[comp.sys.apollo] /etc/syslog goes wild??

danny@idacom.UUCP (Danny Wilson) (05/29/89)

We are running SR9.7 Aegis with Domain/IX (BSD4.2).
Every once in a while I notice that the syslog process is consuming
CPU time like crazy (up to 90% as reported by dspst).

Both /bin/ps and /com/pst show that something is really 
happening there.

Our /etc/syslog.conf contains '1*/dev/log', so trying to 'catf'
/dev/log results in:

	Error: unable to open file "log" - system (or process) crash 
	prevented complete file close.

(or something like that...)

Has anyone seen this before? How do I fix it? (Is it really worth
running syslog at all?)

The answers to these questions and more can (hopefully) be found...
_on_the_network_.

-- 
Danny Wilson
IDACOM Electronics		danny@idacom.uucp
Edmonton, Alberta		alberta!idacom!danny
C A N A D A

weber_w@apollo.COM (Walt Weber) (06/01/89)

In article <674@idacom.UUCP> danny@idacom.UUCP (Danny Wilson) writes:
>Our /etc/syslog.conf contains '1*/dev/log', so trying to 'catf'
>/dev/log results in:
>
>	Error: unable to open file "log" - system (or process) crash 
>	prevented complete file close.

The extensible streams documentation (Open Systems Toolkit) discusses the
32-byte streams header which precedes user data on files of type 'uasc'.
The message above indicates that the /dev/log file has a corrupted
streams header.  Change it's name so that you can create a new one, and
syslogd will revert to it's normal, unobtrusive self...

Note that /com/catf attempts to open a file using constraints of "nR_xor_1W",
which is N-Readers eXclusive-OR 1-Writer.  Since syslogd opens the file for
writing via open(2) (which uses constraint Cowriters, or "ios_$unregulated"),
your catf will fail.  If you use /bin/cat (which also uses open(2)) you will
be able to look at the file, PROVIDING you make the attempt from the same
node which is holding the lock (read: node where syslog is running).

...walt...

>Danny Wilson
>IDACOM Electronics		danny@idacom.uucp
-- 
Walt Weber                    Apollo Division of the Hewlett-Packard Company
(508) 256-6600 x8315          People's Republic of Massachusetts
-The views expressed herein are personal, and not binding on ANYONE-

krowitz@RICHTER.MIT.EDU (David Krowitz) (06/02/89)

Talking about corrupted stream headers ... is their anyway to
retrieve the contents of a file that has that error message
(file not closed properly due to system crash ...)? Is there
some way just to patch up the header of a UASC or REC file?


 -- David Krowitz

krowitz@richter.mit.edu   (18.83.0.109)
krowitz%richter@eddie.mit.edu
krowitz%richter@athena.mit.edu
krowitz%richter.mit.edu@mitvma.bitnet
(in order of decreasing preference)

mishkin@apollo.COM (Nathaniel Mishkin) (06/03/89)

In article <438f3c86.10b48@apollo.COM> weber_w@apollo.COM (Walt Weber) writes:
>In article <674@idacom.UUCP> danny@idacom.UUCP (Danny Wilson) writes:
>>Our /etc/syslog.conf contains '1*/dev/log', so trying to 'catf'
>>/dev/log results in:
>>
>>	Error: unable to open file "log" - system (or process) crash 
>>	prevented complete file close.
>
>The extensible streams documentation (Open Systems Toolkit) discusses the
>32-byte streams header which precedes user data on files of type 'uasc'.
>The message above indicates that the /dev/log file has a corrupted
>streams header.  Change it's name so that you can create a new one, and
>syslogd will revert to it's normal, unobtrusive self...

Because this was always such a nuisance (and for other reasons) at sr10.x,
the default type of file ("unstruct" or "nil") doesn't have a 32 byte
header and hence you tend not to see this problem any more.

-- 
                    -- Nat Mishkin
                       Apollo Computer Inc., Chelmsford, MA
                       mishkin@apollo.com