Kubernetes Name Schemas

Hostname Schema

Domain TLD Usage

Services: example.services
Servers (bare metal, VMs, doesn't matter): example.systems
Network hardware: example.network

Schema Parts

COUNTRY - ISO 3166-1 alpha-3 code, see ISO 3166-1 alpha-3 - Wikipedia and Online Browsing Platform (OBP) - ISO.
PROVIDER - 3 character long abbreviation of provider name.
DC - Optional info about datacenter/region (e.g., FSN1-DC1).
CLUSTER - Cluster designation (in case of Kubernetes, should always have k8s in the beginning) and if there can be multiple a number added (with two digits, e.g., 01, 12).
ROLE - In case of Kubernetes, e.g., master, etcd, node (other "special" roles could be, e.g., ingress, stora (could have a suffix per storage software, e.g. storaceph)).
COUNT_OR_ID - A count is a special "type". It can for servers that are known to have only a maximum of n machines at maximum, the number of servers padded with zeroes (e.g., max 12 servers results in COUNT for the third machine being 03), in case of nodes where there can be an undefined amount of them it should be a shortid.
- To generate a "random" ID (requires bashids to be installed):
```
bashids \
  -e \
  -s 'B_r1KMOASvn_5A1hDKCdPXJfrIBcddpwOnT5orXYaQPV4Ixb1zNSpa-nF6HOw8mii3pqovZUtsnGZ5pqbf59wPfeMp9XagGXc8ViJreL_5J1kvSnDCPfqvuV2bmGsx4DrVV_ef3Gr3MgCMrX86TGUjCDeJmM3LONAfKIH_vv0ZR9WWcJJbLCc5xnxWh7Is8qNq95ORIHS6iU4gKZNV-LIxdYxd7WyO2fKeOn8kApv0FFD2ydkJXdz4KjqBEcN5Fu' \
  $(date +%s%6N) | \
    tr '[:upper:]' '[:lower:]'
```

Servers/ VMs Schema

# When country, provider and dc should be omited:
{CLUSTER}-{ROLE}-{COUNT_OR_ID}.example.systems
# Examples:
## Master/ Primary Kubernetes node
k8s02-master-01.example.systems
## Node/ Worker Kubernetes node
k8s02-node-4ua16bzb6r7.example.systems

# With country, provider and dc in the name:
{CLUSTER}-{ROLE}-{COUNT_OR_ID}-{PROVIDER}-{COUNTRY}{DC}.example.systems
# Examples:
## `deu-fsn1dc1` translates to Germany, FSN1-DC14 (Falkenstein)
k8s02-master-01-htz-deu-fsn1dc14.example.systems
## `deu-fsn1dc1` translates to Germany, FSN1-DC1 (Falkenstein)
k8s02-node-4ua16bzb6r7-htz-deu-fsn1dc1.example.systems
## `eee-west2` translates to GCP Region `europe-west2`
k8s02-master-02-gcp-eee-west2.example.systems
## `usa-west1` translates to AWS Region `us-west1`
k8s02-node-7ca16bnb2r1-aws-usa-west1.example.systems

Script: Gernate Hostname (+ ID)

CLUSTER="k8s02"
ROLE="node"
COUNT_OR_ID="$(sleep 0.00001; \
bashids \
    -e \
    -s 'B_r1KMOASvn_5A1hDKCdPXJfrIBcddpwOnT5orXYaQPV4Ixb1zNSpa-nF6HOw8mii3pqovZUtsnGZ5pqbf59wPfeMp9XagGXc8ViJreL_5J1kvSnDCPfqvuV2bmGsx4DrVV_ef3Gr3MgCMrX86TGUjCDeJmM3LONAfKIH_vv0ZR9WWcJJbLCc5xnxWh7Is8qNq95ORIHS6iU4gKZNV-LIxdYxd7WyO2fKeOn8kApv0FFD2ydkJXdz4KjqBEcN5Fu' \
    $(date +%s%6N) | \
        tr '[:upper:]' '[:lower:]')"
echo "$CLUSTER-$ROLE-$COUNT_OR_ID.example.systems"
echo "$CLUSTER-$ROLE-$COUNT_OR_ID-htz-deu-fsn1dc1.example.systems"

Note: The sleep 0.00001 is used to try to prevent "duplicates" when bashids is run in parallel to generate IDs.

Infrastructure/ Cluster components should be monitored separately from your applications. This allows you to "kill" the application Prometheus in case you have screwed up in some way (e.g., messed up application metrics causing to have a billion labeled metrics).