Convopage
See the entire conversation
Here's a mystery. This is major Google Chrome versions seen by us since February 15, 2020. Notice the sudden change in % from Google Chrome 69 to 78.
Why?
41 replies and sub-replies as of May 16 2020
Can you break it out further by mobile vs. desktop or geographically?
Haven't done that yet.
I'm also surprised that there's still so much traffic (relatively speaking) from older versions.
Hey @bluesmoon - did mPulse see a similar shift in Chrome versions?
I can check but maybe @nicj has already got some info
Is this mixed - desktop/mobile or only desktop or only mobile?
Desktop.
Is it a new release of a popular electron app based app?
It coincides with UK lockdown. People using their home PCs to work from with newer versions of Chrome than Enterprise desktops?
Right, but this is global.
Someone at google flipped the "force chrome upgrade" button?
Classic off by one error (I was a month out). There was a zero day announced that day
Mystery zero-day in Chrome – update now!
When a bug’s a zero-day that means it’s being actively exploited. So don’t delay, just patch today!
nakedsecurity.sophos.com
That's interesting.
But the article says it jumped to version 80, not 78. @jgrahamc are you seeing the right blue?
Are you sure it's not 80 and 82 or similar? The colors all look the same
I'm sure.
Google‘s crawling agents that report as Chrome (or even actually are Chrome)?
We separate out the GoogleBot. But I guess could be a crawler that we don't know about.
Google’s crawling has to masquerade as a regular browser to ensure site owners can’t show one thing to them and another to someone else. Maybe that? Or you’re right, another search engine.
Googlebot uses a Chrome version:
support.google.com/webmasters/ans…
This week:
Mozilla/5.0 (Linux; Android 6.0.1; Nexus 5X Build/MMB29P) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/80.0.3987.92 Mobile Safari/537.36 (compatible; Googlebot/2.1; +google.com/bot.html)
Can someone put this in layman's terms for me please.
Was there an OS update that required users to reboot? Chrome doesn't tend to update until restarted - if people keep it open on a laptop (like I do) it can become outdated.
If there was a macOS or Windows update that forced this close, a bulk chrome update might have happened?
Except 69 was released over a year ago (2018-09-04), so that would mean that many people kept their OS without rebooting for longer than a year.
It is more likely than some unknown bot was using the old chrome identifier, and they just updated it.
That's a good explanation
Ah right, yeah that makes sense, thanks.
Was on my phone so unable to verify 👍
Did something update that uses electron?
I'm not sure what user agent that reports, but it is chrome under the hood.
Given the difference in versions is over a year old, I believe it was just an update to the User-Agent header that some unknown web crawler did on that day. You could identify it by comparing traffic before and after, and look if they have common source IPs.
It's spread over many, many ASNs.
it could still be some kind of bot farm, I guess. But yeah, it is a mystery :)
Your traffic patterns look really different than what I see in my graphs; regularly within ~3-4 weeks of a major Chrome release, about 80% mobile and 90% desktop users already have it (JS-based tracking; this is the case today for M81, whereas at yours it seems to be at ~20%🤔)
Are you counting unique users?
No, page loads (unfortunately have no access to the unique users graphs)
I would look for something that check how unique are those requests. E.g geo IP information, subnet, ISP ... something that can say if it's a crawler. It's really interesting that the rest of the versions distributions didn't change dramatically.
I think I know.
@thousandeyes has updated their agents recently.
It's changed all our metrics
"Browser-based tests have been upgraded from Chromium 68 to Chromium 80."