Skip to content

Instantly share code, notes, and snippets.

@mehrayogesh
mehrayogesh / heat-manage
Created October 18, 2013 17:59
heat-manage --debug db_sync
013-10-18 17:58:16.295 22060 DEBUG heat.openstack.common.lockutils [-] Got semaphore "dbapi_backend" lock /home/ubuntu/heat-hpcloud/heat/heat/openstack/common/lockutils.py:166
2013-10-18 17:58:16.295 22060 DEBUG heat.openstack.common.lockutils [-] Got semaphore / lock "__get_backend" inner /home/ubuntu/heat-hpcloud/heat/heat/openstack/common/lockutils.py:245
2013-10-18 17:58:16.422 22060 DEBUG migrate.versioning.repository [-] Loading repository /home/ubuntu/heat-hpcloud/heat/heat/db/sqlalchemy/migrate_repo... __init__ /usr/local/lib/python2.7/dist-packages/migrate/versioning/repository.py:76
2013-10-18 17:58:16.423 22060 DEBUG migrate.versioning.script.base [-] Loading script /home/ubuntu/heat-hpcloud/heat/heat/db/sqlalchemy/migrate_repo/versions/015_grizzly.py... __init__ /usr/local/lib/python2.7/dist-packages/migrate/versioning/script/base.py:27
2013-10-18 17:58:16.423 22060 DEBUG migrate.versioning.script.base [-] Script /home/ubuntu/heat-hpcloud/heat/heat/db/sqlalchemy/migrate_repo/versions/015_grizzly.p
@mehrayogesh
mehrayogesh / gist:6822849
Last active December 24, 2015 15:49
Extending the Trove/Heat integration and providing the capability of multi-instnace/instance-group creation through externalized heat templates (per service)
This approach is for the generic shift of trove from nova-instance platform to db-instance platform.
The concepts of instance, flavor are considered as the cluster instance, cluster flavor. The create,
delete, list, resize apis are extrapolated to the cluster apis than instance apis. The 3 major chunks are:
1. Externalizing the templates and making them configurable:
The service specific templating package constituting of multiple heat-templates (template/yaml) and a
template-factory (module) encapsulates the templating logic for the service and is responsible for
returning the right template content given a CONTEXT. The template-content is further used for the
creation of the stack. The parameters within the template are also dependent on the CONTEXT. The CONTEXT
can carry the information on number of nodes in the cluster and all the existing parameters (nova-flavors,
@mehrayogesh
mehrayogesh / gist:6798720
Last active December 24, 2015 12:39
Trove/Heat
Trove-Heat Integration Topics:
1. Hardening the heat integration: The instances are not turning ACTIVE on multiple occasions. Further investigation on with the pointer that during some of these cases the prepare method is not called.
2. Heat Events than poll: The polling mechanism to check the status of the heat-stack for completion/failure can be replaced with handling the events from heat and acting accordingly. In the wake of point 1 above, it also needs to be validated that stack- creation-complete status from heat syncs with the instance creation within the stack.
3. Configurable templating: Making the template content as configurable.
4. Trove instance-store updates for multiple instances (governed by the template): In case the template provisions more than one instances then we need a design on the instance state update in trove db.
5. Any further abstractions in trove (may be an instance-group concept) per point 4: Do we need an instance-group like concept in trove for aggregating the multiple instan