control: reopen -1 thanks Hi Vagrant,
the distribution of jobs you chose is suboptimal, eg rpi2c and wbd0 both run workers 7+8, so if one of these two hosts goes down, the other one will have nothing to do as well. similar for opi2b and opi2c (4 workers each, and 33+34 shared) or odxu4, odxu4b and odxu4c, they all share 2 of 4 workers each, meaning if one of them goes down, another one will be half idle. it's not that obvious just by looking at the code in _build_service.sh (and also not by running that for-grep-loop which is there as a comment), it's more obvious at https://tests.reproducible-builds.org/debian/index_nodes_health.html please scramble the worker distribution a little more… -- cheers, Holger
signature.asc
Description: Digital signature
