ClusterManagers.addprocs_lsf
— Methodaddprocs_lsf(np::Integer;
bsub_flags::Cmd=``,
ssh_cmd::Cmd=``,
retry_delays=ExponentialBackOff(n=10,
first_delay=1, max_delay=512,
factor=2),
throttle::Integer=np,
params...) =
Launch np
workers on a cluster managed by IBM's Platform Load Sharing Facility. bsub_flags
can be used to pass flags to bsub
that are specific to your cluster or workflow needs. ssh_cmd
can be used to launch workers from other than the cluster head node (e.g. your personal workstation). retry_delays
is a vector of numbers specifying in seconds how long to repeatedly wait for a worker to start. throttle
specifies how many workers to launch at once.
Examples
addprocs_lsf(1000; ssh_cmd=`ssh login`, throttle=10)