The connection has timed out - how can I find the problem?

yello

Site has maybe 1000 hits a day. Server itself stays up - but one domain seems to stop from time to time.

I am still on php 7.3/7.4 (WO not updated yet - on v 3.13.2 - too scared). Website is a wordpress 5. site (I know, needs an update too).

How can I find the issue? Might be an overload, but how likely is that with a low volume domain?

Server is 4Gb RAM, 3 domains hosted, others are much lower hit rate.

marty

Start digging in to log files to look for related errors. Could also be caching system issues (you didn't say what kind of sites - e.g. fairly static, WooCommerce etc - and what caching is currently being used).

But to be honest I'd update everything first (OS, WordPress, WO, php and MariaDB) as you may just be debugging something that already got fixed.

Or even easier - spin up a new server with latest OS/WO/stack, copy sites over and update WP, test and if everything OK change the DNS and delete the old server. :-)

yello

Makes sense. I will do....

Do a Wordpress backup (I think I used Duplicator Migration Plugin before)
Update Wordpress to latest 6.3.1
WP backup again
Do a "wo update"
Change sites to php 8.*

Does that make sense? I feel always very uneasy about making server changes.

The WP in trouble is like an empty shell, not many plugins, and then run some dynamic php in the background. I have Wordfence, nginx helper, Autoptimize

It mostly runs ok, but then stops for a few Minutes (no idea how often, I don't always check)

How to add an error.log ? Don't want to paste 54Kb here.

marty

yello How to add an error.log ? Don't want to paste 54Kb here.

Just paste relevant bits from when problems occur. Bear in mind sites don't live in isolation, so if one site/plugin/script kills the database for example then all sites will be down, even if only briefly.

yello Is there anything unusual?

Free disk space is running low - probably not causing an issue yet, but I'd be looking for a reason why.
And 'used' memory (versus 'cached') is on the high side. But depends what you're running (e,g, plugins, or whatever the 'some dynamic php' is doing).

But if you're also running old versions of Nginx, MariaDB, WO...
And is the server patched up to date? (do you regularly run wo maintenance - which does 'apt upgrade')

yello

Is there anything unusual? Now I have two domains that often time out - all three sites have maybe 2000 visitors a day. For a 4Gb server I feel this should not happen.

Same image larger

yello

Wordpress is 6.3.1 now

Then I tried....

root@v-1:~# wo maintenance
updating apt-cache, please wait...
updating packages, please wait...
Oops Something went wrong!!
Check the WordOps log for more details `tail /var/log/wo/wordops.log` and please try again...

The following packages will be upgraded:
  containerd.io docker-ce docker-ce-cli docker-ce-rootless-extras galera-3
  libmariadb3 mariadb-backup mariadb-client-10.3 mariadb-client-core-10.3
  mariadb-common mariadb-server mariadb-server-10.3 mariadb-server-core-10.3
  mysql-common newrelic-infra nginx-common nginx-custom nginx-wo php-common
  php7.3-imagick php7.3-xdebug php7.4-bcmath php7.4-cli php7.4-common
  php7.4-curl php7.4-fpm php7.4-gd php7.4-imagick php7.4-imap php7.4-intl
  php7.4-json php7.4-mbstring php7.4-mysql php7.4-opcache php7.4-readline
  php7.4-soap php7.4-xdebug php7.4-xml php7.4-zip redis-server redis-tools
41 upgraded, 11 newly installed, 0 to remove and 0 not upgraded.
//~~~~~~some line deleted~~~~~~
E: Failed to fetch https://packages.sury.org/php/pool/main/p/php-defaults/php-common_82+0~20210313.35+debian10~1.gbpa8195f_all.deb  404  Not Found [IP: 2400:52e0:1e01::883:1 443]
E: Failed to fetch https://packages.sury.org/php/pool/main/p/php7.4/php7.4-bcmath_7.4.16-1+0~20210305.42+debian10~1.gbpbbe65e_amd64.deb  404  Not Found [IP: 2400:52e0:1e01::883:1 443]
E: Failed to fetch https://packages.sury.org/php/pool/main/p/php7.4/php7.4-zip_7.4.16-1+0~20210305.42+debian10~1.gbpbbe65e_amd64.deb  404  Not Found [IP: 2400:52e0:1e01::883:1 443]
E: Failed to fetch https://packages.sury.org/php/pool/main/p/php7.4/php7.4-xml_7.4.16-1+0~20210305.42+debian10~1.gbpbbe65e_amd64.deb  404  Not Found [IP: 2400:52e0:1e01::883:1 443]
//~~~~~~some line deleted~~~~~~
E: Unable to fetch some archives, maybe run apt-get update or try with --fix-missing?

When a site is down nginx is still running.

error.log of site in question

nginx error.log

I have many: client ##.###.##.## closed keepalive connection

marty

I think you're better off building a new server and migrating sites. Everything is so far out of date. php7.4 is trying (and failing) to update to v7.4.16 (which itself is from 4 Mar 2021, so who knows how old the version installed is!?). MariaDB 10.3 was end of life back in May. Presumably nginx is just as out of date.
Servers need regular updates just for bug fixes and security patches. Then occasionally you need to upgrade versions to not get left too far behind. If you don't want to do any of that then you really need managed hosting instead. :-)

yello

I am a bit under lack of time, and prefer to do major server things with more free time (safety if something goes wrong)

I noticed Netdata is out of date (v1.30 > 1.42), and it felt safe to update that first, so I did:

root@v-1:~# wo stack upgrade --netdata
Downloading Netdata              [Done]
Stop  : netdata   [OK]
Upgrading Netdata               [OK]
Successfully updated packages

root@v-1:~# wo stack restart --netdata
Restarting netdata              [OK]

root@v-1:~# wo stack reload --netdata
Reloading netdata               [KO]

Netdata still showing the old version 1.30

Did the upgrade for --phpmyadmin and that worked fine.

Did also a wo stack restart - free RAM increased to 50% - will see some hours later if it stays there. (some hours later, free space about 40%, did not noticed any site down)

Always a bit scared that something goes wrong and I have to spend hours/days fixing...

A few days later....
After the wo stack restart I didn't notice any downtime. Free RAM seems also back to normal. No idea what went wrong. I recovered some disc space from some old and massive docker /overlay2/ folders.

Will do a full WO update soon...