Help needed NSUPDATE issue Solaris 8

Brad Knowles brad.knowles at skynet.be
Tue Sep 11 07:52:54 UTC 2001


At 7:36 PM -0700 9/10/01, Sivakumar Thiyagarajan wrote:

>     I really dont know how the BIND release was
>  compiled. I have been thinking that it ships with
>  Solaris installation CD!! Isnt it so?!? I have never
>  needed to install it except to upgrade BIND.

	If you take the version of BIND that is shipped by Sun, then I'm 
sure it is compiled with their own compiler.  However, that version 
is almost always many years out of date, and if you want to do 
anything with DNSSEC or Dynamic Updates, you really need to be 
running the latest version of BIND from the ISC.

	This means that you will either have to get the source code 
yourself and compile it, or you will have to find a version that 
someone else has gotten, compiled, and turned into a package that you 
can add.

	If you take the package, you have no idea what options they used 
to compile it, or what compiler they used.  If you get the source 
code and build it yourself, then you know exactly how it is built, 
with what options, and with what compiler.

>     BTW we seem to have a grip on the other part of my
>  problem with nsupdate giving up with a 'send error'.
>
>     Looks like in a rapid fire of nsupdate calls, the
>  named does not always respond fast enough. successive
>  nsupdates in turn keeps timing out and often
>  successfully gets a response from the named. But then
>  randomly nsupdate gives up exits with a 'send error'.
>  ( does it suddenly decide that there is no named? )
>
>     The exact msg is "res_send: send error, n=-1"
>
>     Introducing a sleep between nsupdate calls did help
>  but still it feels scarry.

	All I can say is that if you update to the latest version of BIND 
(9.1.3-REL), which is built with threading disabled and using the 
vendor compiler (not gcc), and you still get this error, then you 
should report this fact and as much data as you can to the BIND 9 
developers, whom I'm sure would be very interested to replicate and 
then fix this bug.

-- 
Brad Knowles, <brad.knowles at skynet.be>

H4sICIFgXzsCA2RtYS1zaWcAPVHLbsMwDDvXX0H0kkvbfxiwVw8FCmzAzqqj1F4dy7CdBfn7
Kc6wmyGRFEnvvxiWQoCvqI7RSWTcfGXQNqCUAnfIU+AT8OZ/GCNjRVlH0bKpguJkxiITZqes
MxwpSucyDJzXxQEUe/ihgXqJXUXwD9ajB6NHonLmNrUSK9nacHQnH097szO74xFXqtlbT3il
wMsBz5cnfCR5cEmci0Rj9u/jqBbPeES1I4PeFBXPUIT1XDSOuutFXylzrQvGyboWstCoQZyP
dxX4dLx0eauFe1x9puhoi0Ao1omEJo+BZ6XLVNaVpWiKekxN0VK2VMpmAy+Bk7ZV4SO+p1L/
uErNRS/qH2iFU+iNOtbcmVt9N16lfF7tLv9FXNj8AiyNcOi1AQAA


More information about the bind-users mailing list