[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Treating Nodes like cattle



Hi Dan,

I tested it with 2 nodes in 2 zones with an HA application.

I shutdown the node where the head gear was running.
And I called `oo-admin-repair --removed-nodes`.

The console fails on my application page, until I update mongodb

> a = db.applications.findOne({_id: xxxxxxx })
> a.gears[0].app_dns = true
> db.applications.save(a)


Deployments were not pushed to the broker until I force OPENSHIFT_GEAR_DNS to equal OPENSHIFT_APP_DNS for the new head gear.

$ cp /var/lib/openshift/542132c82ae26179ff00002e/.env/OPENSHIFT_APP_DNS /var/lib/openshift/542132c82ae26179ff00002e/.env/OPENSHIFT_GEAR_DNS

To make my application HA again, I update mongodb

> a = db.applications.findOne({_id: xxxxxxx })
> a.gears[0].host_singletons = true
> a.ha = false
> db.applications.save(a)

And call `rhc app make-ha` again.

Do you kown if this is enough, or am I missing something ?

Cheers
Romain

2014-09-20 16:24 GMT+02:00 Dan McPherson <dmcphers redhat com>:
There is:

oo-admin-repair --removed-nodes

-Dan


On 20/09/14 03:30, Romain wrote:
Hi there,

With the HA features, I was wondering if there is some tools to declare that a gear is lost forever. So it will remove the gear reference in mongodb, gear haproxy, DNS, gear registry...

I would like to setup OpenShift Nodes on ephemeral disks.

Cheers
Romain


_______________________________________________
dev mailing list
dev lists openshift redhat com
http://lists.openshift.redhat.com/openshiftmm/listinfo/dev



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]