gvegayon/parallel

Example 1 does not work

VickTaxiarchi opened this issue · 7 comments

Hello,

I have used the parallel package in Stata earlier with success but currently it's bringing up an error. The error is the same for every command I used it, eg parallel do, and is also the same when trying to reproduce example 1 of the help parallel Stata file:

sysuse bplong.dta
sort patient
parallel setclusters 4
parallel, by(patient): by patient: egen max_bp = max(bp)

The error it produces is:
"cluster XXX has finished with a connection error -601- (timeout)""cluster 000X has finished with a connection error -601- (timeout)"

Same error is produced when working the example on Stata 14 on the desktop, or Stata 16 on a server.

Any ides why this is produced and how I could overcome it please?

Best wishes
Vicky

Please specify what version of Stata and OS you are using. Also please try with latest version (uninstall parallel and then install using the GitHub installation method on the main page).

Please input the results of creturn list, also the output from one of the child logs (e.g., using parallel printlog) after a failure. It looks like it can't find the Stata executable so try specifying that when using parallel initialize.

Thanks for the reply, @VickTaxiarchi. Unfortunately, no images are attached when responding to the issue through your email. You can always include images if posting through the github site.

Hello,

Many thanks for keeping on with this issue. Please see images attached here, apologies for the number of them.

Best regards
Vicky

parallel issue v14 windows
parallel issue v16 1 linux (1)
parallel issue v16 1 linux (2)
parallel issue v16 1 linux (3)
parallel issue v16 1 linux (4)
parallel issue v16 1 linux (5)
parallel issue v16 1 linux (6)
parallel issue v16 1 linux (7)

@VickTaxiarchi, I think the problem is that you are using SSC's version of parallel. We have not been able to post a newer version (I think Dr. Baum was a bit busy at the time!,) so we have just kept the GitHub version up to date. I recommend you remove all source files from personal and other folders related to parallel. The mata library of parallel is called lparallel.mlib, so you'll have to remove that as well. Once you have a clean slate, proceed to install the GitHub version. Instructions to do so are here.