RE: Question Re: Piping and Redirection

Hiya Loyd

I'm still mystified as to why you would want to watch the log files. Seems
to me that constantly watching them would be more disruptive than getting a
break message  - especially when you can't easily filter the log by a
severity level. Or can you ?

I guess you could write a small script to do some filtering, but AS/400
already has this via severity level parameter and you could write a script
there anyway too.

I presume the monitoring of the SQL log is to check progress on a  job so
the equivalent on the AS/400 would be to just check the job log after
locating the job by wrkactjob. Slightly different access to visibility I
guess but much the same result. Same with watching the processes to monitor
jobs completing (my example was intended to appy where there was a
job  that you needed to be alerted the instant it completed not as any kind
of generic procedure)

The approach I tend to take on the AS/400 is to watch nothing but monitor
everything. The critical message queues are monitored for exceptions,
critical jobs are monitored - I have a home grown monitoring system) and
any exceptions to the expected parameters will interrupt me etc etc

This seems to be - at least to me - one more demonstration or symptom that
unix is a much more hands on environment than the AS/400 at a basic level.
This is not necessarily a bad thing - I fondly remember the 38 environment
in the early 80's being kind of similar regarding the level of involvement
it demanded :)

Anyway, I'm not trying to shoot holes in your examples - I appreciate you
providing them and it makes a bit more sense to me than it did.

Regards
Evan Harris

Tail would be a shoehorn in OS/400 and /QSYS.LIB since virtually everything
is a database file. I think tail should stay in the IFS with "real" stream
files.

I imagine the predominant use of tail in a unix environment is to watch log
files. For instance, I "tail -f /var/log/messages" to see new system
messages my Linux system generates. I also "tail -f
/home/mysql/log/mysql.log" to see which SQL statements are being processed.
Both of these files are text files, and tail gives me the flexibility to
view the most recent activity of these files without interrupting my work
(break messages).

I don't necessarily see much use in tailing a database. On a database, I
would rather attach a trigger to it, even though most triggers are tailored
to the table in question. Tail is more of a generic tool, and could be
considered a "poor man's" trigger in the right circumstances.

Could tail be used to monitor when a processes ends? I suppose, but I'd
rather have a /var/run/program.pid file. This file contains the process id
for a running program. When a program ends normally (not kill -9), the .pid
file should be removed automatically.

Loyd