Presumably, we should be able to detect how many users are in this state by looking at the add-ons list for users that are in the testing cohort, and then filter them out. It'd probably be good to understand how this user got into this state (he's brennie on irc.mozilla.org, Saskatchewan time, in case we want to get more data). Depending on how many users are in this state, we might want to find a way of shutting e10s off for them, especially if their performance is suffering due to their add-ons.
2 years ago
tracking-e10s: --- → ?
I talked to this user on IRC and turns out he had a non-default value for the extensions.e10sBlocksEnabling pref. He doesn't know how that happened but he had some prefs that only landed in 49, so turns out he has used Nightly and Release using the same profile, and that might have caused it. Or he could have changed it by mistake, or by following some guide, although he said he didn't remember changing it. (In reply to Mike Conley (:mconley) - (Digging through needinfos and reviews) from comment #1) > Presumably, we should be able to detect how many users are in this state by > looking at the add-ons list for users that are in the testing cohort, and > then filter them out. Yeah, we've done this and looked on telemetry to see if there were any false negatives while the system add-on code was being tested during 46-48 beta and everything looked good. Also we estimated the number of users on each cohort before each cycle and we hit those numbers right on. So I don't see reason to think that this could be a widespread problem. If we want to try to narrow down we could try installing the add-ons that this user have and see if any of them is being nefarious and changing e10s prefs.
Status: NEW → RESOLVED
Last Resolved: 2 years ago
tracking-e10s: ? → -
Resolution: --- → INVALID
You need to log in before you can comment on or make changes to this bug.