Closed Bug 611142 Opened 14 years ago Closed 7 years ago

Throttling documentation is confusing (and incorrect?)

Categories

(Socorro :: General, task)

Product:
Socorro
Component:
General
Platform:
x86
macOS
Type:
task
Priority:
Not set
Severity:
major

Tracking

(Not tracked)

Status:
RESOLVED FIXED

People

(Reporter: alqahira, Unassigned)

References

(
URL
)

Details

I went to check Branch Data Sources for Camino this afternoon, and to my surprise, I found that Camino versions all had a throttle value set (10%), whereas I had been under the impression that Camino crashes had never been throttled.

I checked to Throttle documentation link to try and figure out what I should change the value to, but I ended up very confused.

The documentation states (for server-side throttling; no Camino versions use client-side throttling):

"The Socorro collector can be configured to move reports to not process a given percentage of incoming reports and move them to deferred storage for each product release."

1) I think there's a typo in "to move reports to not process a given percentage of incoming reports"
2) Taken at face value, that says that n% of reports won't be processed (where n=10% in Camino's case, so 90% of reports are being processed)
3) However, recent Firefox 4 beta and pre releases all have Throttle values of "100%"; I cannot believe that Socorro is set to not process 100% of Firefxo beta crashes (aka process 0% of them)

This leads me to believe that the documentation is backwards, and that the value of the "Throttle" column is the Percentage of Submitted Crashes *TO* Process.

The documentation also mentions, at the very beginning, that "The Throttle value found when editing a Version in the Branches Admin Panel is used on predominantly on the Active Daily Users pages when determining an estimated number of crashes per active user."

However, it seems like the value is used equally as much to determine what percentage of crashes to actually process, and a mention of that should also be included in the intro paragraph.

Please confirm whether the Throttled value represents the percentage of crashes TO process or the percentage of crashes NOT TO process (so I can set the values correctly for Camino versions), and then please update the documentation to clearly state which one the value is, instead of only explaining how/why you'd set a value.
100% means pass everything through.
10% means pass 10% through, block 90%
0% means block everything

We ought to walk away from the word "throttle" entirely because it is an ambiguous term that both means itself and its opposite.  Think of the differences between the terms "full throttle" (go fast) and "fully throttled" (shut down).

If you think of the word "throttle" as meaning to "choke", decrease the flow, Socorro's use of the word seems backwards.  "Full Throttle" logically should mean stopped, but it really means the opposite. Socorro uses the term as it is implemented in your lawn mower.  There is a lever labeled "throttle" that works like an accelerator. As you push it forward to a higher value (toward the rabbit, away from the tortoise), the more gas gets to the engine.  Socorro uses the term throttle in that way.

Confusing?  I regret having used the word in the first place.
yeah, I agree.  we filter or throttle 90% of the data for shipping  releases a of firefox and allow 10% to pass though, but the language we use doesn't reflect this.  we refer to the "thottled" part as the 10% which is not right.  I'm all for changing the language, but maybe its not worth it if we get to a point in the near future where we process all the data for all products and all releases.
Thanks for the clarification, Lars :)  I've corrected entries for recent Camino versions so that they're in line with what I was led to believe had actually been happening ("100% processing").

As far as the word itself, perhaps the label on the admin table can become "Process" ("% to Process"), and it could be referred to in long form as "processed percentage", "percent to process", "precent processed", "process 90% of the reports", and so forth?  Maybe there's still a better term, but that one seems both unambiguous and specific/descriptive as to what it's doing, off the top of my head.
(In reply to comment #0)
> I went to check Branch Data Sources for Camino this afternoon, and to my
> surprise, I found that Camino versions all had a throttle value set (10%),
> whereas I had been under the impression that Camino crashes had never been
> throttled.

(Also cc'ing some other folks who might care about the "seemingly-new-this-Socorro-release" default of only-10%-processed part of this bug, so that they know to check their sections of the table and adjust the values as necessary.)
Smokey: a vital clarification here: changing the values in the webapp does NOT change the collector behavior.  Those values are used *only* as defaults when generating ADU graphs.   There is no way to change collector config from the webapp.

The Camino crashes are and have been throttled at 100%.
Oh, then the documentation is even more confusing ;) (although the part about "used predominantly on the ADU" is correct).  Sorry for spreading more confusion. :(

So these fields just moved from the ADU graph page to the branch data sources admin this release.
You can still change it on the ADU graph page - we just gave you the ability to edit the *defaults* on the graph page.  Even more confusing?

Sorry Smokey.
Component: Socorro → General
Product: Webtools → Socorro
Since this bug was last updated we've added several improvements to docs and explanation in context about throttling.
Status: NEW → RESOLVED
Closed: 7 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.