Perl script slower on other server compared to dev (with same spec)

Question

I'm trying to figure out why a Perl script I have, isn't performing anywhere near as well once I put it on the live server.

The whole script is identical on both servers (one is a test server, the other the live one). Both are running this version of Perl:

This is perl 5, version 22, subversion 1 (v5.22.1) built for x86_64-linux-gnu-thread-multi
(with 73 registered patches, see perl -V for more detail)

The dev server is running:

mysql  Ver 14.14 Distrib 5.7.25, for Linux (x86_64) using  EditLine wrapper

Whereas the live server is on a slightly different version:

mysql  Ver 14.14 Distrib 5.7.17, for Linux (x86_64) using  EditLine wrapper

my.cnf files are the same on both servers. Could having such a smaller difference in mySQL versions cause the speed issues? Both servers are 8GB RAM,

The script has a lot of work to do (building static pages for a directory), and takes this long on each server:

As you can see, that's a whopping difference. Any suggestions on what else I could try?

Here is how htop looks on the live server when running the script:

and then on the dev server:

Doesn't seem to be much in it (or any swap being used)

@CloudAce thanks. Its the exact same data, same table structure, same indexes. Really baffling me. The only difference is the dev server has a 2048mb swap, whereas live has 512mb swap - but from what I've read, that's not really even relevant with an 8Gb ram server? — Andrew Newby, May 15 '19 at 06:39
Well, since this is the only difference, its worth to have a look, just to make sure. — Cloud Ace Wenyuan Jiang, May 15 '19 at 06:50
@CloudAce just done it, and starting a build now. Will see how long it takes (and if it made a difference) — Andrew Newby, May 15 '19 at 07:17
@CloudAce unfortunately it didn't help :( I've added the output of `htop` on both the live and dev server. If anything, the dev server has more stuff running on it (so you would expect it to be slower). — Andrew Newby, May 15 '19 at 07:29
Then maybe you should profile the script to see where cost time most — Cloud Ace Wenyuan Jiang, May 15 '19 at 07:36
@CloudAce any suggestions on the best way to do that? Is there a server tool that can do it, or would I need to do it in the Perl script? — Andrew Newby, May 15 '19 at 07:43
For example, `Devel::NYTProf`, or you can just use plain print in your script to measure the execution time — Cloud Ace Wenyuan Jiang, May 15 '19 at 07:46
@CloudAce thanks - will have a play and come back with my finding s:) — Andrew Newby, May 15 '19 at 07:47

0 Answers0