osrf / cloudsim-grant Goto Github PK

Original report (archived issue) by Hugo Boyer (Bitbucket: hugomatic, GitHub: hugomatic).

Add a filter parameter for resources

Make a createResource middleware: it should expect a req.resourceData or the req.error (in case of error) as input. Its work would be to write a new entry in the database with req.resourceData, and format the response (there is no next to call, it completes the route).

Make a deleteResource middleware: it should expect a req.resourceId (or req.error). It would remove the resource from the database and format the response.

Original report (archived issue) by Louise Poubel (Bitbucket: chapulina, GitHub: chapulina).

Data saved to the database using saveData, added in pull request #25, doesn't show on dump, so it's hard to debug them

Original report (archived issue) by Louise Poubel (Bitbucket: chapulina, GitHub: chapulina).

Currently the Redis client just hangs when it can't do something, way after it was initialized

Original report (archived issue) by Hugo Boyer (Bitbucket: hugomatic, GitHub: hugomatic).

Delete user revokes all permissions for every resource.

Revoking permissions can orphan resources if they have a single user.
This needs to be dealt with:

• resources should always have more than one user? (admin user cannot be deleted)
• resources should be deleted before deleting a user (and destroyed, ie: machines shut-down)
• resources should be transfered to another user (admin)

Comments on this issue are welcome.

Original report (archived issue) by Ian Chen (Bitbucket: Ian Chen, GitHub: iche033).

Currently multiple resources with the same name can be created using createResource. There should be a way to check for duplicate resources, e.g. resourceExist?, or prevent duplicate resources from being created.

Original report (archived issue) by Hugo Boyer (Bitbucket: hugomatic, GitHub: hugomatic).

3 characters is small, and will rollover soon.
string compare won't guarantee order.

A better fix would be to split resource name into:

an id (could be a uuid)
a creation time (for sorting, unless the id has increment part)
a type ("simulator", etc)

The current scheme requires to call the database INCR to generate a unique resource name. This operation is asynchronous and could be avoided.

Original report (archived issue) by Hugo Boyer (Bitbucket: hugomatic, GitHub: hugomatic).

grant and revoke routes do not have a isAuthorized check anymore.
We cannot use ownsResource middleware because we don't know the resourceId

solution: change the /permission routes to: /permisssions/:resourceId, and add an ownResource.

This will impact:

• Server route definitions (auth, sim and portal)
• gz-grant (internally)
• widgets (only to change the version of gz-grant)

I also suggest a setPermissionsRoutes that will reduce code duplication.

Original report (archived issue) by Hugo Boyer (Bitbucket: hugomatic, GitHub: hugomatic).

The auth server on the production environment was missing user data.

This problem could happen when the deployment is interrupted.
It was also noticed that deployment of the portal was taking a long time (5 minutes)

These problems may be related to the code that reads the database data (using too many simultaneous asynchronous calls in parallel)

Original report (archived issue) by Louise Poubel (Bitbucket: chapulina, GitHub: chapulina).

It might not

Original report (archived issue) by Louise Poubel (Bitbucket: chapulina, GitHub: chapulina).

Support passing an array of indentities

Original report (archived issue) by Hugo Boyer (Bitbucket: hugomatic, GitHub: hugomatic).

It was observed that deployment is very slow.
It can take a few minutes for the servers to come back online.

A lot of time is spent reading the database (sasc data).

There is extraneous data should be removed (database swoosh). This can be accomplished automatically during each deployment, and it would make the next one faster. We could also trigger the operation manually (when backing up the data)

Original report (archived issue) by Louise Poubel (Bitbucket: chapulina, GitHub: chapulina).

• Create some resources

• Call clearDb

• Call getNextResourceId -> the id count continues from the previous number as if the db had never been cleared.

Not sure if this is a feature or a bug, just thought it should be pointed out.

Original report (archived issue) by Ian Chen (Bitbucket: Ian Chen, GitHub: iche033).

username string -> identities array

Original report (archived issue) by Louise Poubel (Bitbucket: chapulina, GitHub: chapulina).

See this for context

Original report (archived issue) by Hugo Boyer (Bitbucket: hugomatic, GitHub: hugomatic).

Cloudsim-grant should implement data consistency checks, that could optionally be relaxed during production.

• Each resource should have at least one user with read/write access.
• Depending on the collection, each resource could have a unique name

These checks should be run after each database operation, ensuring there is no data corruption possible.

Original report (archived issue) by Hugo Boyer (Bitbucket: hugomatic, GitHub: hugomatic).

The shared database for the dev servers is still using a t2.micro.

Latency is high. The server should be upgraded (like on production)

Original report (archived issue) by Louise Poubel (Bitbucket: chapulina, GitHub: chapulina).

This would be especially useful when testing, the same way we use it for "normal" resources