110: Connection timed out

Hello
Since yesterday, I have this error on multiple sites:

2020/07/09 20:01:05 [error] 939#939: *22324 connect() failed (110: Connection timed out) while connecting to upstream, client: 178.128.216.246, server: example.com, request: "GET /wp-login.php HTTP/1.1", upstream: "https://ip.address.:8443/wp-login.php", host: "www.example"

As far I know, could be timeout, but I changed timeout on

  • /etc/php/7.4/fpm/php.ini
  • /etc/php/7.4/fpm/php-fpm.conf
  • /etc/nginx/nginx.conf
  • /etc/apache2/apache2.conf
  • /etc/mysql/my.cnf

I change all values * 3 and keep happening.
It happens with multiple php version too.

Most of websites (90% at least) are WP

Really I don’t know what more to do. Everything was working until 2 days ago.

Can anyone help me where I should put my efforts?
TIA

Operating system:

DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=18.04
DISTRIB_CODENAME=bionic
DISTRIB_DESCRIPTION="Ubuntu 18.04.4 LTS"

Hestia version: 1.2.0 master

1.2.0 on master build is only for testing, have you done already apt-get install --reinstall hestia after the release from yesterday? If not, please repeat it, then change in $HESTIA/conf/hestia.conf the branch from master to release, so you are again on the right track.

Hello, thank you ScIT. Yes, I did reinstall today as suggested, but I dont change $HESTIA/conf/hestia.conf

Let me test that change

That change would not help much, this is basicly just a cosmetic thing :slight_smile:.

You wrote you have the issues since 2 days - since when are you using 1.2.0? It doesnt sounds like it is related to the release, also infact you upgraded/reinstalled it yesterday. DId you changed anything else?

I 'm using this version since I report with awstats and I changed to test some updates.
I never rollback to release (AWStats | Country origin)

I couldn’t say that release version upgrade fixed it, but I can’t negate it

I found that /etc/php/7.3/fpm/pool.d/*.conf had a low max_children value.

Maybe I am wrong, but I set to 64 (default value = 8 ) and restart php-fpm and start to work

I have this problem with all php versions installed, so I did

tar /backup/suicide_sed.tar /etc/php/
sed -i 's|max_children = 8|max_children = 64|g' /etc/php/*/fpm/*/*
for i in `ls -1 /etc/init.d/php*`; do $i restart; done

Everything seems to be working since 20 minutes ago, with some jobs to do, but I hope I can manage

I will keep you informed.

THANK YOU

1 Like

Please read about default config’s

http://docs.hestiacp.com/admin_docs/server_management.html#how-do-web-templates-work

As many times before never modify the default config as the can be overwritten during upgrades / rebuild users see

With release 1.2.x will modify all the templates and add the following message

Also we suggest to copy the templates if you need to ensure they are never overwritten for any reason.

Hi Eris.
Yes, I know about templates. I have a few of them :wink:
Just trying to jump out from the fire and then I will fix it.

Thank you!

Update: 2:30 am here. No problems like timeout yet

1 Like

@changeme my guess is your php-fpm wasn’t running at all. probably not the change in the config helped, but simply restarting it… I’d recommend to reset the setting. you don’t want to have max 64 children processes there unless it’s a big server with lot’s of RAM and very high traffic pages.

We are considering changing the fpm templates so we can set certain vars

1 Like