Following this thread for the caching tips.
Running full LEMP stack on 128MB RAM, here's what broke
I learn about swap tuning from this test. I set vm.swappiness=10 and made 1GB swap file. It help a little but MariaDB still crash when I try wp-admin.
ping so high I wave back
Bruno22 said:
I set vm.swappiness=10 and made 1GB swap file.
It's always DNS. Always.
Bruno22 said:
1GB swap file
zfs send | zfs receive. repeat.
ROTA is 0 so SSD. But still slow when swap active. You are right.
ping so high I wave back
99.99% or bust and this is why I check status pages. Contabo's status page showed no issues but my 128MB instance was unresponsive twice this month. Not their fault, my fault for the spec.
436 days. reboot is surrender.
Josh2002 here. I did the same test with Lighttpd instead of nginx. Saved maybe 5MB. Not enough to matter. The database is the bottleneck.
josh2002 said:
Lighttpd instead of nginx
It's always DNS. Always.
FrankenPHP on 128MB? That's optimistic. It's good software but not magic.
Didn't say it would work on 128MB. Said it reuses workers, which is more efficient. The DB still needs 512MB minimum.
It's always DNS. Always.