=== zz_CyberJacob is now known as CyberJacob | ||
=== CyberJacob is now known as Guest43720 | ||
=== Guest43720 is now known as zz_Guest43720 | ||
=== ming is now known as Guest66148 | ||
=== plars_ is now known as plars | ||
=== \b is now known as benonsoftware | ||
=== zz_Guest43720 is now known as CyberJacob | ||
=== CyberJacob is now known as Guest68261 | ||
=== Guest68261 is now known as cyberjacob | ||
gennadiy | hi all, i'm going to publish my bundle to juju charm store to user namespace. i pushed it to launchad repo. waited 1 hr but it haven't appeared in charm store yet | 10:25 |
---|---|---|
gennadiy | my repo - https://code.launchpad.net/~tads2015dataart/charms/bundles/tads2015-demo/bundle | 10:25 |
gennadiy | do i need to wait for any review? | 10:26 |
=== JoshStrobl is now known as JoshStrobl|AFK | ||
gennadiy | hi all, i'm going to publish my bundle to juju charm store to user namespace. i pushed it to launchad repo. waited 1 hr but it haven't appeared in charm store yet. should i wait for any review? | 11:20 |
rick_h__ | gennadiy: give it one more hour please | 11:41 |
rick_h__ | gennadiy: oh looks like you did | 11:41 |
rick_h__ | gennadiy: lokoing | 11:41 |
rick_h__ | errr looking | 11:41 |
gennadiy | i made some changes in bundle. i have added quotes for gui position attributes | 11:41 |
rick_h__ | gennadiy: ah, you're caught in our corner case atm. sec I'll get you a quick instruction to work around it | 11:42 |
rick_h__ | gennadiy: ok, so you've got a file named 'bundle.yaml' but your bundle is in the older format. so please rename that file 'bundles.yaml' with an (s) and you should be good | 11:44 |
gennadiy | @rick_h___ also i have question about juju-gui service in bundle. my charm depends on this service and i need to add relation between them. but juju bundle proof return error if bndle contains juju-gui service | 11:44 |
rick_h__ | gennadiy: yes, bundles are deployed through the gui, so it has to exist first to send the bundle to be deployed. | 11:45 |
rick_h__ | gennadiy: so if there's a conflict it'll error there | 11:45 |
rick_h__ | gennadiy: you can try renaming it to smoething else. juju-gui-2 or the like, but not sure if that works. | 11:46 |
rick_h__ | gennadiy: there's WIP to remove that restriction and deploy bundles straight from the juju cli, but it's not ready yet. | 11:46 |
gennadiy | not sure that it's correct way to rename service. in this case we will have 2 different juju-gui :) | 11:47 |
rick_h__ | gennadiy: yes, I'm trying to think of a way around it. Because the user finds the bundle in the store and hits deploy, they've alreaday got a GUI | 11:47 |
gennadiy | it will not be a problem | 11:48 |
rick_h__ | gennadiy: unfortunately I don't see any way around it at the moment. | 11:48 |
rick_h__ | no one's related to the GUI before so it's just not come up heh | 11:48 |
rick_h__ | gennadiy: what does it do with the gui that it's related to it? /me is curious | 11:48 |
gennadiy | another question: if i push bundle with with `juju bundle proof` error - will be it published to store? | 11:49 |
gennadiy | we are trying to implenet autoscale. and we use juju api to increase count of units | 11:50 |
gennadiy | so i need to know ip address of juju api | 11:50 |
rick_h__ | gennadiy: no, it will fail to ingest | 11:50 |
gennadiy | one more question: do we have some page with review errors? i see http://review.juju.solutions/ in the head of page. but i can't find my charms/bundles there | 11:53 |
rick_h__ | gennadiy: for the juju api endpoint can you send an email to the juju mailing list about wanting to do that? | 11:53 |
rick_h__ | gennadiy: long ago another wanted to do this and there were lots of ideas as to how it might be done in the charm's hook context. | 11:53 |
rick_h__ | gennadiy: but I'm having trouble finding it in gmail atm. | 11:53 |
rick_h__ | gennadiy: so that's the review queue to be reviewed and a recommended charm. You have to follow manual steps to go through that and meet stricter requirements. | 11:54 |
rick_h__ | gennadiy: see https://jujucharms.com/docs/stable/authors-charm-store#recommended-charms | 11:54 |
gennadiy | clear | 11:55 |
gennadiy | i will be good to have some page with automatic review result too. | 11:56 |
gennadiy | because today i have some issues with charm updates | 11:57 |
rick_h__ | gennadiy: yes, we're moving to a new model for the charms/bundles that removes the pulling from launchpad and allows users to directly upload to the store | 11:57 |
rick_h__ | gennadiy: so you'd get immediate feedback on issues and not have to wait for automated machinery to come around and pick things up | 11:57 |
gennadiy | cool | 11:57 |
rick_h__ | gennadiy: we've currently got an 'older' deprecated system that's going to EOL in Dec that's causing most of the current delays | 11:58 |
rick_h__ | as the two systems (old and new) have to try to keep in sync so they wait for each other and such | 11:58 |
gennadiy | do you have posibility to check the state of latest commit of this charm - https://code.launchpad.net/~tads2015dataart/charms/trusty/telscale-restcomm/trunk? | 11:58 |
rick_h__ | so come Dec we'll kill the old system and hopefully have new tools for direct uploads | 11:58 |
rick_h__ | gennadiy: sure thing, looking | 11:58 |
gennadiy | store shows 2 revisions only - https://jujucharms.com/u/tads2015dataart/telscale-restcomm/trusty | 11:59 |
gennadiy | also do i need to change revision number manually before publishing? | 12:00 |
rick_h__ | gennadiy: it takes about 2hrs for both systems to get into sync. Looks like that was updated 1hr ago. Give it a bit more and watch the file here to see your new config and you'll know it's pulled in. https://api.jujucharms.com/charmstore/v4/~tads2015dataart/trusty/telscale-restcomm-1/archive/config.yaml | 12:01 |
rick_h__ | gennadiy: the charm should come in fine though, no issues I can see. | 12:01 |
gennadiy | clear, i will wait 2 hrs in the future | 12:02 |
gennadiy | thanks alot for your help | 12:02 |
rick_h__ | gennadiy: very sorry, it's painful | 12:02 |
rick_h__ | gennadiy: the folks are working on the new stuff as fast as they can so when we can kill the old system it'll be a ton better. | 12:03 |
rick_h__ | lazypower: <3 the video and blog post. you're my hero man. That's awesome stuff. great work katco wwitzel3 natefinch ericsnow http://blog.dasroot.net/2015-charming-2-point-oh.html | 13:24 |
lazypower | :) Happy to help | 13:24 |
lazypower | all the credit goes to the people who did the work, i just blog about it | 13:25 |
rick_h__ | lazypower: pulls together nicely | 13:25 |
=== gmeszaros_ is now known as gmeszaros | ||
Icey | is there a way to get juju to deploy SSD EBS backed AMIs instead of traditional EBS backed? | 13:39 |
Icey | on AWS | 13:39 |
Icey | in essence, can I specify the AMI ID to deploy? | 13:41 |
Icey | oh god, apparently the AMI we're deploying should be an SSD EBS backed instance but we're deploying it with a standard type EBS volume | 13:42 |
urulama | hey, any pointers on how to set private simplestream, i'm getting: | 13:50 |
urulama | read metadata index at "http://cloud-images.ubuntu.com/releases/streams/v1/index.sjson" | 13:50 |
urulama | 2015-10-23 12:45:26 DEBUG juju.environs.simplestreams simplestreams.go:433 skipping index "http://cloud-images.ubuntu.com/releases/streams/v1/index.sjson" because of missing information: index file has no data for cloud {RegionOne http://10.5.0.195:5000/v2.0} not found | 13:50 |
rick_h__ | Icey: I think the 1.25 release with storage support lets you do that. see https://jujucharms.com/docs/master/storage and search for ebs-ssd | 14:01 |
rick_h__ | Icey: the 1.25 pending release I should say | 14:01 |
Icey | thanks | 14:01 |
Icey | I'm so looking forward to the storage stuff but it isn't really what I'm talking about, I mean the root disk stuff, not added storage | 14:02 |
Icey | like: '--constraints "root-disk=250G cpu-cores=2 root-disk-type=ssd"' | 14:02 |
Icey | I now root-disk-type doesn't exist | 14:03 |
Icey | but illustrates what I'm wanting | 14:03 |
Icey | know* | 14:03 |
=== JoshStrobl|AFK is now known as JoshStrobl | ||
rick_h__ | Icey: is there an instance type that's ssd backed? maybe have to go that route. I'm not sure if there's another work around. Might be worth an email to the juju list to see if anyone's got a trick up their sleeve | 14:50 |
Icey | the AMI we're using _should_ be using an SSD backed EBS for the root drive | 14:50 |
Icey | somehow, we're initializing it with a standard EBS | 14:50 |
Icey | rick_h__, | 14:50 |
rick_h__ | Icey: oh, really not sure then :/ | 14:51 |
Icey | how hard is it to setup an aws environment that will let you colocate charms? For example I may want a cache host that runs redis and memcached on the same host instead of two separate hosts | 15:19 |
marcoceppi_ | Icey: juju deploy --to lxc:<machine-#> | 15:19 |
marcoceppi_ | well, | 15:20 |
Icey | and if I wanted to do something like that with a bundle :) ? | 15:20 |
marcoceppi_ | Icey: you can | 15:20 |
Icey | spread a bundle around on x machines with LXC? | 15:20 |
marcoceppi_ | yup | 15:20 |
* marcoceppi_ gets example | 15:20 | |
rick_h__ | Icey: marcoceppi_ https://jujucharms.com/docs/1.24/charms-bundles#service-constraints-in-a-bundle shows the lxc bit | 15:23 |
rick_h__ | sorry this one https://jujucharms.com/docs/1.24/charms-bundles#bundle-placement-directives | 15:23 |
marcoceppi_ | Icey: https://gist.github.com/marcoceppi/58a5598f038fda5fc8cd | 15:24 |
Icey | more broadly, say I wanted to deploy an existing bundle with colocated services: https://jujucharms.com/rails-example-scalable/6 | 15:24 |
marcoceppi_ | Icey: so, that's something you can do in the GUI, which allows you to modify placement before execution | 15:24 |
marcoceppi_ | Icey: so, bundle -> gui, modify placement, commit, then you can export the new bundle | 15:25 |
Icey | gotcha | 15:25 |
marcoceppi_ | rick_h__: correct me if I'm wrong, placment and pre-commit bundles exist in gui now, right? | 15:25 |
rick_h__ | marcoceppi_: yes | 15:28 |
marcoceppi_ | \o/ | 15:28 |
narindergupta | arosales: jamespage: Nuage marked their charms as fix committed. But i think they need to create a new charm for neutron-api. I have send an email copy to let Nuage and me know in case thats not the case. | 15:31 |
gennadiy_ | @rick_h__ are you still here? my bundle is still not visible in charm store - https://code.launchpad.net/~tads2015dataart/charms/bundles/tads2015-demo/bundle , my page in charmstore - https://jujucharms.com/u/tads2015dataart/ | 15:58 |
rick_h__ | urulama_: can someone help gennadiy_ please? ^ | 16:02 |
urulama | gennadiy_: hey | 16:03 |
gennadiy_ | hi | 16:03 |
gennadiy_ | i have publiched my bundle 3 hrs ago | 16:03 |
gennadiy_ | but it's still not visible in charm store. seems something wrong with it. | 16:03 |
urulama | gennadiy_: bundles are publish quicker, if they pass validation | 16:04 |
gennadiy_ | but i checked it with juju bundle proof. everything was ok | 16:04 |
urulama | gennadiy_: ok, let me check | 16:04 |
urulama | gennadiy_: yes, only {"Id":"cs:~tads2015dataart/trusty/telscale-restcomm-2","PublishTime":"2015-10-23T12:16:21.185Z"},{"Id":"cs:~tads2015dataart/trusty/monitor-server-1","PublishTime":"2015-10-23T10:46:30.363Z"} were published | 16:05 |
urulama | gennadiy_: i'll have to check the logs | 16:05 |
=== jam1 is now known as jam | ||
jrwren | gennadiy_: : bundle verification failed: ["service \"calls-consumer\" is subordinate but has non-zero num_units","service \"conference-call\" is subordinate but has non-zero num_units","service \"monitor-agent-mesos\" is subordinate but has non-zero num_units","service \"monitor-agent-mesos-master\" is subordinate but has non-zero num_units","service \"sms-feedback\" is subordinate but has non-zero num_units" | 16:13 |
jrwren | ] | 16:13 |
jrwren | gennadiy_: subordinates should have num_units: 0, or not that property at all. | 16:13 |
arosales | narindergupta: so should we wait for the neutron-api charm before we review the Nuage charm? | 16:58 |
narindergupta | arosales: James mentioned in PM that not needed. So you can go ahead to review. | 17:01 |
narindergupta | arosales: he mentioned no need to follow new design currently as they already integrated few sdns into neutron-api as well like calico | 17:02 |
arosales | narindergupta: thanks for the ping | 17:15 |
Icey | can somebody help me out with an interface? I'm writing a new simple interface but my charm doesn't seem to want to use it? | 17:19 |
jrwren | Icey: have it listed in metadata.yaml ? | 17:19 |
Icey | aye (I think?) | 17:19 |
Icey | https://gist.github.com/ChrisMacNaughton/12fe21abab4ddc3a880d | 17:20 |
jrwren | Icey: i assume you mean relation interface? | 17:20 |
Icey | I'm writing the reactive interface for influxdb | 17:20 |
Icey | or at least, I'm trying to | 17:20 |
jrwren | Icey: and some influxdb charm provides influxdb interface? | 17:20 |
Icey | https://jujucharms.com/u/chris.macnaughton/influxdb | 17:21 |
Icey | need to change my name in the charm | 17:21 |
Icey | but 'm trying to build an interface like:http://interfaces.juju.solutions/ | 17:21 |
jrwren | oh that kind of interface. i don't know about those. sorry | 17:22 |
Icey | yeah, trying to avoid any actual hooks in this charm, since it's literally just dropping some html under apache | 17:22 |
Icey | but it needs the config for that database | 17:22 |
marcoceppi_ | Icey: I can try to help you out | 17:24 |
Icey | want to do a hangout for a min? | 17:25 |
marcoceppi_ | Icey: sure | 17:25 |
Icey | https://plus.google.com/hangouts/_/canonical.com/can-i-make-a-name-for-this-hangout-that-is-longer-than-the-one-marcoceppi-made-earlier-to-talk-about-ceph-benchmarking-things-so-we-can-figure-out-interfaces | 17:26 |
Icey | hyphens do make that name easier to read :) | 17:26 |
Icey | cury_fu if I want to embed a tgz in a charm using the apache-php layer instead of having it install from a remote source? | 17:44 |
bdx | hey whats going on everyone? Can I get a review or two on the MR's associated with https://bugs.launchpad.net/charms/+source/nova-compute/+bug/1509267 | 18:30 |
mup | Bug #1509267: Ceph-relation-joined hook error <ceph> <nova-compute> <openstack> <storage> <nova-compute (Juju Charms Collection):New> <https://launchpad.net/bugs/1509267> | 18:30 |
bdx | thanks | 18:30 |
* GQWER slaps Zetas around a bit with a large fishbot | 18:42 | |
* GQWER slaps X-Rob around a bit with a large fishbot | 18:43 | |
marcoceppi_ | cory_fu bcsaller any ideas on this? http://paste.ubuntu.com/12905838/ | 19:19 |
marcoceppi_ | ugh, nvm. Apparently, 'layer: basic' is not the same as 'layer:basic' | 19:19 |
marcoceppi_ | is that a bug or a feature? | 19:19 |
bcsaller | same as cs:foo vs cs: foo | 19:20 |
marcoceppi_ | bcsaller: makes sense, thanks | 19:20 |
* marcoceppi_ can imagine adding lint rules to charm proof for layers | 19:20 | |
ejat | marcoceppi_: http://paste.ubuntu.com/12906168/ | 19:43 |
ejat | btw .. how r ya ? | 19:43 |
ejat | i've try using juju resolved | 19:44 |
ejat | retry .. but nothing ... :( | 19:44 |
marcoceppi_ | ejat: I'm not sure about that, but try cory_fu kwmonroe and admcleod- | 19:46 |
marcoceppi_ | they may be able to help | 19:46 |
ejat | ive tried : juju resolved yarn-master/0 --retry | 19:47 |
ejat | but its not retry ... | 19:47 |
kwmonroe | ejat: can you paste the output from 'juju debug-log -i yarn-master/0 -n 100' | 19:48 |
kwmonroe | *pastebin | 19:48 |
ejat | http://paste.ubuntu.com/12906269/ | 19:50 |
kwmonroe | hm.. a problem importing jujuresources.. gimme a sec | 19:56 |
ejat | kwmonroe: ok | 19:58 |
kwmonroe | ejat: let's go back farther.. pastebin the output from 'juju debug-log -i yarn-master/0 -n 1000 | grep jujuresources' to see if we can see jujuresources getting installed | 19:59 |
ejat | http://paste.ubuntu.com/12906441/ | 20:02 |
cory_fu | kwmonroe: Probably should have used --replay instead of -n 1000 | 20:03 |
cory_fu | ejat: I take it you ran `juju resolved` several times? | 20:04 |
ejat | i try click at juju-gui | 20:04 |
ejat | :( | 20:04 |
ejat | my bad | 20:04 |
ejat | cli 2 times | 20:04 |
cory_fu | ejat: No worries. But we do need the first error from the log. Can you look through `juju debug-log --replay -i yarn-master/0 | less` and get us the very first traceback error? | 20:05 |
cory_fu | marcoceppi_: Whatever happened to that "juju share" or "juju mate" project, or whatever it was called? | 20:06 |
cory_fu | Would be very useful heree. :) | 20:06 |
cory_fu | ejat: I don't suppose you've used the dhx plugin before, have you? | 20:07 |
ejat | http://paste.ubuntu.com/12906522/ | 20:07 |
ejat | cory_fu: havent use dhx before yet | 20:09 |
ejat | u want me to install the plugin ? | 20:09 |
cory_fu | ejat: That's definitely the right error. There ought to be a few lines above that that give info about what caused the apt-get install to fail. | 20:10 |
ejat | u want the line before the traceback | 20:11 |
cory_fu | Maybe 5 lines before, yeah | 20:11 |
ejat | http://paste.ubuntu.com/12906609/ | 20:13 |
cory_fu | I've never seen that "packages cannot be authenticated" before. | 20:15 |
kwmonroe | marcoceppi_: are we back in your wheelhouse yet? | 20:15 |
kwmonroe | ejat: can you try "juju ssh yarn-master/0 'sudo apt-get -y install python-pip'" | 20:18 |
kwmonroe | and see if you get the same unauthenticated pkgs warning | 20:18 |
ejat | E: There are problems and -y was used without --force-yes | 20:19 |
kwmonroe | ejat: are you using any http or apt proxies in your environment? | 20:20 |
ejat | nope | 20:21 |
ejat | my environment at Azure | 20:21 |
cory_fu | From this http://ubuntuforums.org/showthread.php?p=7001019#7001019 it seems like it might be an issue with the debian-keyring. | 20:21 |
cory_fu | No idea what might have caused it, though | 20:22 |
kwmonroe | ejat: paste output from this: juju ssh yarn-master/0 'apt-key list' | 20:23 |
kwmonroe | and we'll compare keys | 20:23 |
ejat | http://paste.ubuntu.com/12906786/ | 20:24 |
Icey | can I check the current state (eg: 'database.available') from within debug-hooks for a charm? | 20:24 |
cory_fu | Icey: Yes. You can use the charms.reactive CLI tool and call get_states | 20:24 |
cory_fu | Icey: $CHARM_DIR/bin/charms.reactive get_states | 20:25 |
cory_fu | I think that's the right syntax | 20:25 |
Icey | $CHARM_DIR is the location of my current charm, which has no charms;reactive (the charm itself is in python), do I need to grab charms.reactive to look at state in bash? | 20:26 |
cory_fu | Icey: If you're using the basic layer, it should include charms.reactive by default | 20:26 |
ejat | is the key differ? | 20:27 |
kwmonroe | ejat: those are the same keys i have.. i was hoping they'd be different (though i'm not sure how they would be) | 20:27 |
cory_fu | Icey: Oh, there was a bug in charm-tools that it didn't include the CLI tools | 20:27 |
ejat | :( | 20:27 |
kwmonroe | ejat: long shot, but try this: juju ssh yarn-master/0 'sudo apt-key update && sudo apt-get update && sudo apt-get install -y python-pip' | 20:27 |
Icey | -_- cory_fu | 20:27 |
kwmonroe | ejat: and see if you still get the "E: There are problems" message | 20:28 |
cory_fu | Icey: If you do a manual `pip install charm.reactive` then you'll get the CLI tool | 20:28 |
cory_fu | Sorry, charms.reactive | 20:28 |
kwmonroe | it's a long shot because i don't think "apt-key update" is gonna do anything, but worth a try i think | 20:28 |
ejat | its installing \0/ | 20:29 |
Icey | thanks cory_fu | 20:30 |
cory_fu | ejat, kwmonroe: My guess is some sort of transient network error causing the packages to be corrupted. And once the apt-get failed, it was just trying to use pip which wasn't installed | 20:30 |
ejat | should i rerun resolved? | 20:30 |
kwmonroe | hmph.. weird. yeah ejat: juju resolved -r yarn-master/0 | 20:30 |
cory_fu | ejat: Actually, you still need to get jujuresources installed | 20:31 |
cory_fu | ejat: I'd recommend doing this: juju run --unit yarn-master/0 'hooks/install' | 20:31 |
marcoceppi_ | kwmonroe cory_fu unauthenticated packages means somethings up with apt. Should just be able to run sudo apt-get update on the node to sort it out | 20:31 |
cory_fu | I think that will redo the important bit and get jujuresources installed | 20:31 |
cory_fu | marcoceppi_: If only you'd said that 20 minutes ago | 20:32 |
ejat | http://paste.ubuntu.com/12907066/ | 20:42 |
ejat | ? | 20:46 |
ejat | error installing the hook | 20:47 |
cory_fu | ejat: I think that's a difference in the PATH or something between `juju run` and a proper hook context. It looks like it got past the pre_setup bit that I was concerned about. You should be able now to use `juju resolved --retry` on the unit, I think | 20:50 |
ejat | http://paste.ubuntu.com/12907214/ | 20:52 |
ejat | \0/ | 20:53 |
ejat | https://jujucharms.com/apache-flume-twitter/trusty/3#revisions | 20:54 |
ejat | i cant test | 20:54 |
cory_fu | What do you mean, you can't test? | 20:56 |
ejat | To verify this charm is working as intended, SSH to the flume-hdfs unit, locate an event, and cat it: | 21:00 |
cory_fu | Are you saying that you cannot ssh into flume-hdfs? Or that no files are showing up in HDFS? | 21:05 |
ejat | can ssh | 21:05 |
ejat | no file/directory | 21:05 |
cory_fu | ejat: I assume you set your twitter access credentials on the flume-twitter service? | 21:06 |
ejat | secret.yaml | 21:07 |
ejat | yes | 21:07 |
kwmonroe | ejat: assuming your creds are set, you can "juju ssh flume-twitter/0" and have a look through /var/log/flume-twitter logs | 21:07 |
kwmonroe | dont' pastebin those because i'm not sure if they expose your twitter creds or not | 21:07 |
cory_fu | ejat: And if you do `juju get flume-twitter` you see the correct credentials? | 21:08 |
ejat | hdfs dfs -ls /user/flume/events # <-- find a date | 21:08 |
ejat | ? | 21:16 |
kwmonroe | ejat: how about "hdfs dfs -ls /user/flume" | 21:16 |
kwmonroe | do you see a 'flume-twitter' subdir in there? | 21:18 |
ejat | http://paste.ubuntu.com/12907539/ | 21:20 |
kwmonroe | ok ejat, "hdfs dfs -ls -R /user/flume/flume-twitter" | 21:21 |
kwmonroe | looks like the readme incorrectly points to an 'events' subdir for the test, vs 'flume-twitter' | 21:22 |
ejat | :) | 21:23 |
=== wolsen_ is now known as wolsen | ||
kwmonroe | ejat: i'm about to EOW, but wanted to make sure you have tweets in hdfs.. things looking ok? | 22:15 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!