#Dagger gc is not kicking in after upgrade

1 messages · Page 1 of 1 (latest)

violet torrent
#

Hello Folks

I have recently upgraded to 0.11.8 and using go sdk.

We are seeing all our dagger fleet servers are at 100% disk utilization, causing production outage.

This .toml config

insecure-entitlements = ["security.insecure"]

[worker.containerd]
  enabled = false

[grpc]
  address = ["tcp://0.0.0.0:8080"]

[worker.oci]
  enabled = true
  gc = true
  snapshotter = "overlayfs"
  max-parallelism = 64
  cniPoolSize = 32

  [[worker.oci.gcpolicy]]
    keepBytes = "3200000MB"
    keepDuration = "0"
    all = false```
can anyone please help, this was smoothly working with engine 0.6.2 and now stopped working with 0.11.8?
keen dirge
#

See the release notes for v0.11.9 - this has been fixed in that release

violet torrent
#

does it need go sdk to be at same version or can i use v0.11.8

keen dirge
#

I'd recommend keeping all the versions the same, but you only need the engine to be on v0.11.9 in this specific case

violet torrent
#

I am also not quite sure if new version solve dagger engine server disk problem
as stated in release notes - prune local cache after session disconnects, it could be related to client side caching where dagger cli runs
correct me if i am wrong here

keen dirge
#

do you still see the caching issues when you upgrade?

violet torrent
#

i am still preparaing setup
Just in case if you know, what is fastest way to know if GC works, shall i reduce keepBytes = "100MB" lets says and run few builds

keen dirge
#

that approach you suggest seems good to me

violet torrent
#

Now, I am monitoring GC