We currently have two probes to measure performance of A-S queries: - FENNEC_ACTIVITY_STREAM_TOPSITES_LOADER_TIME_MS measures amount of time top site query takes - FENNEC_ACTIVITY_STREAM_HIGHLIGHTS_LOADER_TIME_MS measures amount of time highlights query takes ... and I can't seem to find any data for the latter. Former probe is collecting data: https://telemetry.mozilla.org/new-pipeline/dist.html#!cumulative=0&end_date=2016-12-12&keys=__none__!__none__!__none__&max_channel_version=nightly%252F53&measure=FENNEC_ACTIVITY_STREAM_TOPSITES_LOADER_TIME_MS&min_channel_version=null&product=Fennec&sanitize=1&sort_keys=submissions&start_date=2016-11-22&table=0&trim=1&use_submission_date=0

I've stepped through in a debugger, and we're definitely passing "FENNEC_ACTIVITY_STREAM_HIGHLIGHTS_LOADER_TIME_MS" into nativeAddHistogram. And we seem to have correctly defined the histogram in Histograms.json: https://dxr.mozilla.org/mozilla-central/source/toolkit/components/telemetry/Histograms.json?q=path%3AHistograms.json&redirect_type=single#6262 (I.e. there doesn't seem anything obvious causing this...)

(In reply to Andrzej Hunt :ahunt from comment #1) > (I.e. there doesn't seem anything obvious causing this...) Maybe it's not "us"? I.e. could it be that there's something amiss on the data pipeline side of things? A la Bug 1330179.

Frank, any idea what might be causing this?

So I made a small CEP to check incoming messages for this field, and it's not finding anything [0]. I'd recommend checking "about:telemetry" and see if it shows when expected. [0] CEP which just searches for the string "FENNEC_ACTIVITY_STREAM_HIGHLIGHTS_LOADER_TIME_MS" in the message payload: https://pipeline-cep.prod.mozaws.net/dashboard_output/graphs/analysis.frank.fennec_element_finder.messages_per_minute.html

(In reply to Frank Bertsch [:frank] from comment #4) > So I made a small CEP to check incoming messages for this field, and it's > not finding anything [0]. I'd recommend checking "about:telemetry" and see > if it shows when expected. Oops, I'd forgotten about about:telemetry. I see both _TOPSITES_ and _HIGHLIGHTS_ there, so it looks like the data gets at least that far. (Reloading Activity Stream results in additional readings appearing on about:telemetry too, for both of those histograms.)

Gah, I'm so sorry. We switched the aggregator over to main pings in late 2016, and as such fennec data stopped being aggregated. We only recently discovered this. The last day was 2016/12/12. I can't find the bug for this at the moment.

(In reply to Frank Bertsch [:frank] from comment #6) > Gah, I'm so sorry. We switched the aggregator over to main pings in late > 2016, and as such fennec data stopped being aggregated. We only recently > discovered this. The last day was 2016/12/12. I can't find the bug for this > at the moment. That explains why I haven't seen any data for 54 yet.

I've got a list of counts of pings containing a histogram for each histogram (from Fennec) [0]. This contains a few counts for each FENNEC_ACTIVITY_STREAM*, so I can confirm it's being received. I'm going to be adding functionality to the aggregator for you, in the meantime, if you need to analyze this data, you can use ATMO. Docs are here [1]. [0] List of histograms and their ping counts: https://pipeline-cep.prod.mozaws.net/dashboard_output/analysis.frank.histogram_counts.types.txt [1] Docs for ATMO and Spark: https://wiki.mozilla.org/Telemetry/Custom_analysis_with_spark

I just want to note that if you guys need specific questions answered soonish, feel free to ping me about doing some analysis. I understand that it is sometimes a big undertaking to start doing analysis on ATMO with Spark.

(In reply to Frank Bertsch [:frank] from comment #8) > I've got a list of counts of pings containing a histogram for each histogram > (from Fennec) [0]. This contains a few counts for each > FENNEC_ACTIVITY_STREAM*, so I can confirm it's being received. Thanks for checking this out, Frank. Counts for those events look correct (they're supposed to be very similar and, for now, super low). We're aiming to have our telemetry story for A-S in a good place by mid-month, so don't worry about analyzing these numbers for now - but it'll be great to have them aggregating. Please link to any additional bug(s) which track relevant work.