Hi All,
I have essentially 2 questions.
I am looking for a more efficient way to match a
particular string starting at the bottom of an application
log file.
According to the Perl Cookbook "Reading a File
Backwards by Line or Paragraph", (8.4), you need to read all
the lines into an array in reverse order:
@lines= <FILE>;
foreach $line (@lines) {
# Do something with $line
}
This works fine. However, I generally have to look at 8-9
different log files, each at an average of 10 - 50
megs in size. Since I only want to match the first
occurrence of the particular string I'm looking for, I
decided against using "grep" but instead, something like this:
@lines =<FILE>;
foreach $line (@lines) {
if ($line =~ /Error/ ) {
print "Error Message is $line";
last;
}
}
This seems to work OK, except I have some cases
where the log files can reach 100-200 megs in size, and I
start to see a serious hit on performance.
So I says to myself, Kenny-boy, you don't need
the whole log, just grab the last 500 lines or so, then search on
that. But first in order to get the "right" log to search (there
can be several different ones for each application a day), I need
to get the latest log file. To do this a use the following
method:
$wanted_log = qx(ls -rt /usr/local/error_logs/app-*$log_date.log | tail
-1);
The log files have the following format:app-Month-ddhhmm.log.
This also works fine. Interesting enough, I thought "qx" was the same as
using back ticks `, but the back ticks do not interpolate the variable,
where the "qx" does.
This leads to my second problem. The next line of the script creates
the
smaller version of the log. Thus:
system ("tail -500 /usr/local/error_logs/$wanted_log >/tmp/half_log");
Instead of creating the half_log file, the output is printed to the
screen
(STDOUT) instead. I was using "glob" instead of "qx" above
to get the file name, but it was not returning the latest one by date,
only alphabetically. However, the system command was not printing
out to the screen.
If you made it this far, the questions again:
1) What is the most efficient or quickest way,
performance wise, to search from
the bottom of the file for the first occurrence of
a string.
2) Why does the system command print to STDOUT
when using "qx" in place of "glob" instead of
creating the file?
--
Ken Hammer
Information Technology Central Services
University Of Michigan
[EMAIL PROTECTED]
--
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
