December 31, 2018 
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | >
December 31, 2018 [00:05:40] *** jcea <jcea!~Thunderbi@2001:41d0:1:8a82:7670:6e00:7670:6e00> has quit IRC (Quit: jcea)[00:21:20] *** noahmehl <noahmehl!~noahmehl@dsl-74-83-72-196.fuse.net> has quit IRC (Quit: My MacBook has gone to sleep. ZZZzzz…)[00:33:38] *** noahmehl <noahmehl!~noahmehl@dsl-74-83-72-196.fuse.net> has joined #smartos[01:05:31] *** andy_js <andy_js!~andy@94.3.60.133> has quit IRC (Quit: andy_js)[01:14:13] *** lgtaube <lgtaube!~lgt@46.165.251.115> has quit IRC (Ping timeout: 245 seconds)[01:24:41] *** fxhp <fxhp!~fox@138.207.192.33> has joined #smartos[01:40:42] *** Gathis <Gathis!~TheBlack@unaffiliated/gathis> has quit IRC (Remote host closed the connection)[01:43:56] *** jcea <jcea!~Thunderbi@2001:41d0:1:8a82:7670:6e00:7670:6e00> has joined #smartos[01:56:22] *** jcea <jcea!~Thunderbi@2001:41d0:1:8a82:7670:6e00:7670:6e00> has quit IRC (Quit: jcea)[02:36:26] *** noahmehl <noahmehl!~noahmehl@dsl-74-83-72-196.fuse.net> has quit IRC (Quit: Textual IRC Client: www.textualapp.com)[02:39:34] *** v_a_b <v_a_b!~volker@p4FECCEC8.dip0.t-ipconnect.de> has quit IRC (Ping timeout: 250 seconds)[03:28:43] *** ingenthr <ingenthr!~ingenthr@47.150.213.46> has joined #smartos[04:10:17] *** jubal_ <jubal_!~jubal@154-3-237-24.gci.net> has quit IRC (Read error: Connection reset by peer)[04:46:20] *** jubal <jubal!~jubal@154-3-237-24.gci.net> has joined #smartos[05:26:41] *** Kurlon <Kurlon!~Kurlon@98.13.72.207> has quit IRC (Remote host closed the connection)[06:28:12] *** ingenthr <ingenthr!~ingenthr@47.150.213.46> has quit IRC (Quit: Leaving.)[06:29:07] *** ingenthr <ingenthr!~ingenthr@47.150.213.46> has joined #smartos[08:45:42] *** ingenthr <ingenthr!~ingenthr@47.150.213.46> has quit IRC (Quit: Leaving.)[08:59:59] *** ingenthr <ingenthr!~ingenthr@47.150.213.46> has joined #smartos[09:00:15] *** ingenthr <ingenthr!~ingenthr@47.150.213.46> has quit IRC (Client Quit)[09:29:50] *** ingenthr <ingenthr!~ingenthr@47.150.213.46> has joined #smartos[09:32:58] *** mpana <mpana!~mpana@89.136.157.123> has quit IRC (Ping timeout: 245 seconds)[09:33:39] *** mpana <mpana!~mpana@89.136.157.123> has joined #smartos[10:00:16] *** ingenthr <ingenthr!~ingenthr@47.150.213.46> has quit IRC (Quit: Leaving.)[10:30:21] *** ingenthr <ingenthr!~ingenthr@47.150.213.46> has joined #smartos[10:30:35] *** ingenthr <ingenthr!~ingenthr@47.150.213.46> has quit IRC (Client Quit)[10:31:31] *** Latr_work <Latr_work!uid4039@gateway/web/irccloud.com/x-tzlibqwggwwifaux> has joined #smartos[10:38:22] *** jellydonut <jellydonut!~jelly@148.122.187.2> has quit IRC (Read error: Connection reset by peer)[10:56:34] *** andy_js <andy_js!~andy@94.3.60.133> has joined #smartos[11:00:06] *** bens1 <bens1!~bens1@161.122.2.81.in-addr.arpa> has joined #smartos[11:52:54] *** danmcd <danmcd!~danmcd@static-71-174-113-16.bstnma.fios.verizon.net> has quit IRC (Read error: Connection reset by peer)[11:53:29] *** danmcd <danmcd!~danmcd@static-71-174-113-16.bstnma.fios.verizon.net> has joined #smartos[11:54:25] *** mmlb <mmlb!~mmlb@76-248-148-178.lightspeed.miamfl.sbcglobal.net> has quit IRC (Quit: Ping timeout (120 seconds))[11:57:18] *** mmlb <mmlb!~mmlb@76-248-148-178.lightspeed.miamfl.sbcglobal.net> has joined #smartos[12:13:13] *** jellydonut <jellydonut!~jelly@148.122.187.2> has joined #smartos[12:33:14] *** mmlb <mmlb!~mmlb@76-248-148-178.lightspeed.miamfl.sbcglobal.net> has quit IRC (Quit: Ping timeout (120 seconds))[12:36:05] *** mmlb <mmlb!~mmlb@76-248-148-178.lightspeed.miamfl.sbcglobal.net> has joined #smartos[12:37:15] *** mpana <mpana!~mpana@89.136.157.123> has quit IRC (Read error: Connection reset by peer)[12:38:08] *** mpana <mpana!~mpana@89.136.157.123> has joined #smartos[13:16:10] *** jellydonut <jellydonut!~jelly@148.122.187.2> has quit IRC (Read error: Connection reset by peer)[13:31:11] *** Kurlon <Kurlon!~Kurlon@98.13.72.207> has joined #smartos[13:35:26] *** Kurlon <Kurlon!~Kurlon@98.13.72.207> has quit IRC (Ping timeout: 246 seconds)[14:17:39] *** Kurlon <Kurlon!~Kurlon@98.13.72.207> has joined #smartos[14:53:10] *** ingenthr <ingenthr!~ingenthr@47.150.213.46> has joined #smartos[15:03:31] *** ingenthr <ingenthr!~ingenthr@47.150.213.46> has quit IRC (Quit: Leaving.)[15:07:18] *** Kurlon <Kurlon!~Kurlon@98.13.72.207> has quit IRC (Ping timeout: 268 seconds)[15:18:09] *** v_a_b <v_a_b!~volker@p57A27EA7.dip0.t-ipconnect.de> has joined #smartos[15:21:43] *** axonpoet <axonpoet!~axonpoet@fsf/member/axonpoet> has joined #smartos[15:43:11] *** polishdub <polishdub!~polishdub@207.86.38.254> has joined #smartos[15:58:14] *** gotwf <gotwf!~gotz@unaffiliated/gotwf> has quit IRC (Quit: Lator gator... ;-))[16:10:28] *** andy_js <andy_js!~andy@94.3.60.133> has quit IRC (Read error: No route to host)[16:11:26] *** andy_js <andy_js!~andy@94.3.60.133> has joined #smartos[16:16:35] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has joined #smartos[16:32:16] *** andy_js <andy_js!~andy@94.3.60.133> has quit IRC (Read error: Connection reset by peer)[16:32:52] *** andy_js <andy_js!~andy@94.3.60.133> has joined #smartos[16:40:51] *** fxhp <fxhp!~fox@138.207.192.33> has quit IRC (Quit: WeeChat 1.9.1)[17:11:05] *** Gathis <Gathis!~TheBlack@unaffiliated/gathis> has joined #smartos[17:24:01] *** bens1 <bens1!~bens1@161.122.2.81.in-addr.arpa> has quit IRC (Quit: bens1)[17:38:08] *** jellydonut <jellydonut!~jelly@148.122.187.2> has joined #smartos[18:51:18] <Smithx10> HmMmmmm[19:17:02] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has quit IRC (Ping timeout: 246 seconds)[19:52:56] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has joined #smartos[20:27:45] *** jessfraz_ <jessfraz_!~jessfraz@unaffiliated/jessfraz> has quit IRC (Remote host closed the connection)[20:29:11] *** jessfraz_ <jessfraz_!~jessfraz@unaffiliated/jessfraz> has joined #smartos[20:31:14] *** jessfraz_ <jessfraz_!~jessfraz@unaffiliated/jessfraz> has quit IRC (Remote host closed the connection)[20:34:37] *** jessfraz_ <jessfraz_!~jessfraz@unaffiliated/jessfraz> has joined #smartos[21:09:37] <m1cr0man> Is there a way to use filesystem mounts to add a single file to a zone?[21:10:30] <Smithx10> m1cr0man: I don't think so.... but I could be completely wrong[21:11:42] <m1cr0man> Looks like a no q_q[21:11:51] <Smithx10> I do know that manta uses hyperlofs[21:11:56] <Smithx10> Check out hyperlofs[21:12:00] <m1cr0man> ok[21:12:15] <Smithx10> Manta maps objects into memory using hyperlofs[21:13:20] <Smithx10> m1cr0man: https://github.com/joyent/illumos-joyent/tree/master/usr/src/uts/common/fs[21:16:16] <m1cr0man> Are all of these usable? I thought it was just the ones in /usr/lib/fs[21:16:33] <m1cr0man> oh..derp I think they are the same list XD[21:17:09] <Smithx10> m1cr0man: what are you trying to accomplish?[21:17:18] <Smithx10> like a RO of a file in the zone?[21:17:38] <Smithx10> I believe that is what hyperlofs does.[21:17:48] <m1cr0man> yeah trying to add a custom hosts file to a docker zone in /etc.[21:17:52] <Smithx10> oooo[21:17:57] <Smithx10> you can just use the docker options.[21:17:58] <Smithx10> lol[21:18:00] <m1cr0man> ?![21:18:13] <m1cr0man> vmadm? docker options?[21:18:19] *** rsully <rsully!~rsully@unaffiliated/rsully> has joined #smartos[21:18:36] <Smithx10> https://docs.docker.com/engine/reference/run/[21:18:55] <Smithx10> I believe they take it as an arg.... but if you are doing it in vanilla smartos[21:18:55] <m1cr0man> what? I can't use docker run on SmartOS can I?[21:19:01] <Smithx10> But[21:19:14] <Smithx10> smartos sets these as tags[21:19:33] <Smithx10> So i think they can be interpreted[21:19:58] <Smithx10> 1 sec.... i have triton running ill do a run and see what they set[21:20:02] <m1cr0man> Yeah I am already reading the cmd, entrypoint and some other tags in my playbook[21:20:39] <m1cr0man> You know I think should just give up and setup sdc instead of messing with vanilla smartos..[21:21:15] <Smithx10> hahahaha[21:21:20] <Smithx10> Depends on your goals[21:21:23] <Smithx10> what is your goal?[21:22:25] <arekinath> hyprlofs fwiw makes a magic directory where you can map a bunch of files in together dynamically and quickly (without a new mount point for each one)[21:22:38] <arekinath> if you just want a single file then lofs is good for that[21:22:45] <arekinath> (and you can put the definition in the vmadm config for a zone)[21:23:22] <m1cr0man> Honestly I just wanted to learn SmartOS/Solaris, but then I decided to put it on my blog's server. I'm looking to have an easy way to deploy zones/containers to it.[21:23:36] <Smithx10> m1cr0man: here ya go[21:23:49] <Smithx10> https://gist.github.com/Smithx10/8c6c7198978f783e78158c419a4f41de[21:23:57] <Smithx10> "docker:extraHosts": "[\"bob:10.1.10.45\"]",[21:24:15] <arekinath> dockerinit writes /etc/hosts with that stuff, yeah[21:24:53] <Smithx10> arekinath: by chance do you know fi there is a way to apply tags at docker run that opt out of the" docker:" prefix?[21:25:11] *** noahmehl <noahmehl!~noahmehl@dsl-74-83-119-141.fuse.net> has joined #smartos[21:25:18] <m1cr0man> oh, well that partially solves the problem then thanks.[21:25:44] <Smithx10> right now in the k8s virtual-kubelet i have to do a docker run, and then do a cloudapi apply, id like to be able to do everything in 1 api call[21:26:06] <Smithx10> I was storing stuff in metadata, but I dont think you can set metadata via the dockerapi[21:26:20] <arekinath> Smithx10 (IRC): I don't believe so, no[21:26:31] <Smithx10> Do you know if you can apply metadata via the docker api?[21:26:44] <arekinath> I'm also not 100% sure[21:26:44] <Smithx10> labels are just put into tags[21:26:51] <Smithx10> from what i've seen[21:26:56] <arekinath> but I don't think either of those are possible today (it prefixes all the tags it sets)[21:27:03] <Smithx10> yea[21:27:12] <Smithx10> Ill just run the second apply[21:27:15] *** polishdub <polishdub!~polishdub@207.86.38.254> has quit IRC (Quit: leaving)[21:27:21] <arekinath> (this was why CNS had to be tauggt to look for the docker: prefixed versions of tags it interprets as well as the normal version)[21:27:32] <Smithx10> Im trying to make the vkube provider have the same behaviour for cloudapi as dockerapi[21:27:38] <Smithx10> bingo[21:27:41] <Smithx10> thats what im trying to avoid[21:27:56] <Smithx10> im just doing an extra call upfront so i dont have to change all the functions[21:28:50] <Smithx10> it would be nice if dockerapi had like a label that is prefixed like "cloudapi.metadata" or something[21:30:07] <Smithx10> hopefully, if I get this PoC rolling and we starting paying joyent.... we can have mgerdts fix affinity and maybe fix up these smaller issues :P[21:31:33] <mgerdts> Sorry about the delay on the affinity stuff. Holidays and mad dash for some end of year deadlines have kept us from paying enough attention to that. I'll have someone dig into it soon.[21:32:40] <mgerdts> FWIW, I've been looking at Kata containers a bit. It is an example of what I've had in mind for a better docker experience by transparently using bhyve.[21:33:05] <mgerdts> Of course, that makes transparently putting a k8s pod in a bhyve instance pretty straight forward too.[22:05:07] *** rzezeski <rzezeski!uid151901@gateway/web/irccloud.com/x-udsksbivkthwruyg> has joined #smartos[22:06:10] <m1cr0man> `/dev/zfsds0 on /etc/hosts type lofs (rw,nodevices)` lol so /etc/hosts is already being configured by...dockerinit?[22:06:27] <m1cr0man> I was hoping to stick to using the baremetal's host file because I'm using ansible to write it[22:06:45] <m1cr0man> UGH damn I'm just gonna have to setup a dns server[22:09:58] *** jellydonut <jellydonut!~jelly@148.122.187.2> has quit IRC (Remote host closed the connection)[22:10:57] <m1cr0man> https://github.com/joyent/smartos-live/blob/1cfc365023de88bd81388a7fc3790cf612ba53f9/src/vm/node_modules/VM.js#L2052 i see...[22:31:03] *** jellydonut <jellydonut!~jelly@148.122.187.2> has joined #smartos[22:50:18] <Smithx10> mgerdts: wouldnt the firecracker stuff be easier to use?[22:51:30] <Smithx10> i think the k8s crowd is going to jump all over that, i saw them advertising the idea here https://thenewstack.io/this-week-in-programming-kubernetes-future-in-virtual-machines/#disqus_thread[22:52:59] <Smithx10> m1cr0man: yea, you are kind of over complicating this[22:53:13] <Smithx10> m1cr0man: do you come from a Linux Container Background?[22:53:19] <m1cr0man> yes XD[22:53:54] <m1cr0man> But I'm not trying to run a 1000 node DC either. I just have a little hetzner server I want to run games on. I think SmartOS wasn't the smartest choice..[22:54:17] <m1cr0man> and by "games" I mean managed containers, not the standard hand bombing[22:54:58] <Smithx10> yea, so LX comes with a lot of bad habits like mounting things into the contianer[22:55:11] <Smithx10> most of that doesn't need to be done and probably shouldnt be done[22:55:34] <m1cr0man> Oh I had this whole stack deployed without the docker stuff but since influx and my blog stopped compiling for solaris I had to try LX[22:55:48] <Smithx10> ahhh ok[22:55:51] <m1cr0man> I had no single file mounts, it was configuring a joyent-minimal image from the playbook[22:55:54] <Smithx10> nevermind then ./ goes back to hole[22:56:07] <m1cr0man> XD[22:56:29] <m1cr0man> It was a fun experiment, I just don't think I can use smartos though[22:57:33] <Smithx10> cant use smartos for what?[23:03:51] *** Kurlon_ <Kurlon_!~Kurlon@bidd-pub-04.gwi.net> has joined #smartos[23:05:31] <m1cr0man> For running mostly linux things on not VMs[23:05:56] <Smithx10> What is the Application that doesn't work?[23:07:07] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has quit IRC (Ping timeout: 246 seconds)[23:08:04] *** Kurlon_ <Kurlon_!~Kurlon@bidd-pub-04.gwi.net> has quit IRC (Ping timeout: 250 seconds)[23:11:07] *** fejfighter <fejfighter!~fejfighte@124-171-74-70.dyn.iinet.net.au> has joined #smartos[23:12:49] <m1cr0man> It's not even that, with enough effort I can get everything to work - and I have. But just writing this playbook has been so difficult and I've had so many issues idk if I can justify keeping it up. Like I had my blog (nodejs app)deployed in a Joyent zone then I added a feature which required an img processing library which didn't compile on solaris. I can go through and modify things to use an LX zone, but then I'm losing more and more[23:12:49] <m1cr0man> of any benefits of using smartos in the first place.[23:14:11] <m1cr0man> If I was using a regular Linux server I could just use docker + some orchestration tool to manage containers and run all the things I need like that. Plus, most tools already support such a stack. For example, I had to write my own metrics collector for Telegraf to get per zone system stats.[23:14:38] *** noahmehl <noahmehl!~noahmehl@dsl-74-83-119-141.fuse.net> has quit IRC (Read error: Connection reset by peer)[23:14:43] <Smithx10> What are the benefits of SmartOS that you are after?[23:15:32] <Smithx10> I usually like having CrossBow, ZFS, DTrace, LX Zones, FMA, etc[23:15:44] <Smithx10> Bhyve is there now also, so thats sweeeeeeet[23:15:58] <Smithx10> Is there a reason you are using docker instead of just a regular lx branded zone?[23:16:07] <m1cr0man> The things I've liked are the consistent command line interface, ZFS, and the ease of configuring ipfilter. I am looking forward to learning + using DTrace.[23:17:01] <m1cr0man> I'm using Docker zones for influx and chronograf because (I thought) it is less work. When I ran a joyent brand zone for those I had a third zone just to compile the binaries. That worked well. Idk why I didn't start with using lx zones honestly.[23:17:22] <m1cr0man> I thought there was some performance loss I guess, or features that wouldn't work (lofs, but I found out later it does work)[23:17:38] <Smithx10> ahhhh ok :)[23:34:00] <_Tenchi_> LX is fast :D[23:35:29] <mgerdts> @Smithx10 Kata can use kvm or firecracker or qemu. Both use firecracker and qemu use kvm. Kata provides a lot of abstraction that makes it feel like a VM is a container or sandbox. To use Kata, we would need to go down one of two paths: 1) make virtcontainers aware of bhyve or 2) make firecracker use the bhyve kernel interface rather than the kvm interface.[23:36:28] <mgerdts> Firecracker has an explicit goal of supporting one implementation of each thing it supports. That is, virtio-blk or virtio-scsi; not both. I suspect they will have the same feeling about bhyve vs. kvm.[23:37:07] <mgerdts> In contrast, virtcontainers is part of Kata. Kata is specifically welcoming of supporting multiple VM backends.[23:38:54] <Smithx10> understood, you probably could just speed up the "Time to Shell" in kata-bhyve on SmartOS[23:41:04] *** Kurlon <Kurlon!~Kurlon@98.13.72.207> has joined #smartos[23:41:17] <Smithx10> mgerdts: Was there ever thoughts on using K8S CRD (Custom Resource Definitions) and a K8S Controller for them specifically for Triton? It would be a better fit then the Virtual Kubelet which is tied to the Pod Definition atm.[23:41:39] <Smithx10> Or does K8S talk to Kata to talk to another backend?[23:41:43] <mgerdts> There are certainly optimizations that could be made. While DAX is really cool, getting it to work is probably not as high of priority as other things. It may save a few milliseconds and a few MB of RAM.[23:42:17] <Smithx10> Ouch, DAX?[23:42:51] <mgerdts> https://www.kernel.org/doc/Documentation/filesystems/dax.txt[23:43:05] <Smithx10> ahhh thank you.[23:43:19] <mgerdts> https://github.com/kata-containers/documentation/blob/master/architecture.md[23:43:39] <mgerdts> that needs to be more prominent in the "kata containers" search results.[23:44:32] <Smithx10> https://github.com/kata-containers/documentation/blob/master/architecture.md#dax[23:44:34] <danderson> as a general rule, the k8s ecosystem are terrible at explaining how things work on the inside[23:44:43] <Smithx10> danderson: Ive discovered[23:44:46] <danderson> there's a lot of focus on the interfaces for cluster users[23:45:00] <danderson> (and I say this as one of GKE's SREs, we're part of the problem)[23:45:19] <mgerdts> yeah, look at this shiny black box. Isn't it pretty?[23:45:20] <danderson> we keep trying to sell k8s as this magic that's going to revolutionize business, instead of a software system with tradeoffs and costs[23:45:49] <Smithx10> danderson: I fight my managers all day because of that[23:46:06] <Smithx10> "You still need to know your application"[23:46:10] <Smithx10> "Thats not what heptio said"[23:46:15] <danderson> heh.[23:46:29] <danderson> our fight is trying to get k8s devs to instrument anything in core k8s[23:46:32] <Smithx10> There is a "Operator" for that :)[23:46:39] <Smithx10> lololol[23:46:39] <danderson> it's eventually consistent, it'll fix itself eventually![23:46:55] <danderson> yeah, but what if it doesn't, because the system is hard to test and so bugs creep in all the time?[23:47:11] <danderson> it's fine, customers will find that their cluster's broken and tell us! Ugh.[23:47:18] <mgerdts> I've gotta run to be festive or something like that. Happy New Year everyone![23:47:33] <Smithx10> Happy New Year mgerdts :) Have fun, and remember.[23:47:36] <danderson> happy new year! And sorry to pollute the channel with my k8s rant :)[23:47:37] <Smithx10> DONT DRINK AND DRIVE!!!!!![23:47:49] <Smithx10> lol, on new years..... dont even get close to roads :P[23:47:51] <danderson> I've been free of it for about 2 weeks, but work beckons and the wounds are going to reopen[23:48:28] <Smithx10> danderson: Did you find the stuff about ZFS?[23:48:46] <mgerdts> Oh, fine. Plans aren't really that exciting. Having dinner with the elderly inlaws before their 8 pm bedtime. Then back home awaiting the chance to try out my new snowblower.[23:49:43] <danderson> Smithx10: stuff about ZFS?[23:53:26] <Smithx10> @danderson https://twitter.com/bruce_smith_it/status/1079746970137309186[23:53:38] <Smithx10> This is you riiighhhtt??[23:54:14] <danderson> oh, right, that[23:54:51] <danderson> yeah, found some things. A recent-ish FreeBSD codewalk is the closest to a modern full overview I've found so far[23:55:14] <danderson> it's still missing a few things like compressed ARC and prefetching, but it's about 100% more detail about ZFS than I had before :)