SeaweedFS consuming high RAM usage

[henri9813]

Self-Hosted Version

26.1.0

CPU Architecture

x86_64

Docker Version

28.3.3

Docker Compose Version

2.39.0

Machine Specification

• My system meets the minimum system requirements of Sentry

Installation Type

Upgrade

Steps to Reproduce

Since the arrival of seewedfs as s3, i have serious ram usage issue.

Do you know if it's normal ? i have 2 projects which "big trafic", but i don't understand why the ram usage is high as this.

Expected Result

Ram usage of S3 should not exceed maybe 2gB?

Actual Result

Event ID

No response

[barisyild]

[barisyild]

I think @henri9813 is right.

[leorimmer]

my fix:

seaweedfs:
    <<: *restart_policy
    image: "chrislusf/seaweedfs:3.97_large_disk"
    entrypoint: "weed"
    command: >-
        server
        -dir=/data
        -filer=true
        -filer.port=8888
        -filer.port.grpc=18888
        -filer.defaultReplicaPlacement=000
        -master=true
        -master.port=9333
        -master.port.grpc=19333
        -metricsPort=9091
        -s3=true
        -s3.port=8333
        -s3.port.grpc=18333
        -volume=true
        -volume.dir.idx=/data/idx
        -volume.index=leveldbLarge
        -volume.max=0
        -volume.preStopSeconds=8
        -volume.readMode=redirect
        -volume.port=8080
        -volume.port.grpc=18080
        -ip=127.0.0.1
        -ip.bind=0.0.0.0
        -webdav=false
    environment:
      AWS_ACCESS_KEY_ID: sentry
      AWS_SECRET_ACCESS_KEY: sentry
      GOMAXPROCS: 2
    volumes:
      - "seaweed:/data"
    healthcheck:
      test: [
          "CMD-SHELL",
          # Manually override any http_proxy envvar that might be set, because
          # this wget does not support no_proxy. See:
          # https://github.com/getsentry/self-hosted/issues/1537
          "http_proxy='' wget -q -O- http://seaweedfs:8080/healthz http://seaweedfs:9333/cluster/healthz http://seaweedfs:8333/healthz || exit 1",
      ]
      interval: 30s
      timeout: 20s
      retries: 5
      start_period: 60s
    deploy:
      resources:
        limits:
          memory: 2G
        reservations:
          memory: 1G

• shutdown sentry
• apply this to compose file
• clean seaweed volume or attach filesystem path
• run install.sh - will create new buckets
• start sentry
• ???
• Profit!

[henri9813]

Hello,

Can you "reformat this please" ?

And explain the difference from the initial compose ?

Thanks !

[leorimmer]

Hello,

Can you "reformat this please" ?

And explain the difference from the initial compose ?

Thanks !

well. Differences are:

      resources:
        limits:
          memory: 2G
        reservations:
          memory: 1G

ENV variable: GOMAXPROCS: 2

And process itself. Biggest issue is when you start to limit seaweedfs with memory - indexes become corrupted, so you need new clean directory to build new idx, so FS has mounted volume instead of docker volume.

[aldy505]

I would recommend to just set GOMAXPROCS: 2 on the environment variables. That limits the CPU usage.

If you have 8 cores of CPU, seaweedfs will only use 2 cores.