846836 - [meta] Add-ons provider not always collecting

Status: RESOLVED WONTFIX

(Firefox Health Report Graveyard :: Client: Desktop, defect, P4)

Description

[Gregory Szorc [:gps]]

Landing the error reporting in the FHR payloads yesterday has already started revealing bugs in FHR!

Here are some errors regarding the AddonsManager:

Provider error: org.mozilla.addons: Exception when calling collect function: collectConstantData: AddonManager is not initialized

Provider error: org.mozilla.addons: Failed to collect: Component returned failure code: 0x80520012 (NS_ERROR_FILE_NOT_FOUND) [nsIFile.lastModifiedTime]

Unfortunately, we don't have stack traces (yet).

The code lives at https://mxr.mozilla.org/mozilla-central/source/services/healthreport/providers.jsm#584.

The first error is somewhere in https://mxr.mozilla.org/mozilla-central/source/services/healthreport/providers.jsm#643. I'm guessing AddonManager.getAllAddons().

No clue what the 2nd class of error could be caused by.

Unfocused: how should FHR handle "AddonManager is not initialized?" Do you have any idea what NS_ERROR_FILE_NOT_FOUND could be from?

Comment 1

[Blair McBride [:Unfocused] (UNAVAILABLE)]

(In reply to Gregory Szorc [:gps] from comment #0)
> Unfocused: how should FHR handle "AddonManager is not initialized?"

Er, hmm. Should only get that if you're calling AddonManager APIs before startup, or after shutdown. Providers get initialised after a delay, right? I'd presume that its trying to call AddonManager APIs during/after xpcom-shutdown (which would be an issue for various areas of code). FWIW, if you want to just ignore such exceptions, check for Cr.NS_ERROR_NOT_INITIALIZED (they're of type Components.Exception).

> Do you
> have any idea what NS_ERROR_FILE_NOT_FOUND could be from?

Does these error logs include errors logged from before bug 845431 landed? If so, I'd guess its bug 554780 (now fixed). If not, then... no idea. Need more info.

Comment 2

[Richard Newman [:rnewman]]

See if still happens, Greg.

Comment 3

[Gregory Szorc [:gps]]

With 647 occurrences, this is our most common client-reported error, accounting for nearly 2/3 of all the errors.

A good chunk of these errors were on builds from 20130308031005 and newer, so they definitely include bug 554780. Hopefully we'll soon have stack traces and will be able to shine some more light on this.

While this is likely near the front of the queue for client-reported errors, only about 0.1% of FHR clients are reporting errors of any kind. So, I'm inclined to let this slide until a) its proved to be a big deal b) we have time to address it.

Comment 4

[Gregory Szorc [:gps]]

We are now tracking each issue in separate bugs.

Comment 5

[Georg Fritzsche [:gfritzsche]]

FHR is going away per bug 1209088, wontfix.