=== liam_ is now known as Guest89217
=== Spads_ is now known as Spads
=== CyberJacob is now known as CyberJacob|Away
=== CyberJacob|Away is now known as CyberJacob
=== CyberJacob is now known as CyberJacob|Away
noodles785I was wanting to experiment adding egress rules to all units in an environment. I can't do that via constraints or similar juju functionality right? (I'm assuming I'll need to update the default secgroup for the env, which will only work when deploying to openstack/aws, not local)03:47
=== noodles785 is now known as noodles775
=== Spads_ is now known as Spads
rick_h_noodles775: you could add a subordinate, I've thought of doing that04:11
rick_h_noodles775: having a egress filter subordinate that get attched in dev/qa to create firewall rules manually04:11
=== kadams54 is now known as kadams54-away
noodles775rick_h_: yeah - I was hoping to do it outside the units themselves (ie. secgroups), but wanted to check whether it's something juju manages (I'm assuming not). Thanks!06:46
rick_h_noodles775: I don't think so atm06:47
=== CyberJacob|Away is now known as CyberJacob
=== CyberJacob is now known as CyberJacob|Away
=== natefinch is now known as nate-tablet
=== roadmr is now known as roadmr_afk
=== enterprisedc_ is now known as enterprisedc
=== kadams54 is now known as kadams54-away
=== roadmr_afk is now known as roadmr
=== liam_ is now known as Guest73106
skayI think the charm helper docs for charm create are out of sync with the package, but could use a sanity check. here's what results from following the docs, http://paste.ubuntu.com/9125147/15:30
skayexpected behavior should be http://pythonhosted.org/charmhelpers/getting-started.html#creating-a-new-charm right?15:30
tvansteenburghskay: too many "create" in your commands15:35
lazyPowerskay: are you looking at using the services framework as well15:36
lazyPowerthe python-basic template is more likely what you're looking for.15:36
skaylazyPower: when I tried to use the -t to specify a template, I got a usage error15:36
tvansteenburghskay you have "charm create create"15:37
skaytvansteenburgh: woohoo, I like it when it's a typo instead of a complete crazy misunderstanding15:46
tvansteenburgh :)15:46
skaylazyPower: python-basic is indeed what I was looking for (it also looks like the docs should indicate python-basic for hte template rather than python?)15:47
skaythanks all15:47
lazyPowerskay: It's a bit confusing I agree - python being an alias for the services framework.15:48
skaylazyPower: for feedback: when I use the python template the dir tree does not resemble the dir tree on the docs.15:49
* tvansteenburgh makes a note to update docs15:50
tvansteenburghthanks for pointing that out skay15:50
skaytvansteenburgh: I wasn't sure if I should open an issue in case I had misunderstood something. thanks for helping.15:50
tvansteenburghsure thing15:51
skayavoine: adding support for pip_extra_args in order to use --no-index --find-links=/some/path, isn't as straight forward as I thought due to pip not being able to find /some/path.15:54
skayavoine: I am not sure exactly what to do, and maybe I need to write a subordinate charm to populate /some/path on the python-django unit?15:55
skaybecause I don't think that functionality belongs in python-django15:55
skayI think adding support for pip_extra_args belongs, and python-django would just pass that to pip15:56
skayand it would be the responsibility of something else to ensure the path is reachable15:56
avoineskay: Have you tried to upgrade pip?16:21
skayavoine: I did try, that did not resolve the problem. (and also would be a problem if one applies pip_extra_args to all calls, if it is the case that the existing pip doesn't support the extra args, but while I investigated I introduced a temp kluge to ignore it when installing pip)16:23
avoineskay: just to be sure I understand, your installing a tar archive with your django project + your dependencies and then your pointing pip ... --find-links to your dependencies dir, right?16:31
avoinealso pip don't need to be upgrade on trusty since "Added support for bare paths (including relative paths) as argument to –find-links" was added in version 1.3 and trusty is 1.5.416:32
skayavoine: not exactly. I tarred up my django project and had dependencies in the local file system. oh derp, I should just tar them up too?16:32
skayavoine: I'm using precise at the moment (I thought the gunicorn charm was precise, but if I'm mistaken I can use trusty)16:32
avoinegunicorn is in trusty too16:34
avoineskay: how do you add your dependencies, with scp?16:34
avoine–find-links needs a dir where dependencies are already downloaded16:35
skayavoine: no, I ignorantly thought that juju would be able to provide access to them in the local file system16:35
avoineoh like vagrant does16:35
avoineno indeed it's not16:35
avoineI would put them in the tar16:35
skayyes, it is clear to me now16:36
skaythis is simpler.16:36
skaythanks for walking me through that16:36
avoinebut I think that's a feature they want to add16:36
avoinethat would be usefull16:36
avoinelazyPower: mounting local directory inside a local deployement (with nfs for example) is still a feature you wanted to add to juju?16:38
lazyPoweravoine: we found a work around as documented in the NFS charm16:39
avoineskay:  ^16:39
lazyPowerit requires some manual intervention - so adding something without the manual intervention would be nice - but i can see it as being lower priority16:39
lazyPowerskay: https://jujucharms.com/nfs/trusty/0 - take a look at "On the LXC host" in the readme16:40
* skay takes a look16:41
avoineI guess that would be a good use case for a juju plugin16:42
jw4lazyPower, avoine this sounds like what Juju Actions is intended for16:52
lazyPowerjw4: i dont think actions will be able to modify the HOST on behalf of the user.16:53
lazyPowerthose particular edits take place on the LXC host so the containers can receive the nfs kernel modules - otherwise apparmor blocks them16:53
jw4lazyPower: I see - and actions run in the context of the units16:54
lazyPoweractions are intended to be helpers that dont necessarily belong in the hooks - like database backups, adding users to an app, etc.16:55
jw4lazyPower: oh well.  I got excited for a minute there16:55
=== kadams54_ is now known as kadams54-away
=== utlemming_kitche is now known as utlemming_mobile
=== kadams54-away is now known as kadams54_
avoineI forgot, is there a way I can get the data of a relation outside a hook or with a juju command?17:19
avoinejuju-run on the vm did the trick17:31
=== kadams54_ is now known as kadams54-away
=== kadams54-away is now known as kadams54_
mwenninghi juju team, can anyone help with a juju hang problem?18:41
lazyPowermwenning: hey o/ whats going on with juju?18:47
mwenninglazyPower, hi, I'm running juju bootstrap --debug -e maas and it hangs for 10 min and then fails with a timeout18:49
lazyPowermwenning: are the units still spinning up? iirc you're working in a pure hardware environment18:50
mwenningI'm pretty sure I didn't set the proxy right, but it won't take the right value anymore.18:50
mwenningjuju destroy-environment --debug --force maas also hangs18:51
mwenningstby I'll get a pastebin18:51
josemwenning: is it still giving the 502?18:51
mwenningafter I set the proxy correctly it worked for awhile, now it's hung again18:52
mwenninglazyPower, jose, it hangs at line 9 for probably 10 min, then dumps the rest of that stuff.18:55
josemwenning: lazyPower will give you a hand with this - cannot access canonical services and I wouldn't risk pasting any private details on a public site18:56
mwenningjose, ok thx18:56
lazyPowermwenning: can your workstation reach the node its attempting to bootstrap when you just activate it in the MAAS gui?18:58
mwenninglazyPower, yes, maas works fine, I can start/stop all my nodes18:58
jcastrois `instance-type` ignored in environments.yaml?18:59
lazyPowermwenning: and workstation connectivity to the nodes works as well?18:59
jcastroI put m3.mediums but it's asking for m1.smalls18:59
lazyPowerjcastro: thats a depreciated constraint18:59
lazyPoweryou should add proper constraints18:59
mwenninglazyPower, yes, I can ssh into all the nodes from my maas server18:59
jcastroI thought we were putting it back in?19:00
jcastroor at least defaulting to m3's?19:00
lazyPowermwenning: hmm... this is troubling19:00
lazyPowermwenning: which juju revision are you on? 1.20.x i assume?19:00
mwenningjuju-core 1.20.1119:01
lazyPowerok, latest stable19:01
lazyPowerhmm this sounds like a reachability issue though19:01
mwenningwhat's wierd is that it doesn't seem to take anything from environments.yaml or environments/maas.jenv19:02
josewhat do you mean?19:03
mwenningI tried changing bootstrap-timeout to 120  and it didn't take it.19:03
mwenningI've got http-proxy and https-proxy set in the environments.yaml, that should point to the correct proxy but it doesn't seem to pay attention19:04
mwenningIs there somewhere else that juju is reading that info?19:05
lazyPowermwenning: im' asking for a core dev to come take a look with us - 1 moment19:06
lazyPoweri'm not using any proxies and have limited interaction with using a proxy in juju19:06
mwenninghmm, what the heck?  It's trying to access .19:10
mwenningThat's not my MAAS server, it's at
lazyPowermwenning: stale maas jenv?19:10
mwenningwhere is that?19:10
mwenningwhere is maas.jenv?19:11
josetvansteenburgh: ping, mind a quick pm?19:11
tvansteenburghjose: sure19:11
mwenningactually maas.jenv is gone.19:12
lazyPowermwenning: ok so if its not reading a stale jenv file - something else is awry19:15
lazyPowercan you verify the server config in environments.yaml is the correct ip address?19:15
mwenningI lazyPower , yes.  stby I'll paste the whole thing19:17
lazyPowermwenning: do you have another PROXy export set?19:22
lazyPowerecho $JUJU_PROXY i think19:22
mwenninglazyPower, $JUJU_PROXY shows nothing19:23
* lazyPower snaps19:23
lazyPowerthat was the last thought i had on what it could be19:23
lazyPoweri unfortunately dont know why you're seeing that behavior :(19:23
mwenninglazyPower, ok.  I've got to drive downtown to a meeting, I'll look at this later19:24
jcastrohey bcsallerm can you add your +1 to this bug?19:25
mupBug #1373516: Switch default instance type from m1.small to t2.small/m3.medium for EC2 provider <ec2-provider> <juju-core:Triaged> <https://launchpad.net/bugs/1373516>19:25
jcastrohey natefinch, can we get this on someone's radar? ^^^^19:27
jcastroThis will only get worse and worse19:27
=== CyberJacob|Away is now known as CyberJacob
mwenninglazyPower, if you can think of anywhere that juju might be cacheing that information please let me know, that's all I can figure19:33
lazyPowermwenning: my only other suggestion is to do a full env dump and sift through whats in there19:34
lazyPowerjuju reads from env => .jenv => environments.yaml - in that order.19:34
mwenninglazyPower, do you know if it reads http_proxy in the env?19:35
lazyPoweri believe so19:35
mwenningok, I think that's new, but I've set it there too.19:36
natefinchjcastro: I think we addressed that... lemme double check, thought I saw that in the 1.21 release notes19:36
natefinchjcastro: hmm. maybe not19:37
=== kadams54 is now known as kadams54-away
natefinchjcastro: it's definitely on our radar.  I'll see if I can get someone to put a little time into fixing it.  I think the main problem we had was that there wasn't really a no-brainer replacement.  both t2.small and m3.medium had slight problems (I believe t2.small is a burstable instance, and m3.medium is slightly more expensive)19:39
jcastroit was just reiterated to us that m1's are going away much faster than we thought19:40
natefinchjcastro: personally, I'd just go with m3.medium.... it's not our fault amazon doesn't have cheap prices :)19:40
natefinch(for reasonable machines)19:40
natefinchjcastro: I'll bring it up right now.  I agree it needs to get fixed ASAP.  Of course, the problem is that it won't be fixed in older jujus19:41
jcastroI'd rather have people pay more than having it not work because m1's are being culled19:41
natefinchjcastro: we also should handle that error better by requesting from a different AZ in the region and/or trying a different instance type19:41
natefinchjcastro: who said the m1's are going away faster?  Just want to give proper attribution.19:43
jcastroben and robert19:43
jcastroaka the CPC team19:43
jcastrorobert filed the initial bug19:43
natefinchjcastro: just sent a mail to juju-dev.  Assuming no one objects, I'll try to make sure someone gets on it ASAP.19:50
jcastro-<3 thanks19:50
=== kadams54-away is now known as kadams54
=== roadmr is now known as roadmr_afk
=== kadams54 is now known as kadams54-away
=== roadmr_afk is now known as roadmr
=== arosales__ is now known as arosales
=== CyberJacob is now known as CyberJacob|Away

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!