Leonardo team queue

s_tra_ncc connecting --
Where our jobs sit on the hackathon reservation.

Node capacity

--nodes free
--nodes busy
--total
--A100 free (est)
A whole node = 4 A100. Other teams pack 1-GPU jobs onto shared nodes, so "busy" means at least one GPU in use, not fully allocated. For a TP=4 vLLM run you want a fully free node.

Reservation

account--
partition--
state--
window--

Team jobs

JobRoleStateTimeLimitWhere / whyStart

Reservation queue · running first, then pending by priority

JobUserStateNTimeLimitStart