/srv/irclogs.ubuntu.com/2014/02/23/#juju.txt

=== xnox_ is now known as xnox
=== msx____ is now known as msx
cjohnstonhazmat: doubting your still around, but I just got bug #1269519 again00:42
_mup_Bug #1269519: Error on allwatcher api <juju-core:Fix Released by rogpeppe> <juju-deployer:Fix Released> <https://launchpad.net/bugs/1269519>00:42
cjohnstonis there a way of knowing which machine log is needed00:43
hazmatcjohnston, traceback from deployer run .... ideally with -vWd00:43
cjohnstonhazmat: http://paste.ubuntu.com/6979251/00:44
cjohnstondoesn't have -vWd tho00:44
hazmatcjohnston, juju ssh 000:44
hazmatlog is in /var/log/juju/ machine-0.log afaicr00:44
cjohnstonsounds right00:44
cjohnstonhazmat: http://paste.ubuntu.com/6979254/ and all: http://paste.ubuntu.com/6979255/00:45
hazmatcjohnston, interesting.. thanks. i'm done for the night but thats helpful.. i'll talk to rog re the errors there but it sounds like given the synchronous py api..  short term env connections are the to go... will review and give feedback tmorrow00:47
hazmatseems like i/o timeout is  the eventual err from masking the previous ping timeout.. tbd..00:49
cjohnstonthanks00:51
=== freeflying is now known as freeflying_away
=== CyberJacob is now known as CyberJacob|Away
=== freeflying_away is now known as freeflying
=== freeflying_away is now known as freeflying
=== OutOfControl is now known as benonsoftware
hazmatcjohnston, hmm.. actually the log level on these pastebins is missing the api level.. if you have a chance to do run it again.. can you set JUJU_LOGGING_LEVEL="<root>=DEBUG"08:30
hazmator just on the enviornment via juju set logging-config="<root>=DEBUG"08:30
hazmatcjohnston, whats the instance size on your state server in these envs08:37
=== CyberJacob|Away is now known as CyberJacob
=== CyberJacob is now known as CyberJacob|Away
=== CyberJacob|Away is now known as CyberJacob
cjohnstonhazmat: for juju set it wants a service name15:55
hazmatcjohnston, juju set-env15:59
cjohnstonack15:59
rick_hccccccbtujivtdenulrrvgtjjlcnnjrledtruujfgnhr16:06
cjohnstonagreed16:18
rick_h:)16:20
cjohnstonhazmat: http://paste.ubuntu.com/6982593/ and http://paste.ubuntu.com/6982594/16:46
hazmatcjohnston, you reproduced the state watcher gone aaay.. i think the logs might be too big for pastebin16:48
hazmatcjohnston, er.. did you reproduce the state watcher gone away?16:48
cjohnstonI don't think so now that I look closer, but I do see:  WARNING discarding API open error: read tcp 127.0.0.1:37017: i/o timeout16:50
cjohnstonERROR connection is shut down16:50
=== timrc is now known as timrc-afk
hazmatcjohnston, what's the load on the state server machine like..18:15
cjohnstonnot sure18:15
cjohnstonthis 5 minute timeout sucks18:44
cjohnstonthe highest load I've seen so far has been 218:44
cjohnstonavg18:44
cjohnstonhazmat: I saw it hit almost 318:58
=== 7JTAAALD7 is now known as lifeless
=== zz_mwhudson is now known as mwhudson
=== Mantle is now known as Darkmantle
Darkmantleo/20:46
DarkmantleNeed some help please. I have Ubuntu desktop with Ubuntu Server 12.04 VM running MAAS - juju successfully bootstrapped but status hangs20:46
DarkmantleLogs show that it can't connect to mongodb, i've double checked the DNS configuration and its ok20:47
=== mwhudson is now known as zz_mwhudson
=== zz_mwhudson is now known as mwhudson
Darkmantle?20:54
Darkmantle2014-02-23 19:35:03 INFO juju.state open.go:68 opening state; mongo addresses: ["localhost:37017"]; entity "" 2014-02-23 19:35:03 DEBUG juju.state open.go:88 connection failed, will retry: dial tcp 127.0.0.1:37017: connection refused21:08
* Darkmantle redeploys21:10
=== ev_ is now known as ev
=== thumper is now known as thumper-afk
DarkmantleOk I can't do juju status, it just hangs, any idea why?21:39
DarkmantleI've configured the DNS correctly - I get mongodb connection issues on juju node startup but they are all working ok, or so it seems21:40
DarkmantleOr it could be a DNS issue, if I try to do mongo name.master:37017 I get no address associated21:41
Darkmantleor via IP either21:42
Darkmantle?21:51
hazmatDarkmantle, what version of juju?21:51
Darkmantle1.16.621:52
Darkmantleprecise-amd6421:52
hazmatDarkmantle, can you pastebin  $ juju status --debug21:52
Darkmantlesure21:52
hazmatDarkmantle, in the maas ui .. do you see the bootstrap/state server allocated?21:53
DarkmantleYes thats all done21:54
DarkmantleI can SSH into it and do anything21:54
DarkmantleFull network access, full connection to MAAS21:54
hazmatk21:54
hazmatDarkmantle, re pastebin.. apt-get install pastebinit .. handy cli pastebin client21:54
Darkmantleah thanks21:55
DarkmantleIts a re-install so got nooo tools lol21:55
hazmatDarkmantle, --upload-tools is a reasonable workaround21:55
DarkmantleI did that21:55
DarkmantleI did once with once without21:55
Darkmantlehttp://pastebin.com/5i7rtX8U21:56
Darkmantlethe cloud-init-output log has the normal cant connect to 127.0.0.1:37107 error too21:58
Darkmantleeven though mongo can connect using mongo localhost:37017/juju21:58
DarkmantleOr no wait sorry, it gets init call() failed error21:58
hazmatDarkmantle, the mongo you put output for is not the same mongo that juju is running for itself..21:59
DarkmantleSurely it can connect?22:00
* Darkmantle shrugs22:00
hazmatDarkmantle,  its a different process.. /etc/init/mongodb.conf vs /etc/init/juju-db.conf22:00
Darkmantleah yeah22:00
hazmatDarkmantle, could you pastebin the /var/log/cloud-init-output.log22:00
hazmatthat's fixed in the dev (>= 1.17) releases.. no extraneous mongo running.22:01
Darkmantlehttp://paste.ubuntu.com/6984186/22:01
hazmatfwiw22:01
hazmatlooks good22:02
Darkmantlereally?22:02
Darkmantlei assume DNS should point to the MAAS node's IP, right? cause it does22:02
Darkmantleand then MAAS points to .1 in the vlan22:02
hazmatDarkmantle, can your client machine connect to the server? ie.. telnet 97qay.master 3701722:04
DarkmantleYes22:05
=== prismak__ is now known as prismak
DarkmantleI checked all the ports22:05
Darkmantleah no it can't by hostname hazmat , only by IP22:05
Darkmantledefinitely DNS then22:06
hazmatsounds like22:07
Darkmantleugh22:07
hazmatDarkmantle, you could add maas's dns server to your local resolv.conf maybe22:07
Darkmantlei have yeah22:07
Darkmantlei couyld update /etc/hosts since its permanent22:07
Darkmantlewould be easier22:07
Darkmantleand fixed22:08
hazmatcool22:08
Darkmantlequick question, just deployed juju-gui22:09
Darkmantlewhat next? how do i get to it, etc?22:09
hazmatDarkmantle, juju status juju-gui ... go to  https://$ip_address22:10
Darkmantleah duh22:10
Darkmantlemeh, errors22:10
Darkmantlehow long should machines be pending for? :L22:13
hazmatDarkmantle, on maas.. its a bit more dependent on hardware22:13
DarkmantleTrue. I gave the maas VM limited resources22:13
DarkmantleI was told it can cope well on 1g 1CPU22:14
hazmatDarkmantle, you can run juju debug-log -n 10022:14
Darkmantlecannot run instances: gomaasapi: got error back from server: 409 CONFLICT22:14
hazmatDarkmantle, or juju ssh 0  && less /var/log/juju/machine-0.log   which should have any provisioning bits22:14
hazmatDarkmantle, are there other registered/available machines in maas?22:15
Darkmantleok there is only 1 node atm22:15
Darkmantlewith juju on it22:15
Darkmantleim adding a juju-gui noow22:15
hazmatDarkmantle, right.. 409 conflict.. means no additional nodes in maas to hand back to juju22:15
Darkmantleyeah22:15
Darkmantlei have to add a node first then deploy?22:15
hazmatDarkmantle, so there's hulk-smash/manual placement mode.. when deploing..22:15
hazmatDarkmantle, basically.. but re manual-placement.. you can place services onto existing machines with deploy --to=022:16
hazmatfor example.. where 0 is placeholder for any machine id in the juju env22:16
DarkmantleThats true22:16
DarkmantleSo I could add one i MAAS22:16
Darkmantlethen deploy it there?22:16
DarkmantleI thought the point was to have machines for each service22:17
hazmatDarkmantle,  machines or containers..22:17
Darkmantleblah22:17
DarkmantleAH22:17
Darkmantleconfused22:17
DarkmantleOk I get it now22:18
DarkmantleI need to make a new node that I can deploy the services too?22:19
hazmatthumper-afk, ie. you can also do juju deploy --to=lxc:0  and juju will create an lxc container on machine 0 for the service...  deploy --help for more info on that placement stuff22:19
hazmatDarkmantle, yes.22:19
DarkmantleShouldn't it make them automatically?22:19
hazmatDarkmantle, how? juju will request new machines for the provider.. but it can't buy new hardware for maas ;-)22:19
Darkmantleso i hgave to add the new VM's automatically22:20
Darkmantleto host the juju services on22:20
Darkmantlethats loooong :P22:20
Darkmantlebut ok22:20
hazmatDarkmantle, ie. if your on ec2.. it will request new instances for services.. on maas.. the machines have to be registered.. there's maas auto-enlist for racks and dcs,  but if your creating vms as maas machines, you'll have to make new ones to have them show up22:21
Darkmantlefair enough22:21
Darkmantleso make new VM, then add it to maas, then deploy22:22
Darkmantleand it should automatically find it22:22
hazmatyup22:25
DarkmantleI need openstack or something to provision the VM's automatically22:25
DarkmantleTotally should've gone that way22:25
hazmatDarkmantle, or use local provider22:25
DarkmantleTotally stillw ill at some point, MAAS is ok but not as detailed22:25
DarkmantleOr that22:26
DarkmantleMake my juju have the 12gb RAM / 6 CPU's and do it that way22:26
hazmatDarkmantle, maas on vms.. is really just a testing experience.. for charm dev / experiment.. i'd go with local provider.. or a cloud provider22:26
hazmatcloud envs are generally pretty cheap for short lived envs22:26
hazmatie 10 machines for an hr ~ $1 usd.22:27
Darkmantlemhm22:27
Darkmantlehazmat i know its for testing22:28
Darkmantlein fact im running my own test web environment22:28
Darkmantlethats all22:28
Darkmantleblah now PXE borked22:28
Darkmantlethere22:29
* Darkmantle yawns22:29
=== thumper-afk is now known as thumper

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