Skip Menu |

This queue is for tickets about the Net-Async-HTTP CPAN distribution.

Report information
The Basics
Id: 124927
Status: resolved
Priority: 0/
Queue: Net-Async-HTTP

People
Owner: Nobody in particular
Requestors: andy [...] andybev.com
Cc:
AdminCc:

Bug Information
Severity: (no value)
Broken in: (no value)
Fixed in: (no value)



Subject: Suspected memory leak
Date: Mon, 26 Mar 2018 21:25:11 +0100
To: bug-Net-Async-HTTP [...] rt.cpan.org
From: Andrew Beverley <andy [...] andybev.com>
When GETting from some HTTPS hosts, it appears that some memory is not released, in particular arrays (and their references). Strangely this only happens with some HTTPS hosts, not all, but I cannot work out the difference between those that do and don't exhibit the problem. The attached script shows the behaviour, printing the number of arrays in the arena after each set of requests. After running the script for an hour or so, you can see that the number gradually increases. I have tried this with 2 machines on 2 different networks: one is Perl 5.22.1 and the other is 5.24.1. Both use Net::Async::HTTP version 0.41. I would like to sponsor the fix for this problem, if possible. Thanks!

Message body is not shown because sender requested not to inline it.

Subject: Re: [rt.cpan.org #124927] Suspected memory leak
Date: Tue, 27 Mar 2018 10:30:50 +0100
To: bug-Net-Async-HTTP [...] rt.cpan.org
From: Andrew Beverley <andy [...] andybev.com>
I'm not sure whether it helps, but after the process has been running for a long time, a core dump of the process's memory shows lots of data that looks like SSL certificate information. For example: 295487 www.digicert.com1 0 197097 DigiCert Inc1 168802 DigiCert Global Root CA 98622 DigiCert Global Root CA0 98579 digicert inc1 98496 O_lv^ 98490 JL}> 98487 www.digicert.com 98487 DigiCert Inc The numbers above define how many references of that particular string there are in the process's memory.
I suspect this may be a bug in IO::Socket::SSL itself. I have opened at ticket at https://rt.cpan.org/Ticket/Display.html?id=125867 -- Paul Evans
On Tue Jul 17 12:44:34 2018, PEVANS wrote: Show quoted text
> I suspect this may be a bug in IO::Socket::SSL itself. I have opened > at ticket at > https://rt.cpan.org/Ticket/Display.html?id=125867
This was indeed an SSL problem that was fixed upstream. -- Paul Evans