Closed Bug 1189164 Opened 6 years ago Closed 6 years ago

Quick follow up to test that monotonic clock fix solves negative time for last subsession

Categories

(Cloud Services Graveyard :: Metrics: Pipeline, defect, P2)

defect

Tracking

(Not tracked)

RESOLVED WORKSFORME

People

(Reporter: kparlante, Unassigned)

References

Details

(Whiteboard: [unifiedTelemetry][data-validation])

Perhaps write a heka filter to check to see if these negative values are found as soon as the fix hits nightly.

Also good to re-run bcolloran's scripts once enough data is available: https://bugzilla.mozilla.org/show_bug.cgi?id=1186713#c1
This should be in Nightly>=20150730, we should filter for that build id.
I eyeballed this in kibana (the table shows the newer build id pings, sorted by subsessionLength): https://kibana.shared.us-west-2.prod.mozaws.net/#dashboard/temp/AU76r2croEkfX-6KQ_Co

Only 3 pings with negative subsession lengths so far (not counting the "matching" saved session ping).
(In reply to Katie Parlante from comment #2)
> I eyeballed this in kibana (the table shows the newer build id pings, sorted
> by subsessionLength):
> https://kibana.shared.us-west-2.prod.mozaws.net/#dashboard/temp/
> AU76r2croEkfX-6KQ_Co
> 
> Only 3 pings with negative subsession lengths so far (not counting the
> "matching" saved session ping).

Should have mentioned that was over a 2 day period. Here's a 7 day period that includes 7/30:
https://kibana.shared.us-west-2.prod.mozaws.net/#/dashboard/temp/AU76r2croEkfX-6KQ_Co

~15 negative subsession lengths (the earlier builds are generating a lot more).

At any rate, not as good as an actual comprehensive analysis, but early indications suggest improvements.
Odd, with the current implementation i expect no negative subsessionLength fields.
I wonder if the remainders are local builds who have not updated their source yet - if so this should drop down to near-zero soon (and the remainders should only affect a very small set of clients).
Note that both -60 subsessionLengths that i can see now are from the same client id.
Those are both the same value, once in a shutdown and once in a saved-session ping.

Given oddities on Nightly i'd wait a bit for more data before drawing conclusions from that.
Iteration: --- → 43.1 - Aug 24
Priority: -- → P2
I only see 4 submissions (duped as saved-session & shutdown) with that issue in the query:
https://kibana.shared.us-west-2.prod.mozaws.net/#/dashboard/temp/AU76r2croEkfX-6KQ_Co

That is a really low numbers and might just be people building from outdated source etc.
Status: NEW → RESOLVED
Closed: 6 years ago
Resolution: --- → WORKSFORME
Product: Cloud Services → Cloud Services Graveyard
You need to log in before you can comment on or make changes to this bug.