unable to take a subset of defined daemons to assign as a peer for an aggregator.

[emdonat]

I have defined a set of daemons:

daemons:
- names: "compute[1-30]"
  hosts: "compute[1-30]"
  endpoints:
  - names: "compute[1-30]-ib0"
    ports: 411
    xprt: sock
    auth:
      name: ovis
      plugin: ovis

- names: "admin[1-3]-agg"
  hosts: "admin[1-3]"
  endpoints:
  - names: "admin[1-3]-ib0"
    bind_all: True
    ports: 413
    xprt: sock
    auth:
      name: ovis
      plugin: ovis
      conf: "/tmp/inst/auth.conf"

- names: "admin[1-3]-samp"
  hosts: "admin[1-3]"
  endpoints:
  - names: "admin[1-3]-samp"
    ports: 411
    xprt: sock
    auth:
      name: ovis
      plugin: ovis

- names: "gw[1-3]"
  hosts: "gw[1-3]"
  endpoints:
  - names: "gw[1-3]-ib1"
    ports: 411
    xprt: sock
    auth:
      name: ovis
      plugin: ovis

- names: "login[1-2]"
  hosts: "login[1-2]"
  endpoints:
  - names: "login[1-2]-ib0"
    ports: 411
    xprt: sock
    auth:
      name: ovis
      plugin: ovis

and then am attempting to take a subset of these for an aggregator:

aggregators:
- daemons: "admin1-agg"
  peers:
  - daemons: "compute[1-10]"
    endpoints: "compute[1-10]-ib0"
    reconnect: 20s
    type: active
    updaters:
      - mode: pull
        interval: "1.0s"
        offset: "10ms"
        sets:
          - regex: ".*"
            field: inst
  - daemons: "gw1"
    endpoints: "gw1-ib1"
    reconnect: 20s
    type: active
    updaters:
      - mode: pull
        interval: "1.0s"
        offset: "10ms"
        sets:
          - regex: ".*"
            field: inst
  - daemons: "admin1-samp"
    endpoints: "admin1-samp"
    reconnect: 20s
    type: active
    updaters:
      - mode: pull
        interval: "1.0s"
        offset: "10ms"
        sets:
          - regex: ".*"
            field: inst
  - daemons: "login1"
    endpoints: "login1-ib0"
    reconnect: 20s
    type: active
    updaters:
      - mode: pull
        interval: "1.0s"
        offset: "10ms"
        sets:
          - regex: ".*"
            field: inst

and receiving the following error:

Traceback (most recent call last):
  File "/tmp/inst/bin/ldmsd_yaml_parser", line 30, in <module>
    cluster = YamlCfg(None, None, conf_spec, args)
  File "/tmp/inst/lib/python3.6/site-packages/ldmsd/parser_util.py", line 703, in __init__
    self.aggregators = self.build_aggregators(cluster_config)
  File "/tmp/inst/lib/python3.6/site-packages/ldmsd/parser_util.py", line 434, in build_aggregators
    raise ValueError(f'Daemons regex {daemons["names"]} does not match daemon regex "{group}"\n')
ValueError: Daemons regex login[1-2] does not match daemon regex "admin1-agg"

It appears that maybe it doesn't like taking a subset of the defined daemons? This seems like it should be fine to do.

[nick-enoent]

@emdonat Good catch, I implemented a fix today. Testing it on our network. I'll have pull request out this weekend, but it should be ready for you Monday.

[emdonat]

Just so it isn't lost in a merged pull request:

ldmsd_yaml_parser --ldms_config ./cluster.yaml --daemon_name admin1-agg
KeyError: 'compute[1-15]'

During handling of the above exception, another exception occurred:

Exception: Agg config Error: 'compute[1-15]' Line:972

I still don't seem to be able to take a subset of daemons per aggregator stanza as noted in #1583

[nick-enoent]

@emdonat This should be addressed in #1589