home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.apollo
- Path: sparky!uunet!utcsri!helios.physics.utoronto.ca!alchemy.chem.utoronto.ca!system
- From: system@alchemy.chem.utoronto.ca (System Admin (Mike Peterson))
- Subject: Re: named causing tcpd to spin
- Message-ID: <1992Nov18.181813.1527@alchemy.chem.utoronto.ca>
- Organization: University of Toronto Chemistry Department
- References: <1992Nov17.140615.3143@nessie.mcc.ac.uk>
- Date: Wed, 18 Nov 1992 18:18:13 GMT
- Lines: 39
-
- In article <1992Nov17.140615.3143@nessie.mcc.ac.uk> bob@sna.co.umist.ac.uk (Bob Nutter) writes:
- >For the second time lately, named seems to be the cause of tcpd going into a
- >spin with the consequence that all connections are lost. Rebooting has no
- >effect, other than it works for a few minutes and then off it goes again. The
- >only cure is to stop running named on that machine.
- >
- >Has anyone else had this happen to them, and if so, what's the cure? We've got
- >10.4 but I don't really have the time to put it on (we're at 10.3.5 at the
- >moment). Is tcpd/named different at 10.4?
-
- Named does not work properly. Using the $INCLUDE directive in the
- data files causes massive corruption of the data as seen by
- named (many host IP addresses are wrong, some end up with 255
- in them which is illegal). Also, the SOA numeric data fields (time to
- live, etc) are incorrect, and SOA records in the included files are
- not seen (messages in /usr/adm/syslog say:
-
- Oct 28 10:12:53 aurum named[1081]: restarted
- Oct 28 10:12:56 aurum named[1081]: primary/chem.utoronto.ca: no SOA record
- Oct 28 10:12:59 aurum named[1081]: primary/chem.toronto.edu: no SOA record
-
- ). This then causes named to fail to resolve names it should be resolving.
-
- I tried removing all the $INCLUDE from the data files, and that didn't
- fix the problem, so I recompiled named from original source code,
- plus some patches. That version of named is what I run on our HP-UX and
- Domain/OS systems (HP-UX named has the same problem), and it seems to
- work.
-
- This is still broken at SR10.4.
-
- On a related front, note that the 'gethostbyname' system call does not
- return the fully qualified domain name on nodes running a name server
- (it works properly on systems that access remote name servers). This
- caused one of the USENET news programs (nntp or rn) to fall over, since
- the host name didn't match what it expected.
- --
- What are the chances that any HP computer system will ever "work" properly?
- ... and Slim just left town. -*- Mike Peterson, SysAdmin, U/Toronto Chemistry
-