Support #372
closed
Quality/Testing a.k.a. pre production Infrastructure request for creation
100%
Description
So far we utilised a separate VO for pre production testing. However some applications run at Infrastructure level (e.g. HomeLibrary, Social Apps ). Therefore it is impossible to properly test them before production. This is very risky in same cases, especially now that we have a brand new Home Library.
To overcome this situation I propose the following:
create a new Infrastructure (name to decide, /preprod perhaps?)
remove the VO scope Ecosystem from production and create a new VO Scope under /preprod (name to decide, /preprod/D4Science perhaps?)
What do you think?
Updated by Pasquale Pagano almost 10 years ago
To minimize the impact of a new infrastructure to install, maintain, and manage we need to be sure to configure it to reuse MongoDB, Cassandra, and in the future CouchDb of the production infrastructure at least. Clearly, the areas of the storages dedicated to the pre-production should be completely decoupled by the ones used in production and it should be possible to clean all the persisted status immediately after the upgrade of the production.
It should be also analyzed if it is possible to use the same ActiveMq of the production under the same principles.
The above conditions are needed since a pre-prod should ensure the same QoS of the production and the only affordable way to do it in the short-term is to exploit the production technologies that are properly managed and monitored.
As far as the name I would prefer a simple 'd4s' name for the pre-prod.
Updated by Massimiliano Assante almost 10 years ago
- Status changed from New to In Progress
The trade-off is indeed to create a new infrastructure named /d4s, move the resources of Ecosystem into it, and reuse as much as possible the "external" technologies of production where possible.
For instance we can create a new Cassandra KeySpace (Database) on the same Cassandra Production Cluster, the same applies for CouchDb. We already have a JCR Repo for testing, not sure about Mongo and GeoServer.
Updated by Massimiliano Assante almost 10 years ago
- Status changed from In Progress to Feedback
- Assignee changed from Massimiliano Assante to Roberto Cirillo
Rob, can we start this activity after the gCube 3.8.0 rollout?
Updated by Roberto Cirillo almost 10 years ago
I think, before start with the creation of a new infrastructure is better to discuss about it, prepare an estimate of the cost involved and evaluate whether there are alternatives.
Updated by Massimiliano Assante almost 10 years ago
I'm not sure who has to prepare an estimate of the cost involved, but at least 4 machines dedicated to the root VO (agreed name: /d4science.org) has to be created and the current Ecosystem ones will be reallocated to the /d4science.org/$name2DecideVO
Updated by Massimiliano Assante almost 10 years ago
- Assignee changed from Roberto Cirillo to Pasquale Pagano
- Priority changed from Normal to High
Lino I think we can we start the planning of this activity, should we proceed to create tickets and involve the necessary people assigning them tasks? Perhaps the first week of August could be the starting week?
Updated by Pasquale Pagano almost 10 years ago
- Assignee changed from Pasquale Pagano to Massimiliano Assante
Yes, I think you can plan this activity with Roberto. I suggest to look at the development servers no longer used also and to reassign them to the pre-production environment.
Updated by Massimiliano Assante over 9 years ago
- Status changed from Feedback to Closed
- % Done changed from 0 to 100
A new Sprint was created to track the activity https://support.d4science.org/versions/90
the ticket can be closed
