Memory “leak” in whateverable

[AlexDaniel]

This is another ticket in whateverable series. See also: #1278 #1259.

This one, however, is easier to reproduce.

git clone git@github.com:perl6/whateverable.git
cd whateverable
zef install Sake  IRC::Client  File::Temp  File::Directory::Tree  HTTP::UserAgent  IO::Socket::SSL  Pastebin::Gist  SVG::Plot  Stats JSON::Fast  Text::Diff::Sift4  Terminal::ANSIColor  IRC::TextColor  Config::INI  App::GPTrixie  Number::Denominate  HTTP::Server::Async  Digest  Digest::HMAC

DEBUGGABLE=1 PERL6LIB=lib perl6 bin/Unicodable.p6 # it will join #whateverable on freenode, so go there and play with it

Here's an example:

<AlexDaniel> benchable6: uptime
<benchable6> AlexDaniel, 1 day, 1 hour, 13 minutes, and 17 seconds, 1101.556MB maxrss. This is Rakudo version 2017.10-215-g85105077a built on MoarVM version 2017.10-86-g89fae17a6 implementing Perl 6.c.

So by running the uptime command periodically you'll see that the memory is growing. This is a regression and it was not happening earlier.

[AlexDaniel]

After I upgraded rakudo yesterday the leak seems to be gone.

For example:

<AlexDaniel> committable6: uptime
<committable6> AlexDaniel, 15 hours, 53 minutes, and 11 seconds, 255.550781MiB maxrss. This is Rakudo version 2018.01-29-ga2499c90f built on MoarVM version 2018.01 implementing Perl 6.c.

So it settles around 250MiB which is fine. I'll keep an eye on it, and if that's true I'll close the ticket. I think we shouldn't bother with tests for this one because the bots are consistently on the channel and if they start leaking it'll go noticeably bad for my server which does not have enough memory to keep up with 13 hungry bots.

[AlexDaniel]

OK this is not entirely gone, but it got much better. I'm still keeping my eye on this. See #1501 for a memory leak that is more tangible.

[AlexDaniel]

Did I say 13 hungry bots? 17 hungry bots.

Anyway, the leak is no longer there, and IMO this is resolved. I don't think we'll be able to come up with a nice test for the issue, but whateverable is an active project and I'm actively monitoring memory usage of the bots (there's also some logging in place), so if the issue comes back we'll notice.