I'm seeing more and more infra services in need of a full-text search engine:
... possibly more in the future, including a log console, etc.
Current services have to rely on available (Python) libraries and there's a big gap between local file-based libraries such as Whoosh or Xapian and full-blown servers like Elasticsearch or Apache SolR.
Elasticsearch is a pretty good solution for fulltext search, and even has some powerful tools to search logs if we want to use that in the future (Logstash & Kibana). We haven't used it yet because it's a big Java app and it's not a tech we're comfortable maintaining.
I'd like, however, to run a test server with Elasticsearch to see if it can be done. If it can be used for HyperKitty & Hubs & more, see if we can deploy it with ansible properly, see what the maintenance load looks like, etc.
A VM in the infra cloud seems perfect but we may have better options nowadays (openshift maybe?). Depending on how much disk space you can dedicate for it, I could run more realistic tests. If we can, let's start with something between 20 and 50G. I can do with less if it's not possible.
I can spin up a cloud instance for this testing. What OS did you want?
m1.medium has 40GB disk, so sounds like that should be ok?
RHEL7 / CentOS7 seems like the best OS for this use case. 40GB is fine, I don't know yet how much I'll need but I can do quite a lot of testing with that.
elastic-dev.fedorainfracloud.org is ready for your testing.
Let us know how it goes...
Metadata Update from @kevin:
- Issue close_status updated to: Fixed
- Issue status updated to: Closed (was: Open)
to comment on this ticket.