You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
So I have a multii-node Ceph Cluster (Squid Release) deployed using microceph that is running on multiple Arm64 based boards (Orange Pis and Raspberry Pis)
services:
microceph.daemon: simple, enabled, active
microceph.mds: simple, enabled, active
microceph.mgr: simple, enabled, active
microceph.mon: simple, enabled, active
microceph.osd: simple, enabled, active
microceph.rbd-mirror: simple, disabled, inactive
microceph.rgw: simple, enabled, active
snap-id: ct1DgtIGBaljdOwomQGMLr8EJcL5pOP1
tracking: squid/edge
refresh-date: 37 days ago, at 16:19 CST
hold: forever
It's been working great. I have 2 RGW gateways exposed to internet using Cloudflare Tunnels. I am able to use AWS CLI and SDKs (both dart and python) to interact with it using a Cloudflare DNS.. the setup is simple. I have a cloudflare tunnel running on a Kubernetes cluster (separate from the nodes actually running the RGW Gateways) -> Offloads to a Envoy Proxy -> that load balances between my RGW Gateway instances.
I can confirm that Accounts and IAM API are also functional. So this profile for example is created under a new Account, by the root user of the account -> that then created this user.
Issue
The problem is: when I use this with a Terraform s3 backend it errors out only when using the Cloudflare DNS but not when I am using it with IP based endpoint:
terraform {
backend"s3" {
bucket="<redacted-bucket-name>"key="ceph-s3-gateway/terraform.tfstate"region="default"profile="Terraform"skip_region_validation=trueskip_credentials_validation=trueskip_metadata_api_check=trueskip_requesting_account_id=trueuse_path_style=trueskip_s3_checksum=trueendpoints={
# iam = "https://<redacted-record>.arkobasu.space"# s3 = "https://<redacted-record>.arkobasu.space/" <- THIS DOES NOT WORK
s3 ="http://192.168.5.81:80"<- THIS WORKS
}
}
required_providers {
random={
source ="hashicorp/random"
}
cloudflare={
source ="cloudflare/cloudflare"
version ="~> 4"
}
kubernetes={
source ="hashicorp/kubernetes"
version ="2.35.1"
}
}
}
I have been stuck on this for a bit. I have tried everything. The Signature validation does work when I am using something like Presign for objects and accessing them over the browser. I have investigated in Hashicorp's github.. and they say it's a support issue for the vendor (of the S3 provider). Hence writing here.
I would appreciate it bery much if you could give me some direction.
The text was updated successfully, but these errors were encountered:
Summary
So I have a multii-node Ceph Cluster (Squid Release) deployed using
microceph
that is running on multiple Arm64 based boards (Orange Pis and Raspberry Pis)It's been working great. I have 2 RGW gateways exposed to internet using Cloudflare Tunnels. I am able to use AWS CLI and SDKs (both dart and python) to interact with it using a Cloudflare DNS.. the setup is simple. I have a cloudflare tunnel running on a Kubernetes cluster (separate from the nodes actually running the RGW Gateways) -> Offloads to a Envoy Proxy -> that load balances between my RGW Gateway instances.
I can confirm that Accounts and IAM API are also functional. So this profile for example is created under a new Account, by the root user of the account -> that then created this user.
Issue
The problem is: when I use this with a Terraform
s3
backend it errors out only when using the Cloudflare DNS but not when I am using it with IP based endpoint:I have been stuck on this for a bit. I have tried everything. The Signature validation does work when I am using something like Presign for objects and accessing them over the browser. I have investigated in Hashicorp's github.. and they say it's a support issue for the vendor (of the S3 provider). Hence writing here.
I would appreciate it bery much if you could give me some direction.
The text was updated successfully, but these errors were encountered: