jimbaker | such as "put this host in maintenance" | 18:53 |
| jimbaker | so i think this is pretty workable then. the really big piece here is just the dependency we need some sort of auditing support | 18:54 |
| tojuvone | then there is this going into maintenance soon | 18:54 |
| jimbaker | right, so that's part of a state transition graph to be defined | 18:55 |
| tojuvone | I mean in Telco at least, user wants to know that, so actually there is nothing running on host when time comes | 18:55 |
| jimbaker | basically what i would suggest is that we make "active" a computed column | 18:55 |
| jimbaker | there's some other column that captures how a device can from inactive to active | 18:56 |
| jimbaker | and back | 18:56 |
| jimbaker | let's not put time in the picture just yet (not certain if you were suggesting it should be, just seems to complicate) | 18:56 |
| *** VW has joined #craton | 18:56 | |
| jimbaker | but a transition of "running" -> "entering maintenance" (maybe with a note) -> "maintenance", etc | 18:57 |
| jimbaker | that could work | 18:58 |
| jimbaker | iirc, we currently use active to help with quiescence - this host no longer takes new VMs | 18:58 |
| jimbaker | sulo, ^^^, if you're around (not likely on an early fri evening!) | 18:59 |
| tojuvone | yes, the beef is just that user do not suddenly hit maintenance. We have complex chain of different VMs that might be interacting in subscriber session | 18:59 |
| jimbaker | tojuvone, no worries. if we need to refine this state transition to reflect, should be fine | 19:00 |
| jimbaker | my only possible concern is modeling "shutdown in 120 seconds" or something like that. mostly because i haven't thought about such modeling before, vs a simple state machine | 19:01 |
| *** VW has quit IRC | 19:01 | |
| tojuvone | well... how would one do maintenance for group of hosts? | 19:02 |
| tojuvone | would he like to predifene different time for different host | 19:02 |
| tojuvone | and that might be notified as planned maintenance time | 19:02 |
| tojuvone | like there never would be many hosts in mainteancne at the same time | 19:03 |
| jimbaker | tojuvone, that's just a group op | 19:03 |
| tojuvone | I really do not know how ops does this | 19:04 |
| jimbaker | tojuvone, for our public cloud work, we have been able to do this just with an active flag | 19:04 |
| tojuvone | ok | 19:04 |
| jimbaker | what we miss is the ability to use that for the rca aspects | 19:05 |
| jimbaker | so that's why i would describe it as being a *refinement* | 19:05 |
| tojuvone | ok | 19:05 |
| jimbaker | the literature of computer science has all sorts of stuff on "process calculuses" that allow for modeling time and interaction. but if we can avoid that sort of stuff, i think that would be best | 19:06 |
| jimbaker | certainly as a next step :) | 19:06 |
| tojuvone | Was just thinking if set some "planned maintenance time", at that hour it would just roll automatically. (if not cancelled or time redifined) | 19:07 |
| jimbaker | tojuvone, we could certainly set a time-to-live | 19:07 |
| jimbaker | but i was thinking all these changes are best done by an external tool that uses craton | 19:07 |
| tojuvone | I do not have strong opinion and also no ops experience. The importance was jsut for user to know early enough | 19:08 |
| jimbaker | that's the "script" (or set of scripts, eventually to be run by a workflow engine) that would work with craton here | 19:08 |
| jimbaker | tojuvone, yeah, we can make that story happen | 19:08 |
| tojuvone | well that is the notif part, but also talked earlier with Nova that servers API thingy. | 19:10 |
| tojuvone | Anyhow that is also more advanced use case if looking we have already notif | 19:10 |
| *** VW has joined #craton | 19:11 | |
| tojuvone | Meaning "host maintenance state" could be seen trough Nova servers API by having like to Craton API. | 19:12 |
| jimbaker | sure | 19:13 |
| tojuvone | Looking the time frame I see this coming later if feasible. | 19:13 |
| tojuvone | like -> link | 19:14 |
| jimbaker | the big dependency is just getting the auditing/notification done. refining the active column is straightforward, once we know the states | 19:14 |
| tojuvone | I should have IRC with spell check on a fly | 19:14 |
| jimbaker | we will also want to make it easier to do that linkage as well (that's the virtualized variable support) | 19:14 |
| jimbaker | ;) | 19:14 |
| tojuvone | using xchat btw, should there be better client | 19:15 |
| *** VW has quit IRC | 19:15 | |
| jimbaker | presumably possible with emacs... | 19:15 |
| jimbaker | but on ubuntu, i use xchat. on mac, i use textual | 19:16 |
| jimbaker | my good enough tooling i suppose | 19:16 |
| tojuvone | ok, then I should not be that far off with my choice :) | 19:17 |
| tojuvone | you know it is pretty hard to come up with everything myself, as nobody else working with similar stuff at office | 19:18 |
| jimbaker | no worries, we will help each other figure it out | 19:18 |
| tojuvone | But that is all so great also. New stuff :) | 19:19 |
| jimbaker | so i think we have enough to really start working out the blueprint with respect to notifications; and another one to describe maintenance | 19:21 |
| tojuvone | Thanks for all this great discussion. Really great as it is not easy for me to have these hours for "work". | 19:21 |
| jimbaker | i will also take a detailed look at vitrage | 19:22 |
| jimbaker | tojuvone, thanks for your great input! i really look forward to working with you more on this | 19:22 |
| tojuvone | thanks | 19:22 |
| tojuvone | By fast I think at least Thursday works next week. Kid has floorball at same time with meeting and I can take laptop there | 19:24 |
| tojuvone | jimbaker: Also look forward getting all this and trying to help you as I can. | 19:25 |
| jimbaker | tojuvone, sounds great! | 19:26 |
| *** VW has joined #craton | 19:29 | |
| *** valw has quit IRC | 20:51 | |
| *** valw has joined #craton | 20:58 | |
| *** jovon has joined #craton | 21:00 | |
| *** valw has quit IRC | 21:21 | |
| *** valw has joined #craton | 22:16 | |
| *** VW_ has joined #craton | 22:39 | |
| *** VW has quit IRC | 22:42 | |
| *** valw has quit IRC | 22:42 | |
| *** VW_ has quit IRC | 22:43 | |
| *** jovon has quit IRC | 23:10 | |
| *** VW has joined #craton | 23:18 | |
| *** VW has quit IRC | 23:24 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!