Bind 8.1.2 in.named memory leak in Solaris 7

Rob Remus nospam at remus.net
Mon Oct 11 20:39:14 UTC 1999


Thanks for the info!

We had been experiencing the same problem.  The problem lessened after
upgrading from 8.1.2 to 8.2.1 (we could make it a week instead of a
day before running out of memory), hopefully this patch will cure it.
I just installed it, and it seems to be much improved... I'll know in
a few hours... :)

For others who may be having the same problem, the patch mentioned
below is a libresolv patch.  You can find it at
http://sunsolve.sun.com.

-Rob

On 11 Oct 1999 09:46:59 -0700, Joaquim Eudes Mendes Gomide
<jgomide at bancobrasil.com.br> wrote:

>Hi William
>
>Thanks for the information. The patch 106938-02 corrects the problem.
>
>Joaquim Gomide
>
>"William P. Malloy" wrote:
>
>> In article <37FDEA86.C429C61E at bancobrasil.com.br> you write:
>> >I got an article on Sunsolve that says:
>> >"The DNS process in.named continually consumes memory until no memory
>> >is available. At this point in.named terminates leaving a possibly large
>> >core file."
>> >Is this problem a Bind 8.1.2  or a Solaris version of Bind bug? The same
>> >article says that there is no work around, too.
>> >Does the Bind 8.2 solves this problem? Is there any Solaris compiled
>> >version to download? Is there any Solaris patch? I'm looking for a patch
>> >with my reseler too.
>>
>> Does not Sunsolve also list patch 106938-02 (for sparc) 106938-02 (X86)
>> which fixes this bug (it is actually in libresolv)?
>> 4211042 Bind 8.1.2 in.named memory leak in Solaris 7
>>
>> It has been out since June 8th and I thought was one of the Recommended
>> Patches.
>> There is also an in.named patch 107018-01 (but it fixes a different bug)
>>
>> BTW Go to http://access1.sun.com/ and get the Patch reports for each
>> release.
>> I can not believe this is a Recommended Patch, probably because most
>> machines
>> are not running in.named, and a memory leak is not considered "serious"
>> enough?
>>
>> =wpm    William P. Malloy


More information about the bind-users mailing list