Open jakubgs opened 2 years ago
@jakubgs so if you use the -<prefix>.consul.acl-token
command line argument to Cortex everything works, but if you configure the same in YAML everything works?
If possible would you be able to provide examples of config that works and doesn't work?
This is the config I was using:
target: 'distributor'
auth_enabled: false
configs:
database:
uri: 'memory://'
limits:
ingestion_rate: 1000000
ingestion_burst_size: 2000000
max_chunks_per_query: 7000000
cardinality_limit: 30000
max_series_per_metric: 30000
max_series_per_user: 5000000
max_label_names_per_series: 30
server:
http_listen_address: '0.0.0.0'
http_listen_port: 9092
grpc_listen_address: '0.0.0.0'
grpc_listen_port: 9095
log_level: 'debug'
storage:
engine: 'blocks'
distributor:
remote_timeout: '4s'
ring:
kvstore:
store: multi
etcd:
username: cortex
password: cortex
endpoints: ['10.10.0.10:2379', '10.10.0.11:2379', '10.10.0.12:2379']
consul:
host: localhost:8500
# WARNING: Has no effect currently.
acl_token: 00000000-0000-0000-0000-000000000000
multi:
primary: etcd
secondary: consul
mirror_enabled: true
mirror_timeout: 2s
And I didn't use -<prefix>.consul.acl-token
, I used -consul.acl-token
, because brevity is the soul of wit.
I wish there was a global
section in the YAML config file so I could define my ETCD and Consul configuration there for all targets without having to repeat the same section at times.
Describe the bug I've provided the Consul host and token using the flags so far, because it was a good way to avoid cluttering up my config with needlessly repeated sections of
kvstore
config. Today while trying to use ETCD alongside Consul using Multi KV I encountered this error when trying to usekvstore.consul.acl_token
instead indistributor.ring
config:After re-adding the
-consul.acl-token
flag the error went away, which suggests to me that theacl_token
config doesn't work.To Reproduce Steps to reproduce the behavior:
1.13.0
distributorkvstore.consul.acl_token
in configurationExpected behavior It works.
Environment: