Right now the service runs in a screen. :(
Description
Description
Related Objects
Related Objects
- Mentioned In
- Blog Post: Status update (April 25, 2016)
Event Timeline
Comment Actions
Here's the command I use to run the script from the ore-wikimedia-config directory on my local machine. (I have the most recent master of 'ores' installed).
ores precached stream.wikimedia.org https://ores.wmflabs.org --config ores.wmflabs.org.yaml --verbose
Comment Actions
Running even the plain binary ores command does not work:
root@ores-worker-01:/srv/ores# ./venv/bin/ores
Traceback (most recent call last):
File "/srv/ores/venv/lib/python3.4/site-packages/pkg_resources/__init__.py", line 639, in _build_master
ws.require(__requires__)
File "/srv/ores/venv/lib/python3.4/site-packages/pkg_resources/__init__.py", line 940, in require
needed = self.resolve(parse_requirements(requirements))
File "/srv/ores/venv/lib/python3.4/site-packages/pkg_resources/__init__.py", line 832, in resolve
raise VersionConflict(dist, req).with_context(dependent_req)
pkg_resources.ContextualVersionConflict: (requests 2.7.0 (/srv/ores/venv/lib/python3.4/site-packages), Requirement.parse('requests==2.5.3'), {'revscoring'})
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "./venv/bin/ores", line 5, in <module>
from pkg_resources import load_entry_point
File "/srv/ores/venv/lib/python3.4/site-packages/pkg_resources/__init__.py", line 3057, in <module>
working_set = WorkingSet._build_master()
File "/srv/ores/venv/lib/python3.4/site-packages/pkg_resources/__init__.py", line 641, in _build_master
return cls._build_from_requirements(__requires__)
File "/srv/ores/venv/lib/python3.4/site-packages/pkg_resources/__init__.py", line 654, in _build_from_requirements
dists = ws.resolve(reqs, Environment())
File "/srv/ores/venv/lib/python3.4/site-packages/pkg_resources/__init__.py", line 832, in resolve
raise VersionConflict(dist, req).with_context(dependent_req)
pkg_resources.ContextualVersionConflict: (requests 2.7.0 (/srv/ores/venv/lib/python3.4/site-packages), Requirement.parse('requests==2.5.3'), {'revscoring'})
root@ores-worker-01:/srv/ores# lsComment Actions
See https://github.com/wiki-ai/revscoring/pull/150
@yuvipanda, once this PR is merged, then you shouldn't have this version issue.
Comment Actions
nope...
yuvipanda@ores-web-01:~$ /srv/ores/venv/bin/ores
Traceback (most recent call last):
File "/srv/ores/venv/lib/python3.4/site-packages/pkg_resources/__init__.py", line 639, in _build_master
ws.require(__requires__)
File "/srv/ores/venv/lib/python3.4/site-packages/pkg_resources/__init__.py", line 940, in require
needed = self.resolve(parse_requirements(requirements))
File "/srv/ores/venv/lib/python3.4/site-packages/pkg_resources/__init__.py", line 832, in resolve
raise VersionConflict(dist, req).with_context(dependent_req)
pkg_resources.ContextualVersionConflict: (requests 2.7.0 (/srv/ores/venv/lib/python3.4/site-packages), Requirement.parse('requests==2.5.3'), {'revscoring'})
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/srv/ores/venv/bin/ores", line 5, in <module>
from pkg_resources import load_entry_point
File "/srv/ores/venv/lib/python3.4/site-packages/pkg_resources/__init__.py", line 3057, in <module>
working_set = WorkingSet._build_master()
File "/srv/ores/venv/lib/python3.4/site-packages/pkg_resources/__init__.py", line 641, in _build_master
return cls._build_from_requirements(__requires__)
File "/srv/ores/venv/lib/python3.4/site-packages/pkg_resources/__init__.py", line 654, in _build_from_requirements
dists = ws.resolve(reqs, Environment())
File "/srv/ores/venv/lib/python3.4/site-packages/pkg_resources/__init__.py", line 832, in resolve
raise VersionConflict(dist, req).with_context(dependent_req)
pkg_resources.ContextualVersionConflict: (requests 2.7.0 (/srv/ores/venv/lib/python3.4/site-packages), Requirement.parse('requests==2.5.3'), {'revscoring'})Comment Actions
I don't think you are using the most recent version.
See https://github.com/wiki-ai/revscoring/blob/v0.4.10/requirements.txt#L9
Comment Actions
Demonstration:
halfak@ores-staging-01:~$ /srv/ores/venv/bin/ores
Usage:
ores (-h | --help)
ores <utility> [-h | --help]Comment Actions
precached daemon running in a screen on ores-compute. Feel free to kill when you are ready.
Comment Actions
@schana interested in this task? It seems like setting this up in puppet makes sense. Here's the command we run to start the sync server:
ores precached stream.wikimedia.org https://ores.wmflabs.org --config ores-wikimedia-config/config/00-main.yaml
Comment Actions
Note to self from a chat with yuvi:
http://github.com/wikimedia/operations-puppet is the puppet repo
but you can add roles from it using the 'manage instance' link in wikitech
https://wikitech.wikimedia.org/wiki/Help:Instances should hopefully have useful info
You'll just need to create a puppet role that deploys the ores codebase (similar to the others in the ores module) and sets up the precached daemon running via a systemd unit file.
I can't really seem to find examples right now but there's base::service_unit in the repo which should have examples about setting up the systemd stuff.
Not sure if you've already gone through the 'self hosted puppet' dance, if not that's what makes developing puppet roles on that repo easier. http://wikitech.wikimedia.org/wiki/Help:Self-hosted_puppetmaster has info.
Comment Actions
I think this might be helpful for you :)
https://github.com/wikimedia/operations-puppet/tree/production/modules/ores
These are modules, I think they run it using manifests in the manifests folder but I'm probably wrong
Comment Actions
I have coded the part to install systemd service. I have craeted a new class using wikitech "manage puppets group". It is called role::labs::ores::precached.
Now the thing is, I can apply the above role on a server which already has ores installed and this will work fine, precahed will start. I'm not sure if that is way it should be. According to me if I apply role::labs::ores::precached using configure instance link this should complete the setup.
My question is, how should I make sure that ores is installed and then this role is applied?
Comment Actions
Here is the commit for the change:
https://gerrit.wikimedia.org/r/#/c/278555/1
Steps to launch a precached instance:
- Launch an instance inside ores or ores-staging context
- Apply the role role::labs::ores::precached using configure instance
- run fab initialize_server on the instance
Comment Actions
@Halfak Please review when you get a chance. Should I be assigning this card to you (the reviewer)? Not sure what workflow we follow here.
@yuvipanda - do we monitor daemons with tools like monit or supervisord? would like to know your thoughts on how to resurrect precached in case it crashes, or at least we are not blind about its crash.
P.S. Also, sorry for the confusion about the extra traffic to ORES since I launched a preached instance. I should have informed you folks.
Comment Actions
I explained to @Sabya that ores needs to run from source directory (/srv/ores/config/ for fab deploys and /srv/ores/deploy for scap deploys) it's in the puppet, you don't need to install ores.
Comment Actions
Here's what I posted on github:
I just talked this over with @Ladsgroup. We can do this without installing.
python submodules/ores/utility -h works, so we can do "python submodules/ores/utility precached ..." and expect it to work. How much trouble is it to run the precached service from this location?