The main issue is that S.o.S runs two lightweight scheduled searches to populate two lookups, one of which is heavily used in the app:
sos_splunk_instances_info -> splunk_instances_info.csv -> used to populate the "A glimpse of your Splunk instance" info panel in the Home view
sos_refresh_splunk_servers_cache -> splunk_servers_cache.csv -> used to populate the "Server to query" pulldown which exists in most views
At a minimum, you'll need to manually run those searches once to ensure that S.o.S discovers your topology. Given that you are in a search-head pool, you'll need to run this once per search-head in the pool.
Here's how you would run these searches manually:
Go to the flashtimeline within the context of the S.o.S app - http[s]:[splunkweb_hostname]:[splunkweb_port]/en-US/app/sos/flashtimeline
Run the following two searches: | savedsearch sos_refresh_splunk_servers_cache | savedsearch sos_splunk_instances_info | outputlookup splunk_instances_info
Note that you should also run these searches whenever a new indexer or pooled search-head is added, to help S.o.S discover it.