[00:20] thumper: review my race fix plz? https://github.com/juju/juju/pull/8698 [00:47] or wallyworld? ^ [00:47] oh, and this one too https://github.com/juju/juju/pull/8688 [02:06] * thumper looks [02:10] babbageclunk: first one done [02:10] ta [02:11] babbageclunk: other done [02:11] awesome, fanks thumper [02:12] np [02:14] wallyworld: up for a chat about the review any time [02:15] thumper: I thought the unsubscribe would be enough too, but it wasn't. [02:15] um... [02:15] I beg to differ [02:15] chat? [02:17] I mean, feel free to try it - I put the unsubscribe in first, and it didn't fix the race. [02:18] thumper: The other option that worked was to put a mutex into the suite and lock around both s.reqs = make(chan) and the send in the callback. [02:18] But the local variable seemed cleaner. [02:19] thumper: in 1:1? [02:19] coming [02:29] thumper: just got back from meeting witn anastasia, ping when you are free [02:29] wallyworld: ack [02:38] wallyworld: now? [02:38] sure [02:41] thumper: am in HO when you are ready [03:42] wallyworld or thumper, another one to take a look at? https://github.com/juju/juju/pull/8699 [03:43] * thumper looks [04:05] thumper: thanks for the glowing review! [04:05] :) [04:13] jam: has anyone looked at the network PR yet? [04:14] jam: wow... not a small change then [04:16] thumper: yeah. Manadart looked at some of it already [04:16] thumper: most of that is lots more tests [04:18] jam: I guess it looks good to me, although I'm not sure if I'm a good judge or not [04:19] jam: re: https://github.com/juju/juju/pull/8687 - like your comment on whether or not we should have different params structs for v5 and latest [04:19] or just add new field to params [04:24] thumper: thanks for the approval. Ultimately it needs to be tested, and I can at least test the VLAN stuff directly, and I think someone just gave me a bonded machine to test that part on. [04:24] sweet [04:25] hmmm... wondering why our windows reboot tests are trying to initialize a lxd container manager [04:25] that seems dumb [04:48] thumper: I think new types are better. It means generating automated descriptions of API preserves the shape of old APIs [04:48] * thumper nods [04:54] thumper: i would like a second opinion on the use of a bulk remove vs creating a slice of individual removes using removeInCollectionOps() in https://github.com/juju/juju/pull/8700 [04:54] to remove the credentials for a deleted cloud [05:22] fuck fuck fuckity fuck [05:22] * thumper loses the will to keep going today [05:22] will attack tomorrow with gusto... [05:23] laters peeps === frankban|afk is now known as frankban [08:18] externalreality: Want to cast an eye over this one for me? https://github.com/juju/juju/pull/8703 [08:19] Not here. jam ^. Will get on to yours next. [08:20] manadart: not here? [08:20] Eric. [08:48] jam: Approved #8697. Looks like nothing new on OCI. I assume we're taking a pass again... [08:49] yeah [09:38] jam: OK if I forward port https://github.com/juju/juju/pull/8690 ? [09:39] manadart: any reason to not just merge 2.3 into develop? [09:40] jam: Suppose not. I'll give it a lash. [09:55] https://github.com/juju/juju/pull/8704 [12:12] o/ [12:21] Does anyone know of a charm/layer for tensorRT ? [12:28] how do I show the results of an action run with the run-action command? is that juju show run-action by chance? [12:29] figured it out. [12:30] my next question, if an action completes successfully, it updates the config file correct? [12:40] just to clarify, I send all my log files to filebeats in logstash on its port to import logs into ELK correct? kwmonroe [12:41] ooh, and its working! time to feed it o.o [14:11] any update on when we might get 2.3.8 stable? [14:15] admcleod_: working on it, the team's successfully gotten a container in bionic on a vlan and tossing the current updates at the full test suite to see what other gaps needs to be fixed up [14:15] awesome [14:15] thanks [14:31] am I the only one having severe issues with elasticsearch at the moment? [14:32] nothing seems to be able to connect to elasticsearch, regardless of what application I deploy and install. [15:30] rick_h_: so I was looking at a bundle for ELK, andit got me thinking. If i simply take the bundle, and install everything piecemeal, that would allow me to bypass the restriction on deploying bundles with the --to option correct? [15:31] bobeo: sure, but the what restriction are you hitting with --to now? [15:31] I'm trying to deploy arm64 nodes using an amd64 controller and juju-2.4 beta2 [15:31] rick_h_: --to tells me i cant deploy bundles using the --to option. [15:31] I should just need to configure agent-stream=devel on the model right? [15:32] bobeo: oh right, I mean a bundle has --to placement stuff in it so putting a bundle on a machine doesn't work out right [15:32] bobeo: so yea, you just break it down and run the manual commands you want [15:32] rick_h_: so when I was previously trying to deploy elk, it was with --to lxd:, whch was failing, and when I tried to deploy it as is, its when it broke everything, as in I hit that bug that I found yesterday. So I decided today to open the bundle.yaml to try and better understand bundles. [15:33] rick_h_: right. so my question is, how do you translate the relations sections of the yaml files to understand what relations are requirements for which charm? [15:33] bobeo: sec, otp [15:45] bobeo: can you pastebin the bundle that you're working with? [16:12] for instance, in the bundle.yaml for ELK, you have this: - - "openjdk:java" - "logstash:java" - - "kibana:rest" - "elasticsearch:client" - - "logstash:elasticsearch" - "elasticsearch:client" [16:12] kwmonroe: sure! lnk is easier though https://api.jujucharms.com/charmstore/v5/~elasticsearch-charmers/elk-stack/archive/bundle.yaml [16:13] kwmonroe: as for the above, im assuming the -- and the - mean they are related, so for instance, juju relate openjdk:java logstash:java would be one relational pairing [16:13] with each set of -- and - being another pair. [16:14] cool bobeo. you asked how to translate the relation section of the bundle yaml.. for that bundle, you would achieve the same thing with juju from the CLI like this: [16:14] juju relate openjdk:java logstash:java [16:14] juju relate kibana:rest elasticsearch:client [16:14] juju relate logstash:elasticsearch elasticsearch:client [16:14] the -- and - are just the way bundles in the store format that section of yaml [16:15] so the first "-" means here comes a relation, and it's between - openjdk and - logstash. [16:15] kwmonroe: awesome! yay im learning! it feels good to be able to finally realize im understanding how the yamls interact to build charms. [16:16] :) good to hear! [16:17] ok so jumping in further, kwmonroe looking at the config files, I have realized alot of things that youd configure in the files arent always there. so my question is as I go into the files to configure them, if I reset the box, will it blow away those required changes that werent in the charm that I had to add later, or will they be rolled back, even though they arent in the charm config file [16:18] my concern for instance with ELK, I had to go into logstash, kibana, and elastic to make changes that werent included by default with filebeat, kibana, and elastic. While I am still having issues, I hav efewer ones now. If I reset the boxes, I dont want to lose my current configs as I move closer to it working as intended. [16:20] http://paste.ubuntu.com/p/dPmwsHDvMn/ http://paste.ubuntu.com/p/jbR33H84Pc/ http://paste.ubuntu.com/p/jnCBKcZfPk/ http://paste.ubuntu.com/p/3yJFqx67JY/ for reference, juju status, kibana, logstash, and elasticsearch respectively [16:21] kwmonroe: I understand that juju manages the file(s) to manage configurations, or deploys by snaps, but additional configurations will sometimes be required. [16:24] bobeo: if you manually ssh to a unit and update, for example, /etc/logstash/logstash.conf, those changes may very well get overwritten by juju on a subsequent hook. [16:24] bobeo: best practice is to let juju handle the config, which may mean updating the charm to expose config that isn't available. [16:24] bobeo: what kinds of things are you manually configuring? [16:25] logstash, filebeats, and hopefully client and admin systems for elastic so it will work. currently its not responding to web requests or curl requests, which is why mine isnt working [16:26] I wish I knew how to do that. I know how to configure these systems via generic install with ease, but not with juju. if you guys can show me how, I can do a lot to contribute, even if they might need a few tweeks. [16:26] I would love to give back kwmonroe [16:27] sure bobeo -- just tell me what you've had to change. you mentioned elastic wasn't responding to curl -- did you change something in /etc/elasticsearch/elasticsearch.yml to make it work for you? [16:28] similarly with filebeat -- was it something in /etc/filebeat/filebeat.yaml that you had to configure so that logs would ship to logstash? [16:29] kwmonroe: no, its still not responding. Ive had issues with elasticsearch in graylog and in ELK, making me think elasticsearch is having issues. graylog says elastic 6x isnt workign with graylog 2.4 atm, and ELK seems to be having issues with elastic as well, at least as far as I can tell so far. Logstash and kibana are working fine for me. [16:31] kwmonroe: yea, I had to make changes in filebeat. specifically I had to add elasticsearch IP from localhost [16:32] for some reason, it had elasticsearch uncommentd, with correct port, but pointed at local host instead of at the IP for elasticsearch [17:06] rick_h_: where can i find the agent stream on the web? are agents for non x86 architectures built for non released versions of juju? [17:09] jhobbs: typically no we don't. We must at some point but not sure if beta or rc is when that kicks in. I've got a question out but not sure yet. [17:14] jhobbs: ok found http://streams.canonical.com/juju/tools/agent/2.4-beta2/ [17:14] But have to find the stream file to reference it === frankban is now known as frankban|afk [17:15] cool, arm64 agent is what i'm after [17:15] is that devel? proposed? [17:20] jhobbs: searching the various stream files I can see. I don't have it yet. I'm not up on which one is used. [17:20] ok no problem [17:21] i can try them all! [17:21] thanks rick_h_ [17:23] Yea at least feel better I can see arch built agents [17:23] yeah, and now i know to test with beta2 instead of edge [17:23] so i'm in a happier place [17:30] hey rick_h_ is there a way to marry both graylog and elk together? I mean, I like the way graylog handles data a lot better than ELK, especially with indexing, and I like its ease of use, but I like the UI better frrom older kibana instances. is this possible you think? [17:32] jhobbs: ok, they're listed in http://streams.canonical.com/juju/tools/streams/v1/com.ubuntu.juju-devel-tools.sjson [17:32] bobeo: sorry no, I'm not familiar enough with graylog/elk unfortunately. [17:37] rick_h_: woo hoo! thank you [18:08] bobeo: what do you mean marry the two together? [18:50] bdx: so there are several really great portions of both, for instance kibana, I love kibana, way more than the graylog UI, and I love how graylog does indexing, and handles logs, way more than how elk does it. Id like to take the kibana UI, put it on graylog, and use how graylog imports and processes logs, as compared to logstash, which it might be using? Im still digging through graylog docs on what its actually "made of/from" [18:57] bobeo: you can connect kibana to the elasticsearch that you use for graylog and have what you desire [18:58] bdx: I did that actually, remember that "X" i wanted to build? I built that today. Now I just need to make sure elastic is working properly, but im hitting issues. Its not responding to curl requests. [18:59] bdx: it, as in ElasticSearch, my current deployment of it. http://paste.ubuntu.com/p/tKTTVkQ6Mn/ [19:56] bobeo: fwiw, i took the elk-stack bundle and swapped logstash for graylog.. the result is https://jujucharms.com/u/kwmonroe/egk-stack/ [19:57] bobeo: it's totally a dev bundle, but it illustrates how you might swap ELK components to your liking.. (logstash for graylog, in this case) [20:00] bobeo: fwiw, i don't have any trouble curl'ing ES:9200 from within the cluster: [20:00] $ juju run --unit elasticsearch/0 'curl -XGET http://localhost:9200/_cluster/health' [20:00] {"cluster_name":"elasticsearch","status":"yellow","timed_out":false,"number_of_nodes":1,"number_of_data_nodes":1,"active_primary_shards":5,"active_shards":5,"relocating_shards":0,"initializing_shards":0,"unassigned_shards":1,"delayed_unassigned_shards":0,"number_of_pending_tasks":0,"number_of_in_flight_fetch":0,"task_max_waiting_in_queue_millis":0,"active_shards_percent_as_number":83.33333333333334} [20:00] i mean, yeah it's yellow, but yellow ain't red! [20:03] kwmonroe: can I create bundles from local charms? [20:03] you betcha TheAbsentOne! [20:04] TheAbsentOne: have a look at https://jujucharms.com/hadoop-processing/ see 'bundle-local.yaml' in the side bar [20:04] TheAbsentOne: you'll note the "charm:" directive includes a local path for the charms that you want to pull from your local filesystem... you'd deploy that one with "juju deploy bundle-local.yaml", but substitue whatever your bundle name is. [20:06] ahn awesome since you guys were talking about bundles I was asking myself that question xD [20:07] thanks for the info kwmonroe [20:07] np [20:13] Morning o/ [20:20] nmorning veebers [20:21] hi rick_h_ o/ [20:22] rick_h_: saw your email, digging in now [20:22] veebers: cool, definitely let me know if there's something I should see/walk me through something. Gave it a go but still a lot to learn there. [20:49] rick_h_: right, so I figured out why the cleanup wasn't working and updated lxd on grumpig. (still from xenial backports, haven't moved over to snap yet :-\ [20:50] veebers: k, by cleanup you mean old lxds from the failed runs? [21:02] thumper: you coming to caas standup? [21:02] wallyworld: no, I have another call [22:28] wallyworld: https://github.com/juju/collections/pull/1 [22:28] looking [22:29] thumper: lgtm, so we only had set and dequeue? [22:30] wallyworld: https://github.com/juju/names/pull/88 [22:30] yes [22:30] thumper: was making a tea, yep you want to hangout? [22:30] veebers: yeah [22:30] 1:1? [22:30] sounds good omw [22:30] thumper: why oh why doesn't go have generics :-( [22:31] uh ha [22:33] thumper: done [23:27] wallyworld: https://github.com/juju/proxy/pull/1 and yes, I'm leaving the years the same [23:27] * thumper heads to physio [23:27] bbs [23:47] kelvinliu__: there's a couple of more things to remove which I've mentioned in the PR [23:49] wallyworld, yes, i saw the comments, i will fix them soon, thanks.