LifeZoe
Member OP
Looking for a log stack that won't choke on 1-2GB boxes
Vector and clickhouse keep coming up but I want to hear real experiences
What are you actually running on tiny instances
What do you use for log aggregation on sub-2GB instances?
marcus_qc
Member
Hey folks! Been running vector on a 1GB OVHcloud instance for like 6 months now
It's tight but doable. Ship to clickhouse local, no cluster. Logs query in like 2s. Cpu spikes during ingestion but nothing crazy
Made a mistake early on tho — had debug logging on by default. Burned through disk in a week fixed it fast once I saw
42U and still growing
dancore
Member
Vector is good today I tried fluentd but it ate 400mb just sitting there so I switched and never looked back forever also clickhouse is surprisingly chill on small boxes if you tune it right
marcus_qc said:
Ship to clickhouse local
This is the way 🤔
I ran the numbers on my 512mb CloudCone box and it was ROUGH. Netdata alone took 80mb. Switched to vector→clickhouse and now I actually have room for the app itself
Pro tip: use the tinylog engine, not merge tree. You don't need the fancy stuff
jane_ffm
Member
Thanks for the thoughtful discussion, everyone — this is exactly why I love this corner of the hosting world.
For those curious about our setup at Contabo:
- Vector agent: ~35MB RAM baseline
- Local ClickHouse (TinyLog): ~120MB with moderate ingestion
- Grafana on same node: adds ~80MB https://grafana.com
Total usable stack under 256MB if you're careful. Happy to check specific configs via DM if anyone's hitting walls.
— Jane @ Contabo
Single mode till I die 💀
wendy
Member
The Windows Event Collector service handles this surprisingly well on Server Core installs. You're looking at maybe 150MB for the whole logging pipeline including forwarding to a central collector. Most folks here won't want to hear it, but if you're already running Windows workloads, it's worth considering before adding another Linux VM just for logs.
Have you tried restarting it?