bind 8.2.2-p5 ns_req errors with connection refused.

bind 8.2.2-p5 ns_req errors with connection refused.

Post by f.. » Fri, 07 Jan 2000 04:00:00



Hey, all.

I keep getting ns_req connection refused messages like:

messages.Dec1899.gz:Dec 17 17:18:36 www5 named[5484]: ns_req:
sendto([192.168.254.192].2653): Connection refused
messages.Dec1899.gz:Dec 17 17:18:36 www5 named[5484]: ns_req:
sendto([192.168.254.209].1489): Connection refused
messages.Dec1899.gz:Dec 17 17:18:36 www5 named[5484]: ns_req:
sendto([192.168.253.244].1829): Connection refused

My version of bind is 8.2.2-p5 and it works fine, but at times (at what
seems to be random times) I will  get this error message and the DNS
server doesn't respond.

The only way I have found to fix it is to send a HUP to the named or
wait an undetermined amount of time for it to work again (it's
undetermined, because I send a HUP as more then 5 minutes and management
would be on my tail)

So any insight of why the DNS server is stalling would be appreciated.

Thanks. K.J.

P.S. My external DNS server works with no problems.

.

 
 
 

bind 8.2.2-p5 ns_req errors with connection refused.

Post by Mark.Andr.. » Fri, 07 Jan 2000 04:00:00



> > Hey, all.

> > I keep getting ns_req connection refused messages like:

> > messages.Dec1899.gz:Dec 17 17:18:36 www5 named[5484]: ns_req:
> > sendto([192.168.254.192].2653): Connection refused
> > messages.Dec1899.gz:Dec 17 17:18:36 www5 named[5484]: ns_req:
> > sendto([192.168.254.209].1489): Connection refused
> > messages.Dec1899.gz:Dec 17 17:18:36 www5 named[5484]: ns_req:
> > sendto([192.168.253.244].1829): Connection refused

> > My version of bind is 8.2.2-p5 and it works fine, but at times (at what
> > seems to be random times) I will  get this error message and the DNS
> > server doesn't respond.

> > The only way I have found to fix it is to send a HUP to the named or
> > wait an undetermined amount of time for it to work again (it's
> > undetermined, because I send a HUP as more then 5 minutes and management
> > would be on my tail)

> > So any insight of why the DNS server is stalling would be appreciated.

> Sorry I can't offer anything specific, but this looks more like a system
> problem than a BIND bug. sendto() is a fairly low-level networking routine,
> and "Connection refused" is something the networking stack returns. Perhaps
> you are exhausting some sort of resource on the machine, which goes away
> temporarily when you reload named (?). How does the machine look otherwise?

        'sentto' on linux returns a error when the socket has received a
        ICMP port unreachable message.  Most other stacks only return this
        message when the socket is 'connected'.

        As for why named stalled, I suspect that it was the result of
        syslog stalling.  On most platforms syslog does not block.  Linux
        implementations block unless recent patches have been applied.

        By the time you see these error messages the stall had ended as these
        are the result of send messages to clients that have timed out and
        closed their UDP socket.

        Mark
--
Mark Andrews, Nominum Inc. / Internet Software Consortium
1 Seymour St., Dundas Valley, NSW 2117, Australia


 
 
 

1. ns_req Connection refused

Hi,

Can someone help with this question: our bind8 server has been giving
error msgs like this:
ns_req: sendto([171.168.2.10].53): Connection refused  
I've looked around this mailing list for answers and understand that it
means the server stalled and by the time it replied the query timed out
on the querier's sideand therefore was refused.
This was also explained as normal and acceptable to a certain extent.
The question is I'm concerned about the rate at which we're getting
these -- 624 since Sept 28.
and when I look back, these are the counts of ns_req errors since the
beginning of the year.
/var/log/messages-20010131.gz        11
/var/log/messages-20010315.gz         6
/var/log/messages-20010415.gz         1
/var/log/messages-20010421.gz         0
/var/log/messages-20010604.gz         1
/var/log/messages-20010711.gz         0
/var/log/messages-20010820.gz         3
/var/log/messages-20010824.gz       133
/var/log/messages-20010830.gz       138
/var/log/messages-20010905.gz       147
/var/log/messages-20010911.gz       234
/var/log/messages-20010920.gz       699
/var/log/messages-20010928.gz       807

Are these numbers high? Meaning our server is somehow overloaded?
We have also been bombarded by windows boxes trying to do nsupdate.
When I do a grep 'denied update', this is what I see:
/var/log/messages-20010131.gz         0
/var/log/messages-20010315.gz         0
/var/log/messages-20010415.gz         0
/var/log/messages-20010421.gz         0
/var/log/messages-20010604.gz         0
/var/log/messages-20010711.gz         0
/var/log/messages-20010820.gz         0
/var/log/messages-20010824.gz     13074
/var/log/messages-20010830.gz     17955
/var/log/messages-20010905.gz     17964
/var/log/messages-20010911.gz     19656
/var/log/messages-20010920.gz     30360
/var/log/messages-20010928.gz     29554

There may be nothing we can do about the windows boxes, and I'm trying
to decide whether I need
to worry about the servers load...or just turn down the logging. We're
running at debug level 1 permanently.
  Can someone tell me if these are normal...?
We're running a 8.2.3 bind SuSE 7.0 on a pentium3 2-processor smp,
645Hz, 512M.
--
Claire Chan
Publishing Support Center
Heinrich Bauer PKG
Tel: 49-40-3019-5515

2. real time RS232 connections

3. ns_req: Connection refused

4. MSXBoeken

5. HELP - Bind Compile Error ( Bind 8.2.2-P5 / Slackware 7.1 / Ker 2.2.16 / Chroot )

6. Commercial Plan9, again?

7. Compile BIND 8.2.2-P5 makes errors

8. selecting VMs

9. Solaris 2.6 Upgrade to BIND 8.2.2-p5 Error Message

10. RedHat 6.2 and Bind 8.2.2-p5 error

11. Compiling errors for BIND 8.2.2 P5 on Solaris 2.6 with GCC

12. help to find errors please bind 8.22 P5

13. bind 8.2.2.P5 "No such file or directory" error