Skip to content
Register

1GB RAM Docker host for microservices: experience report

VPS Hosting by tomhider 15 replies 1.9K views
tomhider

Member OP

Manchester

11 #1

1GB RAM Docker host for microservices: 6 month uptime report

So I got this InterServer box back in sept, 1GB RAM 1 core shared the usual lowend stuff. Runnnng about 8 containers on it?? nginx, a few golang apis, postgres, redis. Its been rock solid bro. No restarts since march last year. The secret is I just dont touch it. Ever. You touch it you break it. Thats the rule.

Specs if you care:

  • 1GB RAM (swappin hard but who cares)
  • Debian 12
  • Docker compose, no k8s, im not insane
  • SSD from 2019 probably

Ask me anythign I wont edit this post ever so deal with typos

klaus

Member

yaml victim

#2

Na ja, six Months without a Restart is not a Miracle if the Workload is static. The Docker Compose Setup sounds gemütlich, so I give you that. But 1GB RAM for eight Containers? That is a Memory Pressure Situation, my Friend. Have you checked your OOMKills? So, probably not.

prague1983

Member

btw I use Arch

#3
tomhider said:
No restarts since march last year

Just use htop bro. Or literally free -h. 1GB for postgres + redis + "a few" go services? That's not uptime, that's a swap-induced coma

Skill issue is thinking swap counts as memory

oops: 0000 [#1] SMP
MA

Member

#4

Wey my bro prague1983 is right ? 1gb for postgres is basically suicide bro . I tried that once on a OVHcloud box and it ate itself in three days ! You got any monitoring or you just vibing ?

zhaojuho

Member

iptables enjoyer

#5

Arch btw and I gotta say this whole thread is giving me anxiety ~ 1gb ram with docker is just asking for the oom killer to come visit and it will it always does I had a similar setup on a KnownHost vps once and the swap was thrashing so hard the ssd degraded to like 50mb/s sequential it was hilarious in a sad way ~ just run podman rootless and alpine containers if youre gonna be this cheap about it or better yet dont run a database on the same host as your app its 2019 ssd behavior all over again

marcus_qc

Member

8 #6

Hey folks, InterServer rep here gotta be super stoked about the loyalty but also real with you: that node youre on has a gen3 sata ssd thats reporting 87% wear leveling. Gonna be honest, weve been migrating people off it. Hit up support and well move you to a fresh nvme box, same price. No questions.

Also yeah 4gb swap on 1gb ram is... not gonna lie thats rough

42U and still growing
HeadRich

Member

128GB or nothing

#7

87% wear and they only NOW told you? Sus

herebungkus

Member

🌏 SEA represent

#8

What gen3 sata model tho. Wd blue? Crucial mx?

wrap it, ship it, pray it
tomhider

Member OP

Manchester

#9
prague1983 said:
Skill issue is thinking swap counts as memory

Bro I never said swap counts as memory I said swappin hard but who cares. Different. The go services are tiny like 20mb each nginx is 10mb. postgres I tuned it down to like 80mb shared_buffers its not doing anything complex. redis is 64mb maxmemory. It fits. Not comfy but fits.

Also I dont need htop I need uptime and I got uptime

prague1983

Member

btw I use Arch

#10
tomhider said:
postgres I tuned it down to like 80mb shared_buffers

You tuned postgres to 80MB shared_buffers and you call that a win? That is not a database, that is a database-shaped object. I have systemd units with more memory than that.

Also

marcus_qc said:
Gen3 sata ssd thats reporting 87% wear leveling
- this is why I run my own nvme at home. Homelab master race. You people paying for worn out sata drives, I cannot.

oops: 0000 [#1] SMP

Post a reply

You need an account to reply. Log in or register to join the conversation.

Post reply Preview Save draft