Ah interesting.. I reported exactly the same problem with NFS the other day!
@Luke: we also had some issues with 1.8.0, 1.8.3 seems to have fixed things for us though! ________________________________ From: craig w [[email protected]] Sent: 23 October 2015 14:54 To: [email protected] Subject: Re: Docker Private Registry problem when update from 0.23 to 0.25 @Luke - Good to know. I have a similar setup, only different being our timeout is at 5 minutes. We had problems using docker 1.8.x with a private registry that was backed by NFS, apparently we're not alone (https://github.com/docker/docker/issues/15833). On Fri, Oct 23, 2015 at 9:52 AM, Luke Amdor <[email protected]<mailto:[email protected]>> wrote: We currently run Mesos 0.25.0 with Marathon 0.11.1 with a private docker registry (distribution v2.1.1, no creds required) and haven't had any weird problems running docker containers. We however do run our marathon with a really high task_launch_confirm_timeout of 600000 (10 minutes) as we were running into problems with docker pulls and staged tasks. We also run docker version 1.7.1 as we ran into problems with pulls on docker 1.8.0. We haven't yet tried the newest docker release to see if it's helped. On Fri, Oct 23, 2015 at 6:50 AM, craig w <[email protected]<mailto:[email protected]>> wrote: I wasn't sure because of this comment: https://github.com/mesosphere/marathon/pull/2462#issuecomment-148703383 On Fri, Oct 23, 2015 at 7:41 AM, haosdent <[email protected]<mailto:[email protected]>> wrote: According https://github.com/mesosphere/marathon/pull/2462 , I think the patches merge into 0.12.0-RC1 and 0.11.1 doesn't contain yet. On Fri, Oct 23, 2015 at 7:32 PM, craig w <[email protected]<mailto:[email protected]>> wrote: Has anyone confirmed that Mesos 0.25.0 with Marathon 0.11.1 with a private docker registry (without credentials) works? On Thu, Oct 15, 2015 at 4:51 PM, Jan Stabenow <[email protected]<mailto:[email protected]>> wrote: Marathon 0.11.1 + Mesos 0.25.0 + private Docker-Reg (without credentials) works, but with a new error. Marathon (or Mesos) finished each task with an "finished-state" immediately. So let’s wait… https://github.com/mesosphere/marathon/pull/2462#issuecomment-148482471 Am 15.10.2015 um 21:29 schrieb Jan Stabenow <[email protected]<mailto:[email protected]>>: Hey Brian, sorry for my late response. I'll try the latest 0.11.1. This may be one of the problems: https://github.com/mesosphere/marathon/pull/2415 Am 15.10.2015 um 21:14 schrieb craig w <[email protected]<mailto:[email protected]>>: I've successfully pulled images from a private registry using Mesos 0.24.1 and marathon 0.11. I use forcePullImage: true. On Oct 15, 2015 2:25 PM, "Brian Devins" <[email protected]<mailto:[email protected]>> wrote: Has anyone found some resolution for this? I have the mesosphere repo versions of mesos and marathon. 0.24.1 and 0.11.0 respectively. I have also tried 0.10.1 for marathon but still no dice. I can manually pull the images if I log into the box. On Wed, Oct 14, 2015 at 2:30 AM, Joachim Andersson <[email protected]<mailto:[email protected]>> wrote: Those images I have tried with Marathon 0.11 I use "disk": 512, Here is my app config. { "id": "/marathonkempaffb060830d6ad074c66923b77f75e199a4c8a23", "env": { "GIT_CALCULATED_BRANCH": "affb060830d6ad074c66923b77f75e199a4c8a23", "marathonkemp_kemp_password": "!12345!m", }, "instances": 1, "cpus": 0.5, "mem": 1024, "disk": 512, "executor": "", "constraints": [ [ "hostname", "UNIQUE" ] ], "container": { "type": "DOCKER", "volumes": [], "docker": { "image": "dockerregistry.dte.loc:80/marathonkemp:affb060830d6ad074c66923b77f75e199a4c8a23", "network": "BRIDGE", "portMappings": [ { "containerPort": 8080, "hostPort": 0, "servicePort": 10000, "protocol": "tcp" }, { "containerPort": 8000, "hostPort": 0, "servicePort": 10001, "protocol": "tcp" }, { "containerPort": 9010, "hostPort": 0, "servicePort": 10002, "protocol": "tcp" } ], "privileged": false, "parameters": [], "forcePullImage": false } }, "healthChecks": [ { "path": ".", "protocol": "HTTP", "portIndex": 0, "gracePeriodSeconds": 360, "intervalSeconds": 30, "timeoutSeconds": 10, "maxConsecutiveFailures": 3, "ignoreHttp1xx": false } ] } Remove a env parameters <c99478.png> Joachim Andersson Systemutvecklare IT | Utveckling Resurs Bank AB Ekslingan 9 Box 22209, SE-250 24 Helsingborg Direkt Tfn: +46 700 528171 Mobil: +46 700 528171 Växel: +46 42 38 20 00<tel:%2B46%2042%2038%2020%2000> E-post: [email protected]<mailto:[email protected]> Webb: www.resursbank.se<http://www.resursbank.se/> Från: Jan Stabenow [mailto:[email protected]<mailto:[email protected]>] Skickat: den 13 oktober 2015 15:52 Till: [email protected]<mailto:[email protected]> Ämne: Re: Docker Private Registry problem when update from 0.23 to 0.25 Hey! I had the same problems when upgrading 0.23 > 0.24.1 +! marathon 0.10 > 0.11 Have you set the disk-limit in your app-config? If not, set this ("disk": "value") and try it again. Looks like there's a problem with Marathon, but not sure. Best Regards, Jan Am 13.10.2015 um 14:14 schrieb Joachim Andersson <[email protected]<mailto:[email protected]>>: When i deploed with forcePullImage=true it dosn't work in Marathon 0.11 but when I downgraded to Marathon 0.10 it works, <c3b4c2.png> Joachim Andersson Systemutvecklare IT | Utveckling Resurs Bank AB Ekslingan 9 Box 22209, SE-250 24 Helsingborg Direkt Tfn: +46 700 528171 Mobil: +46 700 528171 Växel: +46 42 38 20 00<tel:%2B46%2042%2038%2020%2000> E-post: [email protected]<mailto:[email protected]> Webb: www.resursbank.se<http://www.resursbank.se/> Från: Joachim Andersson [mailto:[email protected]] Skickat: den 13 oktober 2015 13:24 Till: [email protected]<mailto:[email protected]> Ämne: Docker Private Registry problem when update from 0.23 to 0.25 Hi, I updated from 0.23 to 0.25 of mesos and to Marathon 0.11 When running docker containers from the public repository it works fine. But when I try to run from my local repository the images don't start. If I start them on my slave with docker run repository:80/image it works fine. Anyone have any simmular problem or any idea what it could be? //Joachim <image001.png> Joachim Andersson Systemutvecklare IT | Utveckling Resurs Bank AB Ekslingan 9 Box 22209, SE-250 24 Helsingborg Direkt Tfn: +46 700 528171 Mobil: +46 700 528171 Växel: +46 42 38 20 00<tel:%2B46%2042%2038%2020%2000> E-post: [email protected]<mailto:[email protected]> Webb: www.resursbank.se<http://www.resursbank.se/> -- https://github.com/mindscratch https://www.google.com/+CraigWickesser https://twitter.com/mind_scratch https://twitter.com/craig_links -- Best Regards, Haosdent Huang -- https://github.com/mindscratch https://www.google.com/+CraigWickesser https://twitter.com/mind_scratch https://twitter.com/craig_links -- Luke Amdor | Platform Lead Architect | Banno Des Moines IA 50309 | Cell 515.231.4033<tel:515.231.4033> -- https://github.com/mindscratch https://www.google.com/+CraigWickesser https://twitter.com/mind_scratch https://twitter.com/craig_links
