cluster error

[JieBNU]

@vwmaus Hi

I tried the dtwSat in Windows Server (64 cores,256GB memory), and used the function twdtwApplyParallel.
beginCluster(n=27)
r_twdtw1 <- twdtwApplyParallel(x = rts1, y = temporal_patterns1, weight.fun = log_fun1, progress = 'text')
endCluster()

When I set the parameter number of nodes (n) to 27, everything is fine.
But, when I set n to 28, it returns the error that "Error in twdtwApplyParallel.twdtwRaster(x, y, weight.fun, dist.method, : cluster error"

I am wondering how to release hold of the number of nodes(n)?

[Victor Maus]

@JieBNU I will try to reproduce the error here. Can you send me the output of str(rts1) and str(temporal_patterns1)?

[JieBNU]

@vwmaus Sure, I used the code and data that you provided in the dtwSat demo.
The code, output of str(rts1) and str(temporal_patterns1) are in the attach files.
str(rts1).txt

str(temporal_patterns).txt

demo_dtwSat1.txt

[Victor Maus]

@JieBNU the parallel processing splits the raster into chunks by rows. As the data in rts1 has only 27 rows it cannot be divided into 28 processing threads. Please, use the number of rows in the raster as the maximum number of cores beginCluster(n = dim(rts)[2]). Could you also try to set the number of rows beginCluster(n = dim(rts)[2]) in the Linux system #18?

[JieBNU]

@vwmaus Thanks, I used other data and it worked using 60 cores on Windows system.

[Victor Maus]

what about the GNU Linux system?