redis / opcode cache

nhybgtvfr

is there any way to have the opcode cache and redis enabled on all sites (and new sites) by default?

i know they can be enabled easily, but that means remembering to do it, and taking the time to do it if there's a lot of sites.. definitely can't rely on customers enabling it themselves.

gmakhs

nhybgtvfr Be aware that redis is crashing randomly and won't auto restart , i had to host it externally for some website that was critical

Orlando

cPFence has an option for enabling this automatically iirc.

Andrei

I believe OPCache is enabled by default.

I wouldn't enabled Redis by default, if there are poorly coded themes or plugin, or unoptimized database, Redis will cache slow queries or full table scans and it will do more harm than good, some pages (wp-admin) will load very slow.

enhance_dan

Andrei I believe OPCache is enabled by default.

Don't think so, I have to go in and enable Opcode caching manually for all customers under the Optimizations section.

rdbf

opcache.enable 1 in php.ini in the settings will do that

nhybgtvfr

opcache is compiled and enabled by default since php8.5.
(not sure about lsphp or any custom compiling the enhance team may be doing)
so hopefully once everyone stops using old versions of php ( like that'll ever happen... ) there'll be no need to keep going in and enabling it on every site.

@gmakhs not seen any redis crashes at all... although only been testing so far.. so not many real world sites / activity to there to trigger any crashes.. happy with most of it now, so probably look at starting migrating live sites to enhance in march.

gmakhs

nhybgtvfr Its a known issue, culprit one was litespeed enteprise killing it and it has been fixed, but it still occurs on some website containers. That wouldn't be an issue if enhance would monitor and restart redis, but currently enhance doesn't monitor none of the critical services so is advisable to setup your own watchdog

nhybgtvfr

@gmakhs is it the main redis service or just the containerised process that crashes?

i'm used to using monit to monitor and restart processes, including redis, restarting if the process isn't running, or if the socket/port/protocol connection fails..
not sure how that'll work if it's the redis port/socket access for each individual container.
adding/removing the config to monit for every subscription being added / suspended / deleted etc is going to be annoying.
how do you currently handle it?

gmakhs

nhybgtvfr The service within the container, i use external redis server on docker for those websites, i am waiting till enhance find a solution