BIND 10 #1762: loading huge zone crashes all
BIND 10 Development
do-not-reply at isc.org
Tue Mar 13 11:01:25 UTC 2012
#1762: loading huge zone crashes all
--------------------------------------------+----------------------------
Reporter: jreed | Owner:
Type: defect | Status: new
Priority: low | Milestone:
Component: Unclassified | Resolution:
Keywords: | Sensitive: 0
Defect Severity: N/A | Sub-Project: DNS
Feature Depending on Ticket: | Estimated Difficulty: 0
Add Hours to Ticket: 0 | Total Hours: 0
Internal?: 0 |
--------------------------------------------+----------------------------
Changes (by shane):
* priority: medium => low
* milestone: New Tasks =>
Comment:
Replying to [comment:2 jinmei]:
> Replying to [comment:1 jreed]:
>
> > We should have the components recognize when they can't allocate
memory and then shutdown nicely versus having confusing backtraces.
>
> I agree this is the desired behavior, but I'm not sure if it's
> feasible in a python program. Even handling a memory allocation
> failure may require some more memory and there may be no realistic
> exit than a crash.
I also agree that it is not always feasible, but I note that a lot of
these come from our message library, which could perhaps handle it a bit
more gracefully. But yes, basically when we are out of memory in the
general case our only option is to say "out of memory" and exit. We might
be able to drop some cache data or something like that in special cases,
but basically we have to exit.
Note that in Linux you may not have any notice that you are out of memory,
as the kernel overcommits and then kills processes to make more memory
available:
http://lwn.net/Articles/317814/
I'm not sure what specific action we should take here, but I'll leave the
ticket, although at a lower priority.
--
Ticket URL: <http://bind10.isc.org/ticket/1762#comment:3>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development
More information about the bind10-tickets
mailing list