Skip to content

Commit

Permalink
Do not set resources limits by default
Browse files Browse the repository at this point in the history 
I've experienced issues in the past where MetalLB was throttled
for 15s and then replying to all previous ARP request at the same time.
MetalLB is the entry point of the cluster, if it's not working everything
else in the cluster is useless, so let's just not set limits.

Signed-off-by: Etienne Champetier <echampetier@anevia.com>
  • Loading branch information
@champtar @russellb
champtar authored and russellb committed Mar 26, 2021
1 parent de85dc6 commit ed1d259
Showing 1 changed file with 0 additions and 8 deletions.
8 changes: 0 additions & 8 deletions manifests/metallb.yaml
View file
Expand Up @@ -324,10 +324,6 @@ spec:
- containerPort: 7946
name: memberlist-udp
protocol: UDP
resources:
limits:
cpu: 100m
memory: 100Mi
securityContext:
allowPrivilegeEscalation: false
capabilities:
Expand Down Expand Up @@ -378,10 +374,6 @@ spec:
ports:
- containerPort: 7472
name: monitoring
resources:
limits:
cpu: 100m
memory: 100Mi
securityContext:
allowPrivilegeEscalation: false
capabilities:
Expand Down

2 comments on commit ed1d259

@ohauer
Copy link

@ohauer ohauer commented on ed1d259 Jun 25, 2021

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Removing the resorces results in the containesrs running only with best effort and is something you really don't want to have on a cluster with high load.
Wouldn't it be better to set additional requests to reserve CPU and memory to avoid such situations?
E.g. I'm using this values on a high load dev cluster with arp and in another location with bgb.

resources:
  limits:
    cpu: 100m
    memory: 128Mi
  requests:
    cpu: 15m
    memory: 56Mi

@champtar
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You can set limits if you want, but 100m was too small for me / causing prod issue. We can't provide sane CPU limits that fits everyone so better leave it to the users.

Please sign in to comment.