[ruby-core:92641] [Ruby trunk Bug#15844] VM has trouble releasing memory (?) on NoMethodError

From: duerst@...
Date: 2019-05-14 05:41:27 UTC
List: ruby-core #92641
Issue #15844 has been updated by duerst (Martin Dst).


nobu (Nobuyoshi Nakada) wrote:
> On exiting by an exception, ruby tries to print the exception and `NoMethodError` shows its receiver.
> Your `Node` is a recursive construction, so it increases exponentially by the size.

Recursion should not be a problem. Cycles definitely will be a problem.

> Define your own `inspect` method not to include `@left` and `@right` members.

This is definitely the right way to go. But would it make sense to wrap the error output in a timeout, to make it easier to debug such problems in the field? I can create a feature for this.

----------------------------------------
Bug #15844: VM has trouble releasing memory (?) on NoMethodError
https://bugs.ruby-lang.org/issues/15844#change-78004

* Author: alangano (Alan Gano)
* Status: Rejected
* Priority: Normal
* Assignee: 
* Target version: 
* ruby -v: ruby 2.6.3p62 (2019-04-16 revision 67580) [x86_64-linux]
* Backport: 2.4: UNKNOWN, 2.5: UNKNOWN, 2.6: UNKNOWN
----------------------------------------
When the test case is run, the process goes haywire.  I observe RSS bouncing around, and trending upwards.  CPU on the process is at 100%.

The test case is a partial implementation of a binary tree which is also a doubly linked list.  I chopped it down as far as I could.

The issue is triggered by a calling a non-existent method on the Tree object.  The test case only has 100 nodes -- it should be instantaneous.  I kill it after some 6+ minutes with RSS hitting upwards of 5GB.

I am working in 2.6.3.  Duplicated in 2.5.3 and 2.4.3 (with some code modifications)


---Files--------------------------------
issue.rb (2.67 KB)


-- 
https://bugs.ruby-lang.org/

Unsubscribe: <mailto:ruby-core-request@ruby-lang.org?subject=unsubscribe>
<http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>

In This Thread

Prev Next