flux-framework / flux-core

core services for the Flux resource management framework
GNU Lesser General Public License v3.0
167 stars 50 forks source link

`flux overlay status` is slow on large systems #6026

Closed grondo closed 3 months ago

grondo commented 4 months ago

flux overlay status is slow on large system instances with a flat TBON. There is potential that it is unnecessarily contacting all rank 0 children serially to produce the default output.

Also, for large systems the default output of listing every rank individually is perhaps not that useful. Some sort of compressed format (thought it isn't clear what that would be) or summary (number of brokers online vs degraded vs lost etc) would be more useful (whether in a separate subcommand or not)