0

I have a problem that my application "zdaas" installed by cloudify, it has been worked well for a long time, but today, I find the applicaton "zdaas" is disappearing from application commbox(attachment picture 1), but in the host tab, the machine and usm gsa etc exist normally(picture 2)

then I use Admin to restart GSM, and the result is the application tab has nothing appeared(picture 3).

To solve this problem, I try to restart the management, but the problem always exists

At last, I teardown the cloudify, and reinstall "zdaas". But if my application is running in product enviroment(it works very well), it didnot allow me to teardown.

so, how can I resolve this problem?

Thank you very much!

![enter image description here][1]

![enter image description here][2]

![enter image description here][3]

https: //cloudifysource.zendesk.com/attachments/token/DjjxOBl6CIfPLda5fMpGrI67z/?name=1.png https: //cloudifysource.zendesk.com/attachments/token/DjjxOBl6CIfPLda5fMpGrI67z/?name=2.png https: //cloudifysource.zendesk.com/attachments/token/DjjxOBl6CIfPLda5fMpGrI67z/?name=3.png

yzet
  • 11
  • 3

1 Answers1

0

From the screenshots you sent, it looks like you are using a single Cloudify manager. If it is running without persistence (not saving state to disk) then a restart of the Cloudify Manager host would cause all manager state to be lost.

After such a restart, the agent VMs would reconnect to the manager so you would still see them on the hosts page, but the state of installed services would be lost.

It is recommended to run Cloudify in production with two managers running in a highly-available cluster, so that the failure of any one machine will not cause data loss.

To set the Cloudify manager to run with HA, edit your *-cloud groovy file, and set the numberOfManagementMachines field to a value of 2. When you bootstrap cloudify with this value, 2 manager machines will be started up. Both machines are active, and each backs up the other's data.

cloud {
    ...
    provider {
        ...
        numberOfManagementMachines 2
    }
}
Barak
  • 3,066
  • 2
  • 20
  • 33
  • How to avoid state lost when restart managment? or how to run a highly-available cluster? thank you! – yzet Oct 30 '14 at 04:33
  • I config 2 managment machine, and my system use RestClient to connect the manager and control deploy/undeploy ... when I start 2 management, the rest interface also has two, but my RestClient can only accept one url param, how can i do for this? like locate can use "," to split and add more than one – yzet Nov 25 '14 at 07:01
  • The rest client only accepts a single URL. Any action you send to one server is synchronized with the other. If you encounter a connection error, you should create a new rest client initialization with the URL of the second machine. – Barak Nov 25 '14 at 07:40
  • Thank you very much for your prompt reply. – yzet Nov 25 '14 at 08:13
  • there is another problem: to verify HA, I start two managerment, one is master and another is slave, I simulate the master management mathine down, at this time, the slave management as master management runing, when I start the the original "master" down machine, I got an excetpion: "Replication Channel Disconnected", description is: " A repliction channel has been lost between primary space cloudifyManagementSpace.1[1] and Space CloudifyManagementSpace.1[2] – yzet Nov 25 '14 at 08:40
  • I try to use teardown and shutdown managment two command, but both command execute excepion: – yzet Nov 25 '14 at 08:47
  • That error just indicates that there was a connection problem. It is sporadic. Once both managers are up and synchronized with each other, it will stop. – Barak Nov 25 '14 at 10:42
  • Your'e welcome. I'd appreciate it is you accepted the answer (and the other ones) so they will be marked as closed. – Barak Nov 26 '14 at 06:25