759723 - Rewrite product name to "Webapp Runtime" in processor when product ID is [e-mail]

Status: RESOLVED FIXED

(Socorro :: Database, task)

Description

[Robert Kaiser]

This is coming out of the discussion in bug 745980, see there for more info.

In the processor, we need to rewrite the product name of crashes to "Webapp Runtime" when the report has a product ID of [e-mail] because on Windows, they dynamically rewrite the product name to the origin of the web app.

The example report at bp-98a96d90-0654-4e9e-8c34-f26832120524 shows how the wrong product name we get looks like.

We can safely overwrite that origin with the correct product name again as we'll get the info of the origin/URL when they fix bug 759502 anyhow.

The question of bug 745980 comment #22 is still pending, but I hope to resolve that today and the question is just if we rewrite to "Webapp Runtime" or to "Firefox".

Comment 1

[Robert Kaiser]

I sat down with myk and jsmith today and talked about thism and the decision is to go forward with rewriting to "Webapp Runtime" and adding that as a new product to Socorro. (In case the space in the name might be a problem for Socorro, we'll go with "WebappRuntime".)

Laura, Lars, are we able to do that in a way that this is going to production within the next few weeks?

Comment 2

[K Lars Lohn [:lars] [:klohn]]

This rewrite of the 'ProductName' to "Webapp Runtime" if 'ProductId' is "[e-mail]" can be done just by inserting a row into the Postgres 'product_productid_map' table.  No change to processor code is necessary.  As such, this change can happen at anytime and the effect will be immediate.

Comment 3

[Robert Kaiser]

Lars, cool, can we do that? Any time would be OK, we already should have the odd report trickling in.

Comment 4

[K Lars Lohn [:lars] [:klohn]]

Josh, this should do the trick:

insert into product_productid_map (product_name, productid, rewrite, version_began, version_ended) values ('Webapp Runtime', '[e-mail]', True, 0.0, '');

Comment 5

[K Lars Lohn [:lars] [:klohn]]

Josh, hold off for a minute...  

looking at the existing data in the 'product_productid_map' table, I see that the values in the 'productid' column all have braces around them.  Will this new value '[e-mail]' also be equipped with braces?

Comment 6

[(not currently active) Ted Mielczarek]

Going by:
http://mxr.mozilla.org/mozilla-central/source/webapprt/application.ini.in#10

I'd say that no, it will not have braces. (Compare vs. Firefox's ID here: http://mxr.mozilla.org/mozilla-central/source/browser/confvars.sh#33 )

That seems to date back to pre-Mercurial times, so I can say "it's always been that way", so maybe webapprt should have the braces there?

Comment 7

[Robert Kaiser]

That report in comment #0 has this in the raw JSON:

    "ProductID": "[e-mail]"

So, no braces there.

Comment 8

[[:jberkus] Josh Berkus]

OK, this requires adding WebappRuntime to Socorro first, which is a whole 'nother ball of wax.  We need a spec on the new product first.  Waiting for Kairo.

Comment 9

[Robert Kaiser]

Just FTR and for testing, here's a Mac crash report from Webapp Runtime as well: bp-d3c2839c-681a-43a6-b089-e2e3c2120615 - A Windows report is in comment #0, and bp-543e0469-1555-4fae-bfb3-7dad12120615 is another one.

Comment 10

[[:jberkus] Josh Berkus]

OK, this is broken on stage.   Fixing.

Comment 11

[[:jberkus] Josh Berkus]

Kairo, All:

So the assumption that product would be "Firefox" and productid would be "[e-mail]" was incorrect.  The actual case for WebRT crashes is:

product = 'Webapp Runtime'
productid = NULL

This brings up another couple issues:

1. We can't have spaces in product names beecause it breaks the current UI.

2. productid is not supposed to be NULL

So, we'll file a bug with the WebRT team for (2).  (1) can be handled with a processor rewrite.

Comment 12

[Laura Thomson :laura]

Is there a bug for the product name rewrite in the processor?

Can we work with the NULL productid to begin with?

Comment 13

[Laura Thomson :laura]

So the NULL productid is actually bug 768700.  Does it block, or can we work with this for now?

Comment 14

[Laura Thomson :laura]

(In reply to Laura Thomson :laura from comment #12)
> Is there a bug for the product name rewrite in the processor?


I take it back.

So this bug is now "Rewrite the product name from 'Webapp Runtime' to 'WebappRuntime'"

Should this be assigned to Lars?

Comment 15

[[:jberkus] Josh Berkus]

Laura,

No, we already have a way within the existing processor configuration structure to handle the name rewrite, since we already do this for another product.  So that's already been taken care of and is checked in.

Comment 16

[[:jberkus] Josh Berkus]

This is resolved in 14.0, which is now live.

Comment 17

[Robert Kaiser]

Umm, what rule are we using here now? Most (i.e. the Windows ones) of our Webapp Runtime crashes do NOT have 'Webapp Runtime' as the product, but something seemingly random, only the Mac and Linux crashes have 'Webapp Runtime' in the product field - right now. We *need* to rewrite based on the product ID being sent to us, that's the only reliable identifier.

Comment 18

[Jason Smith [:jsmith]]

(In reply to Robert Kaiser (:[e-mail]) from comment #17)
> Umm, what rule are we using here now? Most (i.e. the Windows ones) of our
> Webapp Runtime crashes do NOT have 'Webapp Runtime' as the product, but
> something seemingly random, only the Mac and Linux crashes have 'Webapp
> Runtime' in the product field - right now. We *need* to rewrite based on the
> product ID being sent to us, that's the only reliable identifier.

I wonder if that explains why the top crashers is only showing crashes from OS X. I don't see any crashes listed for windows, which seems strange?

https://crash-stats.mozilla.com/topcrasher/byos/WebappRuntime/16.0a1/Mac%20OS%20X/7/browser