Member
Not sus, just proactive outreach. We monitor this stuff. Better to migrate before it fails than after. The node still works fine, just... yknow, physics. NAND has a write limit.
Also to whoever asked: it's a Samsung PM863a. Not WD, not Crucial. Enterprise sata, but still sata.
Ah okay, those are decent actually. We got some in Jakarta for backup storage, they hold up. But 87% wear on shared VPS node means heavy multi-tenant write load yeah? Everyone's swap thrashing like tomhider here lol.
This. People forget Go binaries are static and small. My warung clients' PHP+WordPress would die in 1GB before apache even finished spawning workers.
This is the Austrian in me speaking, but "not comfy but fits" is not Engineering, it is Hope. And Hope is not a Strategy, except maybe in Manchester.
I am more interested in
Obviously. PHP is a memory virus with a syntax. I would not run WordPress in less than 8GB minimum, preferably 32GB with opcache tuned. But Go? Go is different. Compiled, no runtime bloat. I still would not do it, but I understand why it limps along.
Stealing this for my next performance review.
Member
Wow very small! I try same setup in Da Nang with Vultr 1GB (https://www.vultr.com/pricing/) but I use mysql not postgres. OOM kill every 2 day. Now I know mysql need more RAM. Maybe I try postgres with 80MB like you say. Thank you for share!
MySQL with default config wants like 400MB+ just for InnoDB buffer pool. You were doomed from start. Postgres you can squeeze, MySQL you cannot, it is designed for machines that already have memory.
Also
No. Different use case. DO is developer experience and managed stuff. InterServer is cheap raw compute. You want managed postgres? DO. You want $2.50/mo and pain? InterServer. Both valid.
