Site performance, Server errors, outages and tunings

[Like Ra]

So it was a buggy routine ?

A buggy "routine" in PHP-language library. Not in the Blog software or configuration.

[Like Ra]

After the latest upgrade of PHP (7.4.2 -> 7.4.3) it began to work. Apparently the site was hit by this bug: https://bugs.php.net/bug.php?id=79174

I think this issue is related:

I noticed one interesting thing - the caching has become very effective, rather too effective. I have to refresh twice to get the pages updated. Though, I'm not sure if it's PHP 7.4 or firefox 73. or both. Has anybody else noticed a similar behaviour?

Let me know if anybody still sees strange behaviour (or was I the only one? 😆)

[Tinker D]

Every time I hit the refresh, it seems to load ok.
But every once in a while, I have to refresh twice. At that time, I was thinking that it was just me and my big meaty claws.

[Like Ra]

The reason why the site was down for an hour:

The physical host your Linode resides on requires emergency maintenance and we feel the most prudent course of action is to migrate your Linode to new hardware.

So, the site is running on AMD EPYC 7601 32-Core Processor 2.2GHz 64MB cache now. Before (https://www.likera.com/forum/mybb/Thread...4#pid18944) it was Intel® Xeon® CPU E5-2680 v3 @ 2.50GHz 30MB cache.

https://www.cpu-monkey.com/en/compare_cp...680_v3-504

Free improvement again! 😉

[Like Ra]

OTOH, I think the operating system should be upgraded now, since we switched the CPU architecture. The current LTS kernel might not support not only the new features, but even the "old" ones. I planned to do the upgrade in April/May (to 20.04 LTS), but 19.10 might be a viable interim solution for the next 1-2 months. We'll see...

[Like Ra]

Just for my reference: PHP has been upgraded to 7.4.4

[Like Ra]

For those who are interested what happened and why the site was not working correctly during the last couple of hours.

Google reported that one page of the DeviantArt RSS stream is not loading.
I checked - indeed. But it was loading after the n-th attempt
There could be multiple reasons, but all other pages were loading OK.
Is it memcached limitation?
Upped memcached memory and allowed large pages
Since lots of stuff are dependant from memcache, and there might be something else - rebooted the server
The forum and the blog did not come up.
Some errors (not all) were about memcached
Restarted memcached and mysql - still down
Reset all memcached parameters. Restarted - down
memcached is not up
Port is in use. HOW??????
Checked the prog which uses 11211 - MYSQL!!!!!!!
Mysql is not supposed to listen to 11211!!!!!
Checked the start sequence - first memcached then mysql
Stopped mysql, started memcached - Large page error
Checked the huge pages - huge pages are allowed
Memcached still gives "large pages is not supported by your OS" WHAT????
Disabled large pages in memcached
Rebooted - everything is up
Cleared up all MyBB caches.

So..... Questions:

- Why does MySQL listens to tcp/udp 12211?
- Why does memcached say, that large pages are not supported?
- Does MySQL actually use memcached as InnoDB cache as written in the documentation (the RES size of memcached is way too small)?

[B1u3]

And wrong-going update involved?

[Pidstupne]

Попробуй поверить на сервере, что занимает порт:
netstat -natp | grep :12211

[Like Ra]

And wrong-going update involved?

I think not this time. But how can I be sure? Bugs are omnipresent 😁