The latest upstream version of jvm-tools is 0.8 0.17.0 (we are currently on 0.3.7). The new version adds support for (at least), flame graphs(!)
Wishlist / While we are here:
P.S. Happy to submit a patch for this, but I was unable to locate the source repository
Indeed the source was present only as .dsc, I've imported it into https://gerrit.wikimedia.org/r/#/admin/projects/operations/debs/jvm-tools
I've taken a stab at this, but since I'm not 100% certain what the intention was, procedure-wise, I've pushed it to https://github.com/eevans/jvm-tools for the time being.
What I did:
Obviously, the result doesn't lend itself well to a Gerrit changeset for review.
@fgiunchedi, was this what you had in mind?
Indeed @Eevans that looks good to me. Though I'm open to suggestions on e.g. branch workflow and such, the package should be buildable with gbp now.
Regarding this point, and after having discussed this a bit with @Joe and @fgiunchedi, we seem to have a reluctant consensus that if you merged directly after step #3 above (i.e. without creating a gerrit code review), then #4 could be reviewed in isolation. So the revised process would be:
So, can someone either a) grant me the needed karma to https://gerrit.wikimedia.org/r/#/admin/projects/operations/debs/jvm-tools, or b) perform steps 1-4 above (or alternately, complete step #4 from https://github.com/eevans/jvm-tools/commit/4672fb9 where I have already performed steps 1-3).
I've completed steps 1-3 from above as suggested and pushed the result, not great in terms of coordination required but workable I think for now. If it becomes painful (i.e. more frequent updates needed) in the future we can reconsider. How does that look @Eevans ?
@Eevans @fgiunchedi is there a plan to resume this work or should we close this ticket? Do we still want the newer jvm-tools version? I assume so, hence setting as stalled.
I'm not sure how to parse this. I would love to this done, but it is quite stalled, yes.
I did not sentence well. I'm going to move it to one of the categories that will get triaged into Clinic Duty and hopefully we can line it up.
Question has been answered and this seems still wanted, hence resetting task status as task should not be stalled forever.
looking at the task history it seems this is now all done
- Updated debian/ accordingly
@Eevans are you planning to work on this step? if not which team should we route this too, thanks
Since the idea was to get a new upstream, and many (many, many) new upstream releases have occurred in the six years since, we should probably start anew. :) Honestly, given the amount of time that has passed, it might even be worth looking to see what (Debian-packaged) alternatives exist.
As to where to route it: It's proven quite useful over the years for Cassandra, I honestly don't know if we're using it elsewhere; If it's a Cassandra-only thing, I reckon it should be routed to Data-Persistence.