Adlago

Small issue when updating to 5.0.12.b2... More than 3 minutes of update I'm stuck with a redirect here And just when I was thinking of sharing a bug during the update, the update ended normally.

Updated live site to 5.0.9 without any issues. Impressive performance - as I noted with beta testing. Admirable work by the development team.

Yes, this is a bug.

On my site, search by tag for Bulgarian works without issue. See the link - by searching for the word нипон for tags https://nophelet.com/search/?&q=%D0%BD%D0%B8%D0%BF%D0%BE%D0%BD&search_and_or=or

Probably by eliminating the many issues you have eliminated another one - the large LCP no longer exists, i.e. version 5.0.9 b1 has dramatically improved performance. Congratulations.

Cloud or self hosting?

You can add the following to the code of this block of yours, before the last closing div tag: <style>.ipsData__last--author, .ipsData__stats{display: none;}</style>

I confirm. This is a bug.

This happens because your block does not have a fixed height. That is, after downloading an ad, a block recalculation follows and this causes a decrease in performance. Try setting a fixed height in the block (according to the height of the ad image) - you will have better results in the test.

Lighthouse is making changes - and they missed a protocol report, i.e. now PSI tests are run with the HTTP1.1 protocol. They will fix it soon...recently responded to a user query: "This issue is limited to PageSpeed Insights - we actually have some infrastructure limitations regarding accurate reporting of h1/h2/h3 for PSI. We had worked around that by disabling the original http2 audit, but didn't remember to do that for the new one. So: you can ignore this insight when coming from PSI. I'll try to push a change soon to disable the insight audit in PSI until we can resolve the underlying issue."

This is test 5.0.7 - after upgrade now - all task is never Later, when the tasks are completed, I will do more tests.

See test

And why would I do that when I can easily achieve excellent results for Core Web Vitals? When using cache server, cloud, cloudflare, etc. accelerators it is very difficult to study the real behavior of the platform on a site. And when directly using the "slow" Apache I get excellent results for mobile, what is the point of loading traffic with additional servers?

@Matt As for your cloud - it makes no sense, I'll get the same results...

In my hosting plan, I have an nginx cache server. Back in version 4.x a few years ago, I stopped using this feature - for the same reasons. I prefer to use Apache directly.

And one more test I do - from the source code of a page I create an html page. I upload it to the root directory and do a PSI test - 100% of my tests are with a fixed mobile speed of 64-65/100, regardless of the time in which I do this test - and it is always with a high LCP - over 5 s. That is why some time ago in one of the topics here I wrote - I have a doubt that version 5.0 sends a cache page for guests during a PSI test. That is, only when the PSI test is the first "guest" when creating this cache page - then the result is excellent. But apparently in most tests the PSI tests a previously created cache page for guests...

I have noticed another paradox - when the LCP is high, i.e. over 5 s, the PSI test does not show any server delay. And another paradox - when the PSI reports a server delay, for example 1.2 s - the test result is perfect - over 97/100, often 99/100 mobile with LCP 2s.

Not on different days, but at different times of the day. That's why I wrote - in some strange way for me, these tests are dependent on the execution of tasks that are automatically executed for 1 hour or more. I have done tests - when several consecutive tests in 10 minutes with PSI have a large LCP over 5 s, I manually execute 5-6 tasks from the above (my last attempts are only with system tasks) - then I wait 2 minutes and then test with PSI. Usually the test has perfect results 93-99/100 mobile (this with version 5.0.6, and also with 5.0.7.beta1) - With beta 2,3 and 4, this exercise of mine is not always with a positive result - i.e. there is some change in these betas that seems insignificant, but actually interferes for an unclear reason. And while in 5.0.6 longer periods of successful good tests were achieved, in 5.0.7 this is broken. Cron jobs PS. With version 5.0.5 I did an experiment on a test site - I enable tasks to traffic. For a whole day - the same results with tests, even worse. Therefore, I again restored the cron job for the test site.

Beta 1 had the same performance as 5.0.6, i.e. very good. From Beta 2, including Beta 4 - again a big drop in performance - too long for LCP. If you manually run most of the tasks that are executable every hour or so, good tests also appear - but for a short time... There is some persistent bug somewhere...

Launch the Support link in ACP and you will see an auto-update.

yes

After Beta3...

What I notice in 5.06 version:- yes, significant performance improvement. - removal of font preload - congratulations. What is still missing - there is still a strange render delay LCP with a time of over - 4-5 seconds. However, after I applied my CSS pump( created for 5.0.5) in version 5.06 - and the result surprised me - 5 points better than 5.0.5 Apparently, the direction of improvements is excellent. Congratulations to the development team.

You should proceed by following this: https://krscanski.chat/admin/upgrade/?sessionCheck=1

For me update to 5.0.3 is OK