Not sure if you need add a new parameter default_accepted_resource_roles , even if you start marathon with role dev, it can still get resources from both role dev and *.
Please refer to https://github.com/mesosphere/marathon/blob/master/docs/docs/command-line-flags.md for marathon configuration. - v0.9.0 --default_accepted_resource_roles (Optional. Default: all roles): Default for the "acceptedResourceRoles"attribute as a comma-separated list of strings. All app definitions which do not specify this attribute explicitly use this value for launching new tasks. Examples: *, production,*, production On Sun, May 1, 2016 at 3:21 PM, haosdent <[email protected]> wrote: > could you show the result of curl 10.132.46.71:5050/ > <http://10.132.46.71:5050/slaves>roles It shows consume "dev" in my side. > > On Sun, May 1, 2016 at 3:04 PM, caogaojin <[email protected]> > wrote: > >> @Haosdent, by the way, i’m using mesos:0.28.1, not the master branch. and >> marathon version is 1.1.1 >> >> -- >> Best Regards, >> >> GaoJin CAO >> ChinaMobile(Suzhou) R&D >> Mail：caogaojin at cmss.chinamobile.com >> >> On May 1, 2016, at 14:04, haosdent <[email protected]> wrote: >> >> Hi, @caogaoin. I try in with the master branch of Mesos and Marathon. And >> it works fine for me. >> >> >> ``` >> INFO Finished processing dda04940-55b6-4ba7-a7bb-75c2108f3151-O8. Matched >> 1 ops after 2 passes. cpus(dev) 2.0; mem(dev) 1024.0; cpus(*) 5.0; mem(*) >> 29808.0; disk(*) 915273.0; ports(*) 31000->31254,31256->32000 left. >> (mesosphere.marathon.core.matcher.manager.impl.OfferMatcherManagerActor:marathon-akka.actor.default-dispatcher-2) >> ``` >> >> Let me list my verify steps to help you find your problem when start your >> tasks. >> >> 1. start Mesos master and agent, and reserve resources for dev. >> >> ``` >> $ ./bin/mesos-master.sh --work_dir=/tmp/mesos --ip=127.0.0.1 >> --hostname=127.0.0.1 >> >> $ ./bin/mesos-slave.sh --work_dir=/tmp/mesos --ip=127.0.0.1 >> --hostname=127.0.0.1 --master=127.0.0.1:5050 >> >> >> $ curl -i -d slaveId=dda04940-55b6-4ba7-a7bb-75c2108f3151-S0 \ >> -d resources='[ >> { >> "name": "cpus", >> "type": "SCALAR", >> "scalar": { >> "value": 2.0 >> }, >> "role": "dev", >> "reservation": { >> } >> >> }, >> { >> "name": "mem", >> "type": "SCALAR", >> "scalar": { >> "value": 1024.0 >> >> }, >> "role": "dev", >> "reservation": { >> } >> } >> ]' -X POST http://127.0.0.1:5050/reserve >> >> 2. start Marathon with dev role >> >> ``` >> ./bin/start --master 127.0.0.1:5050 --zk zk://10.10.120.61:2181/marathon >> --mesos_role dev >> ``` >> >> 3. start task >> >> ``` >> $ cat task.json >> { >> "id": "/test-health", >> "cmd": "sleep 200", >> "cpus": 1.0, >> "mem": 128.0 >> } >> >> $ curl -X POST -H "Content-type: application/json" >> http://127.0.0.1:8080/v2/apps -d @task.json >> ``` >> >> On Sun, May 1, 2016 at 12:18 PM, haosdent <[email protected]> wrote: >> >>> Sorry for make you confusing before. The offer looks match resource >>> requirement according to your log, let me reproduce it first. >>> >>> On Sun, May 1, 2016 at 8:57 AM, Guangya Liu <[email protected]> wrote: >>> >>>> Gaojin, >>>> >>>> Some questions and suggestions here: >>>> >>>> 1) How did you deploy the nginx with new marathon URI? I did not found >>>> that I can configure the new marathon URI with nginx. >>>> >>>> 2) Can you please send out the curl output for your master node: >>>> a) curl "http://master-ip:5050/master/slaves >>>> <http://9.21.51.123:5050/master/slaves>" 2>/dev/null|python -m json.tool >>>> >>>> >>>> b) curl "http://master-ip:5050/master/state >>>> <http://9.21.51.123:5050/master/slaves>" 2>/dev/null|python -m json.tool >>>> >>>> >>>> 3) For open source DC/OS issues, it is suggested to post your question >>>> to [email protected] where you can always get instant and professional >>>> support. >>>> >>>> Thanks, >>>> >>>> Guangya >>>> >>>> >>>> On Sun, May 1, 2016 at 1:49 AM, caogaojin < >>>> [email protected]> wrote: >>>> >>>>> Hi list, >>>>> >>>>> I’m currently using dynamic reservation mechanism with mesosphere/DCOS >>>>> 1.7, the opensourced version, >>>>> and something wired just happened. I tried to setup a new marathon >>>>> service from the universe repo, >>>>> and i did a slight modification, setting mesos_role to `dev`. i also >>>>> turned off mesos anthentication >>>>> mechenism. >>>>> >>>>> My steps are: >>>>> >>>>> First i tried to reserve 2 cpu and 1743 ram to `dev` role using http >>>>> endpoint, and it works: >>>>> # curl -i -d slaveId=e6d590b1-5f2d-4be9-93fd-2fac364584cf-S11 \ >>>>> -d resources='[ >>>>> { >>>>> "name": "cpus", >>>>> "type": "SCALAR", >>>>> "scalar": { >>>>> "value": 2.0 >>>>> }, >>>>> "role": "dev", >>>>> "reservation": { >>>>> } >>>>> >>>>> }, >>>>> { >>>>> "name": "mem", >>>>> "type": "SCALAR", >>>>> "scalar": { >>>>> "value": 1024.0 >>>>> >>>>> }, >>>>> "role": "dev", >>>>> "reservation": { >>>>> } >>>>> } >>>>> ]' -X POST http://10.132.46.81:5050/unreserve >>>>> >>>>> Second, the slave shows following infomation: >>>>> # curl http://10.132.46.81/mesos/slaves >>>>> reserved_resources_full: { >>>>> dev: [ >>>>> { >>>>> name: "cpus", >>>>> type: "SCALAR", >>>>> scalar: { >>>>> value: 2 >>>>> }, >>>>> role: "dev", >>>>> reservation: { } >>>>> }, >>>>> { >>>>> name: "mem", >>>>> type: "SCALAR", >>>>> scalar: { >>>>> value: 1743 >>>>> }, >>>>> role: "dev", >>>>> reservation: { } >>>>> } >>>>> ] >>>>> }, >>>>> >>>>> Third, i tried to run nginx using the new marathon service, which i >>>>> also set the task only accept `dev` resources, >>>>> >>>>> ID: /nginx >>>>> ... >>>>> Resource Roles: dev >>>>> >>>>> 4. on marathon ui i just see the waiting status, which means there’s >>>>> no suitable resource offer, marathon reports: >>>>> >>>>> Insufficient resources for [/nginx] (need cpus=1.0, mem=128.0, >>>>> disk=0.0, ports=(1 dynamic), available in offer: [id { value: >>>>> "e6d590b1-5f2d-4be9-93fd-2fac364584cf-O7476" } framework_id { value: >>>>> "e6d590b1-5f2d-4be9-93fd-2fac364584cf-0006" } slave_id { value: >>>>> "e6d590b1-5f2d-4be9-93fd-2fac364584cf-S11" } hostname: "10.132.46.84" >>>>> resources { name: "ports" type: RANGES ranges { range { begin: 8182 end: >>>>> 32000 } } role: "dev" reservation { } } resources { name: "cpus" type: >>>>> SCALAR scalar { value: 2.0 } role: "dev" reservation { } } resources { >>>>> name: "mem" type: SCALAR scalar { value: 1743.0 } role: "dev" reservation >>>>> { >>>>> } } resources { name: "ports" type: RANGES ranges { range { begin: 1025 >>>>> end: 2180 } range { begin: 2182 end: 3887 } range { begin: 3889 end: 5049 >>>>> } >>>>> range { begin: 5052 end: 6326 } range { begin: 6328 end: 8079 } range { >>>>> begin: 8082 end: 8180 } } role: "*" } resources { name: "disk" type: >>>>> SCALAR >>>>> scalar { value: 41497.0 } role: "*" } resources { name: "cpus" type: >>>>> SCALAR >>>>> scalar { value: 1.0 } role: "*" } resources { name: "mem" type: SCALAR >>>>> scalar { value: 896.0 } role: "*" } url { scheme: "http" address { >>>>> hostname: "10.132.46.84" ip: "10.132.46.84" port: 5051 } path: "/slave(1)" >>>>> }] >>>>> (mesosphere.mesos.TaskBuilder:marathon-akka.actor.default-dispatcher-19) >>>>> >>>>> can anyone help me with this? >>>>> >>>>> thank you. >>>>> >>>>> >>>>> GaoJin CAO >>>>> ChinaMobile(Suzhou) R&D >>>>> Phone：18896723732 >>>>> Mail：caogaojin at cmss.chinamobile.com >>>>> >>>>> >>>> >>> >>> >>> -- >>> Best Regards, >>> Haosdent Huang >>> >> >> >> >> -- >> Best Regards, >> Haosdent Huang >> >> >> > > > -- > Best Regards, > Haosdent Huang >
