[00:00] <_mup_> ensemble/expose-provisioning r253 committed by jim.baker@canonical.com [00:00] <_mup_> Make explicit that a couple of methods are only to be used for testing [00:39] <_mup_> ensemble/unit-agent-resolved r275 committed by kapil.thangavelu@canonical.com [00:39] <_mup_> complete the change over to more transitions, fill out coverage [00:48] <_mup_> ensemble/unit-agent-resolved-part-two r274 committed by kapil.thangavelu@canonical.com [00:48] <_mup_> merge and resolve conflict [01:01] <_mup_> ensemble/unit-agent-resolved-part-two r274 committed by kapil.thangavelu@canonical.com [01:01] <_mup_> redo merge resolve conflict, grabbed the wrong resolved file [02:11] Yo! [02:33] hazmat: ping [02:33] niemeyer, pong [02:33] hazmat: Yo [02:33] hazmat: I'm having some trouble to review the new unit-agent-resolved, for lack of a good based to compare it with [02:33] hazmat: ensemble-resolved, which is the pre-requisite, doesn't include the trunk yet [02:33] hazmat: the trunk doesn't include ensemble-resolved [02:34] hazmat: and unit-agent-resolved contains both [02:34] niemeyer, no.. it doesn't, unit-agent-resolved is after ensemble-resolved in my pipeline [02:34] hazmat: So no matter what's the base, there are cross-changes [02:34] hazmat: Yes it is, thus, it includes the changes from it [02:34] niemeyer, the base is ensemble-resolved [02:34] hazmat: Yes it is [02:34] i need to push my latest which merges trunk [02:34] of ensemble-resolved [02:35] if you want to utilize it as a base [02:35] hazmat: unit-agent-resolved contains both the latest from trunk and the latest from ensemble-resolved [02:35] niemeyer, yes [02:35] hazmat: I can't compare it with either without getting irrelevant changes [02:35] hazmat: (in the diff) [02:36] niemeyer, the diff to ensemble-resolved looked fine to me [02:36] hazmat: The solution is to handle the ensemble-resolved merge on trunk, or to update it [02:36] -html_static_path = ['_static'] [02:36] +#html_static_path = ['_static'] [02:36] hazmat: That's part of that diff [02:36] hazmat: (and a lot of other things from trunk) [02:37] pushing the latest ensemble-resolved [02:37] hazmat: That should handle it if you have trunk merged on it, thanks [02:37] niemeyer, let me know if that fixes the issue [02:37] its pushed [02:38] that's current with trunk [02:38] hazmat: Yep, that did it, thank you [02:39] niemeyer, cool, np [03:01] Phew, 2 to go [03:01] Tomorrow, though [03:15] <_mup_> ensemble/config-set r216 committed by bcsaller@gmail.com [03:15] <_mup_> checkpoint [03:53] <_mup_> ensemble/auto-dependency-resolution r217 committed by kapil.thangavelu@canonical.com [03:53] <_mup_> flesh out some tests using a new test ease of use api. [04:17] <_mup_> ensemble/expose-provisioning r254 committed by jim.baker@canonical.com [04:17] <_mup_> Test added to verify opened port parsing [09:05] morning folks [10:37] <_mup_> ensemble/config-get r219 committed by bcsaller@gmail.com [10:37] <_mup_> use get_formula_state in get_relation_endpoints, removes duplicate code [10:39] <_mup_> ensemble/config-get r220 committed by bcsaller@gmail.com [10:39] <_mup_> rename hook's local_* methods as get_local_* [10:39] <_mup_> expanded test coverage [10:41] <_mup_> ensemble/config-get r221 committed by bcsaller@gmail.com [10:41] <_mup_> merge trunk [10:44] <_mup_> ensemble/trunk r219 committed by bcsaller@gmail.com [10:44] <_mup_> Merge config-get [r=niemeyer] [f=776605] [10:44] <_mup_> Merge support for config-get cli tool for use in hooks. This makes local service options available within hooks. [11:40] config-get lovely :) [14:01] <_mup_> Bug #777816 was filed: Ensemble should resolve dependencies upon deploy < https://launchpad.net/bugs/777816 > [14:27] hazmat: Morning! [14:28] niemeyer, morning [14:30] niemeyer: hazmat morning [14:30] kim0, top of the morning [14:30] hehe [14:31] morning always seems a bit arbitrary given the our globally distributed nature [14:31] Yeah .. maybe we should get a bot for translating timezone salutations [14:32] from a symbolism perspective, morning is nice though, expressing at hope and joy at the beginning of a new day [14:37] hmm.. puppet's changing licensing to apache [14:41] so how do I have to bribe to merge my doc branches :) [14:41] s/how/who/ [14:49] kim0, i'll do it now [14:49] niemeyer, do you know if the bzr --author flag on commit takes a full name or an lp name? [14:50] hazmat: I suppose it takes a full name [14:50] hazmat: Or, actually [14:50] hazmat: I think this is just metadata [14:50] niemeyer, it is just metadata, but its displayed onlp [14:50] niemeyer, the gopher ppa has the latest go packages? [14:50] hazmat: Displayed as is, I think [14:50] hazmat: It does [14:50] hazmat: ppa:gophers/go [14:51] niemeyer, thanks [14:51] hazmat: np [14:51] Actually, I have to send those packages again.. some third-party libraries got included by mistake [14:56] <_mup_> ensemble/trunk r220 committed by kapil.thangavelu@canonical.com [14:56] <_mup_> merge kim0's new user guide and faq [a=kim0][r=niemeyer][f=734975] [14:56] <_mup_> Provides a tutorial within the documentation taking the user through [14:56] <_mup_> the process of getting started with Ensemble as well as an FAQ. [14:57] kim0, done, and your noted as the author on lp and bzr history [14:59] kim0, thanks again, that was much needed [15:10] <_mup_> ensemble/auto-dependency-resolution r218 committed by kapil.thangavelu@canonical.com [15:10] <_mup_> additional test for solving deps from both environment and formula repos. [15:20] doing some bug milestone management, sorry about the noise [15:31] <_mup_> ensemble/resolved-state-api r204 committed by kapil.thangavelu@canonical.com [15:31] <_mup_> merge trunk [15:33] niemeyer, should i change the relation resolved api now (it doesn't need the retry flag).. or just wait, since the entire stack uses the api.. [15:33] hazmat: thanks man [15:34] kim0, np [15:35] hazmat: Sorry, ECONTEXT [15:35] hazmat: What's that about again? [15:36] niemeyer, sure.. the set_relation_resolved(relation_map) really just needs to be set_relation_resolved(relation_list) .. since retry hook has no meaning [15:36] hmm.. i think i suggested that i just make the cli barf in this case [15:36] i guess that api has other pending cleanups as well though [15:36] hazmat: Do we plan to make it retry in the future? [15:36] niemeyer, hmm.. good question [15:37] niemeyer, its concievable.. but not nesc. correct [15:37] a hook execution may be obsolete in a relation context [15:37] by the time we want to retry it [15:39] i go ahead cli fix, and note is as something that needs adjustment as part of the staterequest protocol work [15:39] ^it [15:41] hmm. [15:41] i guess i can try it and see how much work it really is [15:44] niemeyer, thoughts on retry? [15:45] uds sessions are posted [15:52] hi all [15:53] hey [15:54] kim0: am at https://code.launchpad.net/~kim0/ensemble/user-tutorial-and-FAQ/+merge/58861 will buzz you later :) [15:55] koolhead17: it's been merged .. http://people.canonical.com/~niemeyer/ensemble/user-tutorial.html [15:56] ok :) [15:58] hazmat: Sorry, was answering that message in the cloud ML [15:58] hazmat: Hmm [15:59] hazmat: So, let's try to come up with a scenario [15:59] hazmat: DNS goes off for whatever reason [15:59] hazmat: db-relation-changed explodes [15:59] hazmat: Intended change done by the other side of the relation isn't acted upon [15:59] hazmat: No other changes are done [16:00] hazmat: DNS goes live [16:00] hazmat: ensemble resolved unit/0 db [16:00] hazmat: Change is lost [16:00] niemeyer, yes [16:00] niemeyer, we could capture enough information to retry the hook [16:00] but then say unit/0 departs [16:01] should we be retrying a hook for something which doesn't exist anymore? [16:01] hazmat: How would that take place? [16:01] hazmat: resolved shouldn't work on a relation which doesn't exist (!?) [16:01] niemeyer, the relation exists but the remote peer that caused the change has departed [16:01] s/peer/unit [16:03] hazmat: Hmm [16:03] hazmat: Thinking [16:03] some of this might be more clear if we do on disk persistence and recovery, since the merge of last known to current is applicable [16:04] hazmat: Hmm.. maybe [16:04] niemeyer, i'll leave the api as is for now [16:04] hazmat: Sounds good [16:05] hazmat: Can you please [trivial] a comment explaining that debate please? [16:05] hazmat: Just enough to give us context for following on that conversation at a latter time [16:05] niemeyer, sure in the commit, merge proposal, bug report, code ? [16:05] hazmat: Code feels more visible.. next to the API which we should change [16:05] niemeyer, sounds good [16:05] hazmat: Thanks [16:16] <_mup_> ensemble/expose-provisioning r255 committed by jim.baker@canonical.com [16:16] <_mup_> Merged trunk & resolved conflicts [16:16] <_mup_> ensemble/expose-dummy-provider r217 committed by jim.baker@canonical.com [16:16] <_mup_> Merged trunk [16:23] * hazmat lunches [16:25] <_mup_> ensemble/expose-dummy-provider r218 committed by jim.baker@canonical.com [16:25] <_mup_> Use fail/succeed instead of inlineCallbacks for dummy methods w/o internal yield needs [16:28] hazmat: Enjoy [16:28] <_mup_> ensemble/expose-provisioning r256 committed by jim.baker@canonical.com [16:28] <_mup_> Merged upstream [16:37] One branch pending [16:38] jimbaker: Will review your lunch after lunch [16:38] Erm [16:38] jimbaker: Will review your branch after lunch [16:38] "Will review your lunch after branch" would be nice too [16:40] niemeyer, i believe i'm to have some sort of chicken soup to deal with my cold symptoms... i will see what my mother-in-law prepares ;) [16:40] jimbaker: ;-) === niemeyer is now known as niemeyer_lunch [16:47] <_mup_> ensemble/trunk r221 committed by jim.baker@canonical.com [16:47] <_mup_> merge expose-dummy-provider [r=niemeyer][f=766241] [16:47] <_mup_> Adds dummy provider support for port opening/closing. [16:48] <_mup_> ensemble/expose-provisioning r257 committed by jim.baker@canonical.com [16:48] <_mup_> Merged trunk [16:49] <_mup_> ensemble/expose-hook-commands r249 committed by jim.baker@canonical.com [16:49] <_mup_> CLI commands [16:50] <_mup_> ensemble/expose-hook-commands r250 committed by jim.baker@canonical.com [16:50] <_mup_> Merged upstream & resolved conflicts [17:14] <_mup_> ensemble/resolved-state-api r205 committed by kapil.thangavelu@canonical.com [17:14] <_mup_> document set_relation_resolved api todo discussion [17:18] <_mup_> ensemble/resolved-state-api r206 committed by kapil.thangavelu@canonical.com [17:18] <_mup_> doc string cleanups per review. [17:47] <_mup_> ensemble/trunk r222 committed by kapil.thangavelu@canonical.com [17:47] <_mup_> merge resolved-state-api [r=niemeyer][f=767762] [17:47] <_mup_> Provides a get/set/clear/watch state api for marking unit errors as resolved. [17:48] <_mup_> ensemble/trunk-merge r191 committed by kapil.thangavelu@canonical.com [17:48] <_mup_> merge trunk === niemeyer_lunch is now known as niemeyer [17:54] <_mup_> ensemble/ensemble-resolved r259 committed by kapil.thangavelu@canonical.com [17:54] <_mup_> update code comments per review comment [17:57] <_mup_> ensemble/trunk r223 committed by kapil.thangavelu@canonical.com [17:57] <_mup_> merge ensemble-resolved [r=niemeyer][f=767948] [17:57] <_mup_> Provides a new ensemble resolved subcommand for marking unit errors [17:57] <_mup_> as resolved. [18:16] Alright, last branch in review! [18:26] niemeyer, i've got some comments on the unit-agent-resolved review that i would like to discuss [18:26] already attached to the review [18:41] hazmat: Ok.. I'll need some time to go over this review I'm doing [18:41] <_mup_> ensemble/unit-agent-resolved r279 committed by kapil.thangavelu@canonical.com [18:41] <_mup_> address review comments re formatting === deryck is now known as deryck[lunch] [19:14] jimbaker: I really wish this branch was pushed in a more fine grained way [19:15] I'm not sure I'll be able to get through it today [19:16] niemeyer, there is an obvious split between the service changes and the specific support for provisioning [19:17] although the service changes are all very much boiler plate cut & paste [19:17] however, if you want me to split the branch in that way, i can certainly do so [19:17] jimbaker: Yeah, hundreds of lines of boilerplate ;) [19:18] jimbaker: That'd take even more time [19:18] niemeyer, i don't like any of it, but hopefully post budapest it can be removed [19:18] jimbaker: The idea is to not get into this state [19:18] jimbaker: As you said, there are several split points.. [19:18] jimbaker: Pushing such a huge branch 3 days before UDS isn't good [19:38] niemeyer, jimbaker, bcsaller test api https://pastebin.canonical.com/47175/ [19:39] i'm implementing parts as part of the dependency resolver work [19:39] most of it is cribbed from other test helpers we have around === deryck[lunch] is now known as deryck [21:02] <_mup_> ensemble/trunk r224 committed by kapil.thangavelu@canonical.com [21:02] <_mup_> merge unit-agent-resolved [r=niemeyer][f=767961] [21:02] <_mup_> Provides a unit lifecycle support for processing resolved unit relations. [21:35] * niemeyer => break === niemeyer is now known as niemeyer_break [21:38] <_mup_> ensemble/unit-agent-resolved-part-two r279 committed by kapil.thangavelu@canonical.com [21:38] <_mup_> update unit agent to use correct transition depending on retry setting. [21:43] <_mup_> ensemble/trunk r225 committed by kapil.thangavelu@canonical.com [21:43] <_mup_> merge unit-agent-resolved-part-two [r=niemeyer][f=776014] [21:43] <_mup_> Provides unit agent support for transitioning resolved units and unit relations. [21:50] <_mup_> ensemble/debug-hook-scope-guard r211 committed by kapil.thangavelu@canonical.com [21:50] <_mup_> merge trunk [22:03] <_mup_> Bug #778134 was filed: debug hook seems to have lost hook cli context when using hook api < https://launchpad.net/bugs/778134 > [22:03] we really do need to have ensemble changed in the environment [22:03] or use byobu.. its hard to tell what the debug is actively working on otherwise [22:09] sigh.. we need more release testing, or feature freezes before a release. so many new features this release, i'm seeing a few behavior regressions [22:11] hazmat, agreed - i haven't fired up AWS recently since i'm working on stuff that just needs unit testing, *for now* [22:11] but experience has shown repeatedly that we don't anticipate enough in our unit tests [22:14] <_mup_> ensemble/debug-hook-scope-guard r212 committed by kapil.thangavelu@canonical.com [22:14] <_mup_> switch to using byobu [22:56] * niemeyer_break waves === niemeyer_break is now known as niemeyer [23:04] <_mup_> ensemble/trunk r226 committed by kapil.thangavelu@canonical.com [23:04] <_mup_> merge debug-hook-scope-guard [r=niemeyer][f=776014] [23:04] <_mup_> Enables live hook debugging via screen/byobu, with better start/end condition handling. [23:05] <_mup_> ensemble/expose-hook-commands r251 committed by jim.baker@canonical.com [23:05] <_mup_> Hook command skeleton, plus parse_port_protocol for that arg [23:58] * niemeyer => dinner