still stuck with named memory leak

Mukund Sivaraman muks at isc.org
Wed Dec 24 00:29:41 UTC 2014


Hi Len

On Tue, Dec 23, 2014 at 03:02:49PM -0600, lconrad at go2france.com wrote:
> sent data to a ISC.org guy, no response.

Is this about the config file that you sent me? I had asked you to send
me dumps over time of the named statistics that are available via HTTP,
as the named process grows. See the "statistics-channels" documentation
in the manual. You can use curl or wget to dump them to a file.

We are very curious about these reports, but so far we do not have
enough information that points at any leaks. Even this week we have been
working with a different BIND user who, along with you, has reported
memory growth issues, but we haven't been able to find any leaks from
the given data.

> filled in the form on isc.org site, got a canned response:  "try bind user
> list ....  pay for support"
> 
> any suggestions?

The best way to report BIND bugs as a tracked ticket is to email
<bind9-bugs at isc.org>. This is the developers' bug tracker. Another way
is to discuss this on the BIND lists, which you have done.

Please do this now:

1. Enable the "statistics-channels" of named so that you are able to
dump XML statistics using HTTP and test that you are able to.

2. When you notice that named has started growing conspicuously and you
think it's process size is something it should not be, note down its
virtual process size (VSZ) and resident set size (RSS) fields which you
can get through ps, along with a timestamp.

3. Along with noting the VSZ and RSS sizes, also dump the corresponding
XML statistics using HTTP to a file.

Please send these stats to us, either via the <bind9-bugs at isc.org>
tracker, or this mailing list.

		Mukund
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 801 bytes
Desc: not available
URL: <https://lists.isc.org/pipermail/bind-users/attachments/20141224/7055379c/attachment.bin>


More information about the bind-users mailing list