INN FAQ Part 3/9

Go to the table of contents

Subject: (3.1) What runs as what?

1. The first "make install" MUST RUN AS ROOT.
2. Other "make install" runs as "root", but sometimes "news" will work.
3. "rc.news" runs as "root" for versions prior to 1.5 and as "news"

	for INN1.5 and later.

HOW TO REMEMBER THIS:

    EVERYTHING RUNS AS "NEWS" EXCEPT:
        "rc.news" runs as root, since it runs from rc.local (which
            runs as root) and since it runs things that start out
            as root but become "news" after doing special things.

        "make install" runs as "root" because it has to do "chown" to
          set up files securely.

If your innd nevertheless runs as root, then also look at 7.52

Go to the table of contents

Subject: (3.2) Why does innd just exit right away with no message?

First, fix your syslog: innd (nearly) always logs a message before it exits. (The INN distribution includes a version of the current UCB syslog, along with instructions on how to install it. Ultrix systems might want to look at the syslog that is available on gatekeeper.dec.com). If innd doesn't syslog, then it did dump a core file, that you can inspect.

Second, the most common cause of this is that you do not have a history file (or no history database). You will see a message like this:

	ME cant dbminit /usr/local/news/history No such file or directory

( Note that in versions > INN1.5, the word ``ME'' is replaced by ``SERVER'' ).
This means that you do not have a history database (or no history.dir and history.pag files). If you are just installing INN for your first time, you might want to run the BUILD script in your INN source tree.
Or, you can read about makehistory in doc/news-recovery.8; if you do the latter, make sure to rename the database files (pay attention to the two "mv" commands).

Another possibility is a bad active file; e.g. one line only has three fields instead of four. You can find such lines by running expire:

expire -n -x -t

which will then show you which lines are bad.

Go to the table of contents

Subject: (3.3) syslog message: inndstart: inndstart cant bind Address already in use

>I try to start INN but got this error in my syslog file:
> inndstart: inndstart cant bind Address already in use
>Can anyone give me a hint on how to fix it ?

Something else is already using port 119. Comment out the line in /etc/inetd.conf for port 119 and "kill -HUP" the inetd program (which makes it re-read the configuration file).

Remember, there are two ways for programs to take control over an incoming port. (1) the program runs and grabs the port (which has to be done as root) and then handles all incoming connections, or (2) the program is listed in /etc/inetd.conf and inetd grabs the port and hands off any incoming connections to the program specified in /etc/inetd.conf (i.e. one process gets started for every incoming connection.)

If you do (1) and (2) at the same time, the inetd will grab the port and then anyone else that grabs the port will get an "Address already in use" error.

INN's "innd" program is type (1). C News+NNTP uses something that is type (2).

Most people that get this error ran C News+NNTP a long time ago and forgot to remove the line from their /etc/inetd.conf file.

Go to the table of contents

Subject: (3.4) syslog message: ME internal no to group

If you set MERGE_TO_GROUPS to "DO", you have to have a "to" group listed in your "active" file or you will get the above syslog message and innd will not start. You can edit the active file directly if innd isn't running (see also 6.3).

Go to the table of contents

Subject: (3.5) syslog message: ME internal no control and/or junk group

You must have a newsgroup named "control" and a newsgroup named "junk" for innd to start. Since innd isn't running, you can edit the active file directly

Add these lines:

control 0000000000 0000000000 y junk 0000000000 0000000000 y

--tal

Also make sure that the directories in the spool either don't exist or are writable by the news user.

Go to the table of contents

Subject: (3.6) syslog message: Can't setup communication (bind failure)

The message "Can't setup communication (bind failure) Permission denied" means that the permissions on your _PATH_NEWSCONTROL directory are wrong. You might want to delete the directory and "make install" to create it again.

Go to the table of contents

Subject: (3.7) syslog message: ME bad_newsfeeds no feeding sites

(Rich Salz replies:) The syslog message is telling you that you are not feeding news to any sites. You have to have at least one feed. (You may consider this to be a bug, it's just that I'm too lazy to make everything work right if you don't have any newsfeeds.)

Until you go into production and start feeding sites, add a line like this:

		dummy-feed:!*::

Go to the table of contents

Subject: (3.8) syslog message: ME cant GetConfigValue

Jan 12 17:38:06 galaxy innd: ME cant GetConfigValue pathhost Error 0

                                                    ^^^^^^^^

This means you don't have "pathhost:" in your inn.conf.

GetConfigValue is the routine that gets data out of the inn.conf file.
If you get the above error, it means you don't have a particular value in your inn.conf. Run "inncheck -v" usually will tell you what you need to do.

Note that there might be a slightly different message:

Nov 1 15:39:30 serv nnrpd[5973]: cant getconfigvalue Error 0

In that case the host where nnrpd was running had DNS resolving problems. Getting DNS to work correctly again solved the problem.

Go to the table of contents

Subject: (3.9) syslog message: ME cant accept RCreader

Situation: INN 1.4sec on ICS2.0 or Interactive Unix System V Release 3.2 or one of the many SVR4 Unix's. You get syslog messages like:

Jul 14 12:07:44 isdn.IFNS.DE innd: ME cant accept RCreader No more Jul 14 12:07:44 isdn.IFNS.DE innd: ME cant accept RCreader Protocol error May 7 12:05:07 salyko.cube.net innd: ME cant accept RCreader Protocol error

when you connect to innd from a remote machine, but not from the local host.

Problem: Well, that's just another bug in the SVR4 networking code.

Chris Munonye <munonye@ifns.de> reports that this patch fixed the problem:

*** RCS/rc.c Sun Jul 18 15:02:50 1993 --- rc.c Sun Jul 18 15:04:35 1993 ***************
*** 173,178 ****
--- 173,181 ----

      /* Get the connection. */
      size = sizeof remote;

      if ((fd = accept(cp->fd, (struct sockaddr *)&remote, &size)) < 0) {
         syslog(L_ERROR, "%s cant accept RCreader %m", LogName);
         return;

Under Solaris, when accept() fails and sets errno to EPROTO, it means the client has sent a TCP reset before the connection has been accepted by the server.

Therefore, the syslog messages Inn produces:

         innd: ME cant accept RCreader Protocol error

Note that in 2.5.1 with the Internet Server Supplement and also in 2.6 the errno for this case has changed from EPROTO to ECONNABORTED.

Go to the table of contents

Subject: (3.10) syslog message: ME cant ioctl(TIOCNOTTY) Invalid argument

Most likely a botched config.data - TIOCNOTTY is a BSDism and probably doesn't work on SVR4-based system. (If you aren't sure if your Unix is SVR4-based, get out of the netnews business).

Go to the table of contents

Subject: (3.11) syslog message: ME cant fork /some/tool Resource temporarily unavailable

User news ran out of processes and wasn't able to fork off the specific tool.
You need either just raise the limit on the number of processes in the shell which starts innd (e.g in rc.news :

unlimit maxproc

or if that doesn't help raise the number of processes in the kernelconfig (modt often by raising maxusers and also raising the # of processes per user).

Go to the table of contents

Subject: (3.12) /usr/local/etc/rc.news: (M-i)^P^L^A.Gd: not found

Q: rc.news reports "/usr/local/etc/rc.news: (M-i)^P^L^A.Gd: not found"

A: Bob Izenberg <bei@io.com> reports:

Here's an unlikely possibility that bit us once. In an almost all-Sun environment, I requested and got /bin/ksh as my shell. When I logged in, I got a burst of control characters and a closed connection after a few seconds. It turned out that ksh followed about fifteen symbolic links to a file that had been moved/removed.

Therefore, make sure that the first line in rc.news refers to a shell that really exists. If you run "sh /usr/local/etc/rc.news" make sure you don't have anything else named "sh" in your path. (And if you have something named "sh" in your path that isn't a Bourne Shell or a clone, worry!)

Go to the table of contents

Subject: (3.13) syslog message: ME cant dbminit /usr/local/news/history

That means that some of the history files are missing -- See 3.2

There is also a message: cant dbminit ME (Math) argument out of domain of func This means that your history.dir file is not formatted correctly. It should look something like this: (from James Brister)

dbz 3 2579 9 C 128 127 24 4 3 2 1 0 4677 1706 1706 0 0 0 0 0 0 0 0

(i.e. two lines of ascii, most of it being numbers).

In the latter case you can do:

cd /where/the/history/is rm history.dir rm history.pag makehistory -r -f history