RE: Bash error log clues?

At 05:52 AM 7/31/2000 -0500, you wrote:
Normally, I'd be looking for a for/do loop. And it's not necessarily on 
line 1 of the script, even though the error says so. In bash, the following 
are equivalent:

That's a hint which may help, thanks

>The "-c" is also a clue. Here's a thought: does either the 
>/etc/rc.d/init.d/httpd file or the apachectl script contain a -c switch on 
>the line that starts up httpd? If so, look at the file referenced and see 
>if there's something odd. (It could even be a line in the httpd.conf file 
>that's supposed to be commented but isn't.)

Can't find apachectl, in /etc/init.d/, but checked httpd and saw nothing 
amiss. The server's configs just get read once at startup, is that right?

>But this is the odd thing -- and I'm really wierded out by it.

You and me both, ate up most of the day looking into this. I was able to 
isolate a section of the access_log and correlate it to a section of the 
error_log and here's what I found:

In an approximately 20 minute period, the error was logged either 120 or 60 
times (depending upon whether the two-line error is one instance or two). I 
cannot correlate *anything* happening in the error_log with something 
occuring 60 or 120 times during that period in the access_log. No programs 
being called fit that bill, but the only programs *I'm* running are Perl 
scripts and none of them appear to be malfunctioning. Am I correct in 
assuming the:

sh: syntax error near unexpected token `;'
sh: -c: line 1: `;'

error *must* come from a shell script? If Perl scripts malfunction, they 
show time-stamped errors in the error log, at least that's my experience. 
These errors are not time-stamped and are all like the above. I'm going to 
check the cron log for that period of time and see if I can get any 
correlation there. If not, I'm out of ideas.

>You might also try posting the question on the bash list 
>(bash(at)moongroup.com) -- see http://www.moongroup.com for subscription 
>information, if you're not already subscribed.

Will do, thanks. If I could just force the shell to time-stamp its errors, 
that'd be some help.
>Best,
>Scott Temaat, President/CEO

HWG: hwg-servers mailing list archives, maintained by Webmasters @ IWA