Shell Programming and Scripting

All,

Is there some way to have my KSH script pause, until the word "DONE!" appears in the logfile /logs/log.txt?

Difficulty: The word "DONE!" may already be present in logs from the past. So I anticipate tail -0f may (somehow?) need to be done in the background, instead of grepping in a loop?

Thanks so much,
CG

would it work to have a counter for the string "DONE!" that is initialized to 0 at the start of your script and then enter a loop containing a sleep statement and then after the sleep statement expires count the number of "DONE!" instances to see if it has increased from previous count.

if it has not increased, loop back for another nap (sleep)

if it has increased, run a conditional code path that will increment your DONE counter by 1 and run anything else before it's time to sleep again?

biker
LiveFire Labs: Online UNIX Training with Hands-on Internet Lab
Online UNIX and Linux Training with Hands-on Internet Lab

Good idea, but unfortunately I can't use a sleep/counter anymore

I am actually currently using a sleep/counter now, but I need to now change it to something that monitors the logs.
...reason being is that the background process that generates this logfile has changed, and the DONE! can happen anywhere from 40 seconds to now 7 minutes.

Although I can technically sleep 7 minutes, because I run this script over and over, it'd be unfortunately way too time consuming if the "DONE!" actually occurs in 50 seconds, for example. -- other things do show in this logfile too, so I can't check for a change in the logfile size either.

What if you just sleep 10 seconds or other short interval, look at the log file, sleep 10 seconds, etc.?

i'm not sure i follow, but i'll probably need to take some time to think about your response in order to ask some relevant follow-up questions. one thing i don't understand is why the script is run "over and over" vs starting it once and letting it run.

in the meantime, have you looked at or considered co-processes/co-routines?

if you're not familiar, here are a few older pages related to the topic...

Co-processes - Part I - Review of Foreground and Background Jobs

Co-processes - Part II - Introduction

Co-processes - Part III - Syntax and Simple Example

biker

I simply suppose you want something like

Code:

tail --retry -f log | while read line
do
grep -q "LOG" 
counter++
done.

or

you may try a simple combination of "tail -f logfile|egrep LOG" and then check the count of output from the command

I suppose I should elaborate:

What I'm modifying is a KSH automation script, that's run on demand, for testing a piece of software.

What this KSH script does/need to do: (Step 2)

1) Shutdown, prepare, and restart the software under test (already scripted)

2) Wait for a message (we'll call "DONE!") to appear in /log.txt that says that the MYSQL DB has bound to the RMI/Software.
a. This can take anywhere from 40 seconds to 7+ minutes, depending on DB size..
b. While I can put a 7+ minute delay there, it would be much better if this KSH script continued "as soon as" the "DONE!" message appears in the /log.txt. *This is the challenge.
c. I cannot simply grep the file for "DONE!" in a while loop, because I cannot delete "old/past" logs in this same file that may already contain this word.
d. I also cannot simply check in a while loop for the filesize to change, since other logs are being appended to this same log file during this wait time.

3) Continuing on with already automated tests...


Apologies I wasn't as detailed the first time around!
Thanks so much
CG