You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@mesos.apache.org by baotiao <ba...@gmail.com> on 2015/05/09 21:01:10 UTC
mesos framework already registered
Hey. we meet the problem
I0510 02:46:36.642375 15327 master.cpp:1625] Framework 20150508-153142-1180331018-5050-15323-0006 (marathon) at scheduler-44923500-1fd5-4686-a1d2-c498db1c0ee3@127.0.0.1:38334 already registered, resending acknowledgement.
I have searched the google for a long time, and I have saw the mesos source code.
I find it cause by we haven't deleted the previous framework infomation, but we don't know how to delete the framework infomation.. any one can help me?
----------------------------------------
陈宗志
Blog:baotiao.github.io
Re: mesos framework already registered
Posted by 陈宗志 <ba...@gmail.com>.
Thanks so much.
After setting the LIBPROCESS_IP, I solve the problem.
On Sun, May 10, 2015 at 1:04 PM, Adam Bordelon <ad...@mesosphere.io> wrote:
> To clear Marathon's frameworkId (and app definitions) from Zookeeper
> state, you can use zkCli.sh to rmr /marathon. Then subsequent attempts to
> register will get a new frameworkId.
>
> Another possibility is that Marathon is able to reach Mesos, but Mesos
> cannot reach Marathon, so its registration acknowledgement messages are
> getting dropped, and Marathon keeps trying to reregister. First make sure
> that you can ping the Marathon host from the Mesos master host (if they're
> not the same). Then make sure that Marathon is registering with a
> non-localhost IP (not 127.0.x.1), by either changing /etc/hosts to map to a
> real IP, or setting LIBPROCESS_IP before starting Marathon. You should see
> in the Mesos master logs what IP it thinks Marathon is registering as.
>
> On Sat, May 9, 2015 at 9:48 PM, baotiao <ba...@gmail.com> wrote:
>
>> Yes, The main problem is after we set up the marathon with mesos.
>> when I deploy a new job in marathon. The job is always in the status of
>> deploying.
>> And after I see the mesos's master log. I guess maybe the framework
>> already registered cause the problem.
>>
>> ----------------------------------------
>> 陈宗志
>>
>> Blog:baotiao.github.io
>>
>>
>>
>>
>> On May 10, 2015, at 04:00, haosdent <ha...@gmail.com> wrote:
>>
>> And when you create MesosSchedulerDriver and pass a FrameworkInfo which
>> contains a vaild FrameworkID, the driver will attempt to "failover".
>>
>> On Sun, May 10, 2015 at 3:27 AM, haosdent <ha...@gmail.com> wrote:
>>
>>> I think this log is a normal info log. Could you provide more details
>>> about you problem? And why you think reregister abnormal.
>>>
>>> On Sun, May 10, 2015 at 3:01 AM, baotiao <ba...@gmail.com> wrote:
>>>
>>>> Hey. we meet the problem
>>>> I0510 02:46:36.642375 15327 master.cpp:1625] Framework
>>>> 20150508-153142-1180331018-5050-15323-0006 (marathon) at
>>>> scheduler-44923500-1fd5-4686-a1d2-c498db1c0ee3@127.0.0.1:38334 already
>>>> registered, resending acknowledgement.
>>>> I have searched the google for a long time, and I have saw the mesos
>>>> source code.
>>>>
>>>> I find it cause by we haven't deleted the previous framework
>>>> infomation, but we don't know how to delete the framework infomation.. any
>>>> one can help me?
>>>> ----------------------------------------
>>>> 陈宗志
>>>>
>>>> Blog:baotiao.github.io
>>>>
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>> --
>>> Best Regards,
>>> Haosdent Huang
>>>
>>
>>
>>
>> --
>> Best Regards,
>> Haosdent Huang
>>
>>
>>
>
--
---
Blog: http://www.chenzongzhi.info
Twitter: https://twitter.com/baotiao <https://twitter.com/#%21/baotiao>
Git: https://github.com/baotiao
Re: mesos framework already registered
Posted by Adam Bordelon <ad...@mesosphere.io>.
To clear Marathon's frameworkId (and app definitions) from Zookeeper state,
you can use zkCli.sh to rmr /marathon. Then subsequent attempts to register
will get a new frameworkId.
Another possibility is that Marathon is able to reach Mesos, but Mesos
cannot reach Marathon, so its registration acknowledgement messages are
getting dropped, and Marathon keeps trying to reregister. First make sure
that you can ping the Marathon host from the Mesos master host (if they're
not the same). Then make sure that Marathon is registering with a
non-localhost IP (not 127.0.x.1), by either changing /etc/hosts to map to a
real IP, or setting LIBPROCESS_IP before starting Marathon. You should see
in the Mesos master logs what IP it thinks Marathon is registering as.
On Sat, May 9, 2015 at 9:48 PM, baotiao <ba...@gmail.com> wrote:
> Yes, The main problem is after we set up the marathon with mesos.
> when I deploy a new job in marathon. The job is always in the status of
> deploying.
> And after I see the mesos's master log. I guess maybe the framework
> already registered cause the problem.
>
> ----------------------------------------
> 陈宗志
>
> Blog:baotiao.github.io
>
>
>
>
> On May 10, 2015, at 04:00, haosdent <ha...@gmail.com> wrote:
>
> And when you create MesosSchedulerDriver and pass a FrameworkInfo which
> contains a vaild FrameworkID, the driver will attempt to "failover".
>
> On Sun, May 10, 2015 at 3:27 AM, haosdent <ha...@gmail.com> wrote:
>
>> I think this log is a normal info log. Could you provide more details
>> about you problem? And why you think reregister abnormal.
>>
>> On Sun, May 10, 2015 at 3:01 AM, baotiao <ba...@gmail.com> wrote:
>>
>>> Hey. we meet the problem
>>> I0510 02:46:36.642375 15327 master.cpp:1625] Framework
>>> 20150508-153142-1180331018-5050-15323-0006 (marathon) at
>>> scheduler-44923500-1fd5-4686-a1d2-c498db1c0ee3@127.0.0.1:38334 already
>>> registered, resending acknowledgement.
>>> I have searched the google for a long time, and I have saw the mesos
>>> source code.
>>>
>>> I find it cause by we haven't deleted the previous framework infomation,
>>> but we don't know how to delete the framework infomation.. any one can help
>>> me?
>>> ----------------------------------------
>>> 陈宗志
>>>
>>> Blog:baotiao.github.io
>>>
>>>
>>>
>>>
>>>
>>
>>
>> --
>> Best Regards,
>> Haosdent Huang
>>
>
>
>
> --
> Best Regards,
> Haosdent Huang
>
>
>
Re: mesos framework already registered
Posted by baotiao <ba...@gmail.com>.
Yes, The main problem is after we set up the marathon with mesos.
when I deploy a new job in marathon. The job is always in the status of deploying.
And after I see the mesos's master log. I guess maybe the framework already registered cause the problem.
----------------------------------------
陈宗志
Blog:baotiao.github.io
> On May 10, 2015, at 04:00, haosdent <ha...@gmail.com> wrote:
>
> And when you create MesosSchedulerDriver and pass a FrameworkInfo which contains a vaild FrameworkID, the driver will attempt to "failover".
>
> On Sun, May 10, 2015 at 3:27 AM, haosdent <haosdent@gmail.com <ma...@gmail.com>> wrote:
> I think this log is a normal info log. Could you provide more details about you problem? And why you think reregister abnormal.
>
> On Sun, May 10, 2015 at 3:01 AM, baotiao <baotiao@gmail.com <ma...@gmail.com>> wrote:
> Hey. we meet the problem
> I0510 02:46:36.642375 15327 master.cpp:1625] Framework 20150508-153142-1180331018-5050-15323-0006 (marathon) at scheduler-44923500-1fd5-4686-a1d2-c498db1c0ee3@127.0.0.1 <ma...@127.0.0.1>:38334 already registered, resending acknowledgement.
> I have searched the google for a long time, and I have saw the mesos source code.
>
> I find it cause by we haven't deleted the previous framework infomation, but we don't know how to delete the framework infomation.. any one can help me?
> ----------------------------------------
> 陈宗志
>
> Blog:baotiao.github.io <http://baotiao.github.io/>
>
>
>
>
>
>
>
> --
> Best Regards,
> Haosdent Huang
>
>
>
> --
> Best Regards,
> Haosdent Huang
Re: mesos framework already registered
Posted by haosdent <ha...@gmail.com>.
And when you create MesosSchedulerDriver and pass a FrameworkInfo which
contains a vaild FrameworkID, the driver will attempt to "failover".
On Sun, May 10, 2015 at 3:27 AM, haosdent <ha...@gmail.com> wrote:
> I think this log is a normal info log. Could you provide more details
> about you problem? And why you think reregister abnormal.
>
> On Sun, May 10, 2015 at 3:01 AM, baotiao <ba...@gmail.com> wrote:
>
>> Hey. we meet the problem
>> I0510 02:46:36.642375 15327 master.cpp:1625] Framework
>> 20150508-153142-1180331018-5050-15323-0006 (marathon) at
>> scheduler-44923500-1fd5-4686-a1d2-c498db1c0ee3@127.0.0.1:38334 already
>> registered, resending acknowledgement.
>> I have searched the google for a long time, and I have saw the mesos
>> source code.
>>
>> I find it cause by we haven't deleted the previous framework infomation,
>> but we don't know how to delete the framework infomation.. any one can help
>> me?
>> ----------------------------------------
>> 陈宗志
>>
>> Blog:baotiao.github.io
>>
>>
>>
>>
>>
>
>
> --
> Best Regards,
> Haosdent Huang
>
--
Best Regards,
Haosdent Huang
Re: mesos framework already registered
Posted by haosdent <ha...@gmail.com>.
I think this log is a normal info log. Could you provide more details about
you problem? And why you think reregister abnormal.
On Sun, May 10, 2015 at 3:01 AM, baotiao <ba...@gmail.com> wrote:
> Hey. we meet the problem
> I0510 02:46:36.642375 15327 master.cpp:1625] Framework
> 20150508-153142-1180331018-5050-15323-0006 (marathon) at
> scheduler-44923500-1fd5-4686-a1d2-c498db1c0ee3@127.0.0.1:38334 already
> registered, resending acknowledgement.
> I have searched the google for a long time, and I have saw the mesos
> source code.
>
> I find it cause by we haven't deleted the previous framework infomation,
> but we don't know how to delete the framework infomation.. any one can help
> me?
> ----------------------------------------
> 陈宗志
>
> Blog:baotiao.github.io
>
>
>
>
>
--
Best Regards,
Haosdent Huang