|operations/puppet : production||Add the visualdiff module + instantiate visualdiffing services|
|operations/puppet : production||Make testreduce generic and instantiate parsoid-rt services|
|operations/puppet : production||Add testreduce module and role|
|Declined||RobH||T116090 Dedicated server for running Parsoid's roundtrip tests to get reliable parse latencies and use as perf. benchmarking tests|
|Resolved||ssastry||T118778 Puppetize parsoid testing services on ruthenium|
|Resolved||ssastry||T124480 parsoid roles: convert upstart to systemd|
|Resolved||jcrespo||T124704 puppetize mysql m5-master access on parsoid-test server|
|Resolved||jcrespo||T124703 Need databases provisioned for parsoid-rt testing, visual diff testing|
Recording some notes from today's puppet investigation.
As it turns out, there are multiple things that need puppetizing ... mysql / mariadb, uprightdiff, testreduce, visualdiff, parsoid, and nginx right now which proxies the parsoid-tests.* urls to one of these services.
So, I am just going to bite the bullet and create a couple new puppet roles, role::parsoid::rt_test, role::parsoid::visual_diff.
parsoid::rt_test will require testreduce, mysql, nginx service and set up template and config files for them.
parsoid::visualdiff will require testreduce, mysql, nginx, uprightdiff, visualdiff and set up template and config files for them.
I don't see an obvious simple way of doing this.
@faidon is suggesting that it might be simpler for us to do this on labs since operations has now made it possible to allocate physical servers via labs. That would be a lot simpler since we'll have root and we can install whatever we want without needing to puppetize this stuff. That might turn out to be a simpler option. We could just throw the necessary config files and some simple bash scripts into a gerrit repo.