Closed
Bug 1305742
Opened 9 years ago
Closed 9 years ago
Add gfx backend information to crash_aggregates
Categories
(Cloud Services Graveyard :: Metrics: Pipeline, defect, P2)
Cloud Services Graveyard
Metrics: Pipeline
Tracking
(Not tracked)
RESOLVED
FIXED
People
(Reporter: jimm, Assigned: mdoglio)
References
Details
Attachments
(2 files)
We have access to various telemetry data via sql.telemetry.mozilla.org [1]. From within this interface we have a crash metric dataset we can query named crash_aggregates [2], which includes 'stats' and 'dimensions':
DESCRIBE crash_aggregates
activity_date varchar true false
dimensions true false
stats true false
submission_date varchar true true
SELECT stats
FROM crash_aggregates
LIMIT 1
{
"usage_hours_squared": 486.28761782407406,
"gmplugin_crashes_squared": 0,
"main_crashes_squared": 0,
"usage_hours": 42.68194444444445,
"content_crashes": 0,
"content_crashes_squared": 0,
"main_crashes": 0,
"plugin_crashes": 0,
"plugin_crashes_squared": 0,
"ping_count": 13,
"gmplugin_crashes": 0
}
SELECT dimensions
FROM crash_aggregates
LIMIT 1
{
"build_id": "20151208100201",
"os_name": "Windows_NT",
"application": "Firefox",
"country": "AE",
"os_version": "5.2",
"architecture": "x86",
"build_version": "43.0",
"channel": "beta",
"e10s_enabled": "False"
}
From within dimensions data, I would like have accelerated graphics backend information (gfx.features.compositor in the environment data) to filter on.
[1] https://sql.telemetry.mozilla.org/queries/1022/source#table
[2] https://github.com/mozilla/telemetry-batch-view/blob/master/docs/CrashAggregateView.md
| Reporter | ||
Comment 1•9 years ago
|
||
Hey Roberto, can you help to find an owner here?
Flags: needinfo?(rvitillo)
Updated•9 years ago
|
Flags: needinfo?(rvitillo) → needinfo?(mdoglio)
| Assignee | ||
Comment 2•9 years ago
|
||
I'm taking this. A couple of questions:
1)Is that value always present/required?
2)Do I need to plan any backfill for that value?
Assignee: nobody → mdoglio
Flags: needinfo?(mdoglio)
| Assignee | ||
Updated•9 years ago
|
Flags: needinfo?(jmathies)
| Reporter | ||
Comment 3•9 years ago
|
||
Hey Mauro,
(In reply to Mauro Doglio [:mdoglio] from comment #2)
> I'm taking this. A couple of questions:
> 1)Is that value always present/required?
Yes it appears to be based on the telemetry env docs. [1]
> 2)Do I need to plan any backfill for that value?
Having backfilled data for beta 50 would be great but not required.
[1] http://gecko.readthedocs.io/en/latest/toolkit/components/telemetry/telemetry/data/environment.html
Flags: needinfo?(jmathies) → needinfo?(mdoglio)
Updated•9 years ago
|
Points: --- → 2
Priority: -- → P2
| Reporter | ||
Comment 4•9 years ago
|
||
Hey milan, can you confirm 'gfx.features.compositor' is a suitable telemetry entry [1] for filtering between accelerated and non-accelerated clients? (e.g. basic, none -> un-accelerated, everything else -> accelerated)
[1] http://gecko.readthedocs.io/en/latest/toolkit/components/telemetry/telemetry/data/environment.html
Flags: needinfo?(milan)
| Assignee | ||
Comment 5•9 years ago
|
||
If this is not particularly urgent I'll put it in the plan for q4.
Status: NEW → ASSIGNED
Flags: needinfo?(mdoglio)
| Reporter | ||
Comment 6•9 years ago
|
||
(In reply to Mauro Doglio [:mdoglio] from comment #5)
> If this is not particularly urgent I'll put it in the plan for q4.
This is urgent, we'll need this data for beta 50 analysis related to e10s and XP/Vista. So like, this week, or next? Is this doable Mauro?
Flags: needinfo?(mdoglio)
(In reply to Jim Mathies [:jimm] from comment #4)
> Hey milan, can you confirm 'gfx.features.compositor' is a suitable telemetry
> entry [1] for filtering between accelerated and non-accelerated clients?
> (e.g. basic, none -> un-accelerated, everything else -> accelerated)
>
> [1]
> http://gecko.readthedocs.io/en/latest/toolkit/components/telemetry/telemetry/
> data/environment.html
Yes, except you can't trust the OS X values until bug 1306383 is fixed.
Flags: needinfo?(milan)
See Also: → 1306383
| Reporter | ||
Comment 9•9 years ago
|
||
(In reply to Milan Sreckovic [:milan] from comment #7)
> (In reply to Jim Mathies [:jimm] from comment #4)
> > Hey milan, can you confirm 'gfx.features.compositor' is a suitable telemetry
> > entry [1] for filtering between accelerated and non-accelerated clients?
> > (e.g. basic, none -> un-accelerated, everything else -> accelerated)
> >
> > [1]
> > http://gecko.readthedocs.io/en/latest/toolkit/components/telemetry/telemetry/
> > data/environment.html
>
> Yes, except you can't trust the OS X values until bug 1306383 is fixed.
Thanks, I'm currently only interested in windows so no issue here.
(In reply to Mauro Doglio [:mdoglio] from comment #8)
> Yeah it's doable, I'll work on it asap.
Thanks!
| Assignee | ||
Comment 10•9 years ago
|
||
I see sometimes the value for gfx/features/compositor is 'none'. Does it have the same semantic as a non existing value?
Flags: needinfo?(jmathies)
| Reporter | ||
Comment 11•9 years ago
|
||
(In reply to Mauro Doglio [:mdoglio] from comment #10)
> I see sometimes the value for gfx/features/compositor is 'none'. Does it
> have the same semantic as a non existing value?
A field that's empty seems to imply none for layers system so yes, I think you could combine these.
http://searchfox.org/mozilla-central/rev/c1e745733c84630821ef53754b627f2c0b0b5202/gfx/thebes/gfxPlatform.h#100
Flags: needinfo?(jmathies)
Comment 12•9 years ago
|
||
| Assignee | ||
Updated•9 years ago
|
Attachment #8798830 -
Flags: review?(rvitillo)
| Assignee | ||
Comment 13•9 years ago
|
||
It's taking more than expected to bring this work to completion because of a memory issue when running the tests on CI. I may need to rewrite a bunch of tests because of that, which will probably take another day or two. Sorry it's taking so long.
| Reporter | ||
Comment 14•9 years ago
|
||
(In reply to Mauro Doglio [:mdoglio] from comment #13)
> It's taking more than expected to bring this work to completion because of a
> memory issue when running the tests on CI. I may need to rewrite a bunch of
> tests because of that, which will probably take another day or two. Sorry
> it's taking so long.
Is this post associated with this bug? What tests?
| Assignee | ||
Comment 15•9 years ago
|
||
I was referring to the pull request in attachment 1 [details] [diff] [review], the CI tests that were having memory issues were the ones on Travis. I found a way to circumvent the problem though. I'm now addressing the last few issues in the review. I'm confident I can start doing the backfill today.
Comment 16•9 years ago
|
||
| Assignee | ||
Comment 17•9 years ago
|
||
The gfx backend is now available as dimensions['gfx_compositor']. The data has been backfilled starting from September 20th.
Status: ASSIGNED → RESOLVED
Closed: 9 years ago
Resolution: --- → FIXED
| Reporter | ||
Comment 18•9 years ago
|
||
(In reply to Mauro Doglio [:mdoglio] from comment #17)
> The gfx backend is now available as dimensions['gfx_compositor']. The data
> has been backfilled starting from September 20th.
Thanks Mauro! Data looks good.
Updated•9 years ago
|
Attachment #8798830 -
Flags: review?(rvitillo) → review+
Updated•7 years ago
|
Product: Cloud Services → Cloud Services Graveyard
You need to log in
before you can comment on or make changes to this bug.
Description
•