-
Notifications
You must be signed in to change notification settings - Fork 10
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
One tool #429
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
So does this completely eliminate the concept of the user creating a profiler engine themselves? I see a couple places where you removed it from a set of roles that are being checked for so I'm guessing it does. If so, we are probably going to have CI issues with this that I will need to resolve since we explicitly test profiler engine creation.
Also, I think I identified a couple of local changes that slipped in that we don't actually want to merge upstream.
I must be inadvertently removing something I did not understand had a purpose. Is there a specific example of a crucible run that has a user creating a profiler engine? I can then use that to make sure it works here. |
I think the intended purpose of this functionality was to be able to launch profilers on something like a KVM host, storage server, etc. where you wanted to collect data but weren't actually running a workload. |
Ahh, OK, I misunderstood your question. I thought you were referring to some kind of user-provided profiler, which we of course don't have [yet], and so it did not make sense to me. I'll make sure we can still include remotehost endpoints which are only profilers. I'll test that next. |
- engine naming needs to be resolved to allow multiple remotehost endpoints
- Engines which run tools need to have a more specific label, one that includes th endpoint-label, to avoid duplicate labels like "profiler-1" - Logic for adding RB followers was a bit broken, only looking at the first RB message with new followers. - There is still work to do to avoid duplicate tool collection when more than 1 remotehost endpoint use the sme host (one could argue to not do this, but I'd like it to just avoid duplicte tools in case someone does).
-reducing some code duplication -documenting globals in each function -cpu-part needs to be fixed still
-other minor cleanups
-it's possible that a user used a remotehost endpoint as a "profiler" only, but then the tool-params.json was empty, resulting in an endpoint that still runs but doe snot launch engines. A small change to not create env-vars.json is needed.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
First pass review. I plan on running some tests with it myself just to better understand it, but I'll get to that later.
This PR will have multiple commits for different areas of work:
plus any changes needed in rickshaw-*: