Running boostrapnet on a cluster

[lmguzman]

Hi Valentin,

Thank you for fixing the previous issue. I was able to install your package and run it perfectly on my laptop.

However, I am trying to run the package on a cluster and I get the following error:

Error in checkForRemoteErrors(lapply(cl, recvResult)) : one node produced an error: object '.doSnowGlobals' not found

I have loaded and re-installed the packages doParallel and doSNOW.

Some googling seems to suggest that I need to do set up the cluster in a different way:

cl<-makeCluster(type="MPI", 4)

Have you come across something like this before?

Melissa

[valentinitnelav]

Hi Melissa,

The bootstrapnet package is already designed with parallel processing in mind, using the foreach package, so conflicts with clusters are not surprising.

Most probably, if you have a cluster, you need to allocate a node with a certain number of cores for your job like you wrote (did that fix the problem?). I have tested the package only on Windows terminals with several cores, but not on Linux clusters (which might be your case).

Also bootstrapnet was not designed for "array jobs" on Linux clusters. That is a very specific case. Is that your case?

Valentin

[valentinitnelav]

Hi @lmguzman , I know this is long time ago, but if you fixed the issue, could you let me know what worked for you?

[lmguzman]

Hi @valentinitnelav, yes I was initially trying to use it so I could run multiple things in parallel. Im currently just using it sequentially.