I do run into the same issue as #3 when supplying the cluster number I do obtain results, though when trying the automatic knee approach it fails in the mykmeans step in the infer_tree_structure call.
I do get the error:
file descriptor is too large for select()
Error in sendMaster(try(lapply(X = S, FUN = FUN, ...), silent = TRUE)) :
ignoring SIGPIPE signal
so the kmeans is creating a lot of threads by default and is using hard set 30 cores, this apparently creates in my case too many open connections.
this functions would need some parameters to allow to adjust the parallelization
I do run into the same issue as #3 when supplying the cluster number I do obtain results, though when trying the automatic knee approach it fails in the mykmeans step in the infer_tree_structure call.
I do get the error:
so the kmeans is creating a lot of threads by default and is using hard set 30 cores, this apparently creates in my case too many open connections.
this functions would need some parameters to allow to adjust the parallelization