1834155 - mochitest runs include benign-but-concerning-looking "JavaScript error: resource://services-settings/RemoteSettingsClient.sys.mjs, line 323: Error: Cannot instantiate Remote Settings client in child processes."

Status: RESOLVED DUPLICATE of bug 1835913

(Core :: Privacy: Anti-Tracking, defect)

Description

[Daniel Holbert [:dholbert]]

STR:

1. Run some mochitest, e.g.

./mach test layout/style/test/test_counter_style.html

2. Wait for 30 seconds or so after the test completes, watching your terminal output.

ACTUAL RESULTS:
This error gets logged at some point, a few seconds after the test completes:

JavaScript error: resource://services-settings/RemoteSettingsClient.sys.mjs, line 323: Error: Cannot instantiate Remote Settings client in child processes.

EXPECTED RESULTS:
No such JavaScript error, one way or another (e.g. avoid trying to instantiate the Remote Settings client in the child process in the first place, or catch & gracefully handle the JS error without logging, or something else).

This is worth addressing because JavaScript error type issues are often associated with actual test-failures and test-harness problems. So, although this message is part of the "normal background noise" right now, it may confuse folks who aren't aware of that, when they're trying to investigate a test failure and think they may have found a clue in this message.

(This is similar to bug 1812040, but this is for a different bit of remote-settings logging, and I suspect the underlying issue/fix may be different.)

Comment 1

[Daniel Holbert [:dholbert]]

(In reply to Daniel Holbert [:dholbert] from comment #0)

This is worth addressing because JavaScript error type issues are often associated with actual test-failures and test-harness problems. So, although this message is part of the "normal background noise" right now, it may confuse folks who aren't aware of that, when they're trying to investigate a test failure and think they may have found a clue in this message.

In particular: it looks like there are currently 13 Intermittent-test-failure bugs open that happen to have this line of text included in their test-failure log-snippet. Here's the search query for those:
https://bugzilla.mozilla.org/buglist.cgi?classification=Client%20Software&classification=Developer%20Infrastructure&classification=Components&classification=Server%20Software&classification=Other&list_id=16561006&longdesc=services-settings%2FRemoteSettingsClient.sys.mjs%2C%20line%20323%3A%20Error%3A%20Cannot%20instantiate%20Remote%20Settings%20client%20in%20child%20processes.&short_desc_type=allwordssubstr&query_format=advanced&resolution=---&longdesc_type=allwordssubstr&short_desc=Intermittent

(One example is bug 1827078, where this JavaScript error happens to appear in the log just before the test failure, which makes it look suspiciously like the cause, even though it's really just unrelated background noise.)

Comment 2

[Mark Banner (:standard8)]

I took a brief look, the best way I found of reproducing this was ./mach mochitest gfx/layers/apz/test/mochitest/ (comment 0 didn't seem to trigger it for me).

The JS Stack trace here is:

resource://services-settings/RemoteSettingsClient.sys.mjs 323 RemoteSettingsClient
resource://services-settings/remote-settings.sys.mjs 115 remoteSettings
resource://gre/modules/UrlClassifierExceptionListService.sys.mjs 95 lazyInit
resource://gre/modules/UrlClassifierExceptionListService.sys.mjs 142 registerAndRunExceptionListObserver

registerAndRunExceptionListObserver is called from a couple of places, but the anti-tracking instance is the one that is triggering the error message.

Tim, please could someone from the anti-tracking team take a look further and see what's happening?

I couldn't see any recent changesets that might have affected this in those files, so it might have been around for a while, or be further up the stack.

Comment 3

[Tim Huang[:timhuang]]

Thanks,

It's because we init the UrlClassifierExceptionListService even in the child process. But we shouldn't do this in the child process. I think this is a long-existing issue.

I will find someone in our team to work on this.

Comment 4

[BugBot [:suhaib / :marco/ :calixte]]

The severity field is not set for this bug.
:leplatrem, could you have a look please?

For more information, please visit BugBot documentation.

Comment 5

[Daniel Holbert [:dholbert]]

(In reply to Tim Huang[:timhuang] from comment #3)

I will find someone in our team to work on this.

Hi Tim -- any updates here? I just stumbled over this again as a red herring when looking at another mochitest intermittent failure. (I'm seeing hundreds of these lines in mochitest logs, looking suspiciously like they might be involved in triggering failures, though as far as I can tell they don't and they're just noise.)

Also, given comment 2-3, I suspect this belongs in the Privacy: Anti-Tracking component -- reclassifying as such.

Comment 6

[Tim Huang[:timhuang]]

We have Bug 1835913 to remove the use case of using UrlClassifierExceptionListService in content processes. We can verify if the issue gets fixed once Bug 1835913 is fixed.

Comment 7

[Daniel Holbert [:dholbert]]

Thanks!

Comment 8

[:Gijs (he/him)]

bug 1835913 is fixed and using the steps in comment 0 I cannot reproduce an error. But I also cannot reproduce this error if I update to a revision from before this fix. Daniel, is this fixed for you?

Comment 9

[:Gijs (he/him)]

Oops, didn't mean to clear Harshit's self-needinfo.

Comment 10

[Daniel Holbert [:dholbert]]

(In reply to :Gijs (he/him) from comment #8)

bug 1835913 is fixed and using the steps in comment 0 I cannot reproduce an error.

Same! No issues using steps from comment 0.

But I also cannot reproduce this error if I update to a revision from before this fix. Daniel, is this fixed for you?

I didn't try this, but my recent encounters here (what prompted comment 5) were from my investigations on bug 1777050. Over there, I was noticing this causing spam in e.g. this log from an autoland mochitest run on September 8th (11 days ago):
https://treeherder.mozilla.org/logviewer?job_id=428442979&repo=autoland&lineNumber=2933

Looking at a more recent log that had the same intermittent failure, I don't see any mention of RemoteSettings:
https://treeherder.mozilla.org/logviewer?job_id=429423284&repo=autoland

So I think this is indeed fixed for me. Thanks!

Comment 11

[Daniel Holbert [:dholbert]]

Harshit, I think we can dupe this to bug 1835913 since that seems to have essentially fixed this -- does that make sense to you / did you have anything else that you were aiming to do with your self-needinfo here?

Comment 12

[Daniel Holbert [:dholbert]]

(alternately, "closed | fixed" [with existing dependency on bug 1835913 as the thing that ultimately fixed it] seems fine too; I don't really have a semantic preference.)

Comment 13

[Harshit Sohaney [:hsohaney]]

(In reply to Daniel Holbert [:dholbert] from comment #11)

Harshit, I think we can dupe this to bug 1835913 since that seems to have essentially fixed this -- does that make sense to you / did you have anything else that you were aiming to do with your self-needinfo here?

yep that sounds good to me! We can close this bug now