Hi Dave,

I can see src variables like biocache_hub_war_url pointing to a war file which makes the step essentially copying files over the internet. In that case, I wonder whether it’s possible to deposit the war file on any cloud storage service(GDrive, Dropbox or any) as part of your release workflow, and then modify the Ansible playbook to retrieve the war file directly from the storage service?

I find this copying step is most time-consuming and I believe putting it on the cloud would speed up a lot and facilitate the testing of the ALA setup. But I might have oversight here since I am not familiar with Java development workflow.

What do you think?

Cheers,

Burke

On 03 Jun 2014, at 14:00, Tim Robertson [GBIF] <trobertson@gbif.org> wrote:

Thanks Dave - a good point I hadn’t considered. 
Let’s leave it for now then, but we might want to consider somehow getting the artifacts cached and replicated (nexus seems like a decent option for that).

Cheers,
Tim




On 03 Jun 2014, at 13:50, <David.Martin@csiro.au> <David.Martin@csiro.au> wrote:

Hi Tim,

I made that change during a big refactor - but for selfish reasons.
Use of local repo from my home internet connection to external VMs (EC2 and Nectar - an Aussie provider) made running the scripts from scratch too slow to use/test. Testing with vagrant is very useful, but up until point.
I think Australia lacks the internet speeds (particularly upload speeds) of the scandinavia :)
Perhaps we need a toggle to use local repo or not. Or alternatively a way of replicating artefacts to other maven repositories so that it isnt slow from other countries.

Cheers

Dave
________________________________________
From: ala-portal-bounces@lists.gbif.org [ala-portal-bounces@lists.gbif.org] on behalf of Tim Robertson [GBIF] [trobertson@gbif.org]
Sent: 03 June 2014 21:33
To: ala-portal@lists.gbif.org
Subject: [Ala-portal] Ansible changes removes local caches - intended?

Hi ALA folks

With this commit: https://github.com/gbif/ala-install/commit/d5b6394f44aa82451079d84f646e200caafbd2a8

Things like the following were removed:

-- name: determine if the application is in our local repo
-  local_action: stat path={{local_repo_dir}}/{{biocache_service}}.war
-  register: biocache_war_path
-
-- name: fetch application if it is not in our local repo
-  local_action: get_url url={{biocache_service_url}} dest={{local_repo_dir}}/{{biocache_service}}.war
-  when: biocache_war_path.stat.exists == false

The original design was to pull the large artifacts to a local directory once (the .ala folder) so that if you were provisioning different environments (locally, uat, production etc) you weren’t troubled with huge download times for each environment.

Was this an oversight perhaps?  In Europe we see 10s of minutes in this stage, hence the original design to cache locally.

If we get into proper maven releases and versioned artifacts deployed to nexus, we can start mirroring nexus repositories to bring back data locality too (plus it’ll use the local .m2 repo directory on the host).

Any thoughts on this?  Would you like me to revert those changes?

Thanks,
Tim


_______________________________________________
Ala-portal mailing list
Ala-portal@lists.gbif.org
http://lists.gbif.org/mailman/listinfo/ala-portal


----------------------------------------------------------------------------------------
Tim Robertson - GBIF Head of Informatics - trobertson@gbif.org
Global Biodiversity Information Facility http://www.gbif.org/
GBIF Secretariat, Universitetsparken 15, DK-2100 Copenhagen Ø, Denmark
Tel: +45 3532 1487  Mob: +45 2826 1487  Fax: +45 2875 1480
----------------------------------------------------------------------------------------

_______________________________________________
Ala-portal mailing list
Ala-portal@lists.gbif.org
http://lists.gbif.org/mailman/listinfo/ala-portal