Parallel insert from R to MongoDB - r

Question

Question
I am using the package mongolite to connect from R to MongoDB database. I want to see whether it is possible to do multiple insert in parallel into the database? While mcmapply in the parallel package works in other cases, it doesn't seem to work for multiple insertion into MongoDB.
Example codes
I have a data frame grid.df containing grid IDs and geographical locations for the grid points. I want to insert the id_grid, lon and lat columns to the database.
insert.grid <- function(i){
# check which data to insert
criteria <- sprintf('{"id_grid" : %s}', id_grid[i])
# values to be inserted
newdoc <- sprintf('{"$set" : {"id_grid" : %s, "loc" : {"type" : "Point", "coordinates" : [%s, %s] } } }',
grid.df$id_grid[i],
round(grid.df$lon[i], digits = 4),
round(grid.df$lat[i], digits = 4)
)
inserted <- F
while (!inserted) {
inserted <- gcon$update(query = criteria, update = newdoc, upsert = T)
}
}
Attempts
1) Using mcmapply
mcmapply works fine with 1 core. However, as soon as I try to use multiple cores, it fails.
results <- mcmapply(insert.grid, 1:nrow(grid.df), mc.cores = 4)
Setting mc.preschedule parameter to FALSE didn't help neither.
results <- mcmapply(insert.grid, 1:nrow(grid.df), mc.cores = 4, mc.preschedule = FALSE)
Both are giving the following error:
Warning message: In mclapply(seq_len(n), do_one, mc.preschedule =
mc.preschedule, : all scheduled cores encountered errors in user
code
2) Using file locks
I also followed the approach introduced here: http://www.quintuitive.com/2014/11/20/synchronization-for-r-with-the-flock-package/ and while it works with multiple cores, I do not really gain much in the performance since making the locks for every insert completely slows down the process and it similar to use 1 core.

Answer 1

This was solved by moving the gcon connection inside the function definition. The problem was simply that only one connection was defined and could not work for parallel insert.