Log Script

[heprox]

I'm working on a AIX 4.3.3 server running Oracle 8i. I need to create a script (preferably in Korn) that does the following:

1. Identifies files (log files and *.err files)in several directories, in a loop, that are older than 24 hours.

2. Places all of the identified files from step #1 and places them in a compressed tar file (with a file name in mmddyy format)in a specific directory.

3. Initiates an FTP session that moves the file to a remote server.

Step #3 is optional (I may do this in VB from the parent FTP server and just get the file in reverse).

Ideas:
Using "find $dirs -type f -name "*.err" -atime +1" or someting similar, but can't figure out how to designate what directories to look in (handled by a variable?) and how to move them through a loop. I'm thinking of just designating the directories and the filenames (to be looked for) by comma-deliminated variables? Any help would be appreciatted. Thanks.

 

[Annihilannic]

There's no need for a loop as find accepts multiple directories. So you could use something like:

[tt]#!/bin/ksh
DATE=$(date +%d%m%y)
tar cvf - $(find /var/adm /other/logs /more/logs \( -name "*.err" -o -name "*.log" \) -atime +1 -print) | compress -c > /tmp/logs_${DATE}.tar.Z
ftp -nv logdrophost << HERE
user username password
bin
cd /somewhere
put /tmp/logs_${DATE}.tar.Z logs_${DATE}.tar.Z
HERE[/tt] Annihilannic.

 

[heprox]

I'm using:

#!/bin/ksh
DATE=$(date +%m%d%y)
DATADIR=<data directories>
FILENAME=<filenames>
tar cvf - $(find ${DATADIR} -type f -name $ -mtime +1 -print) | compress -c > /tmp/logs_${DATE}.tar.Z

...but I continue to get empty archives. I'd like to be able to have two variables DATADIR and FILENAME and be able to just edit the script to add different directories and/or filenames (i.e. comma delminated?). In addition when I create this archive I would like the original(s) (*.err) to be deleted from the DATADIR location(s). I intend to either have the script set up an FTP session and send the file to a remote server or create a second script that sends the files every few days to a remote server. Any help is appreciatted. Thanks.

 

[Annihilannic]

&quot; -name $ &quot;? Isn't something missing there?

find won't understand -name filename1 filename2 etc. You need to use the brackets as per my example with -o -name between each filename.

Try getting the find command, i.e. the part between $( and ), to list the correct files on its own before getting the tar and compress parts working. Annihilannic.

 

[heprox]

Correct, I made a typo, sorry. I've got this working and it creates the archive correctly:

#!/bin/ksh
DATE=$(date +%m%d%y)
DATADIR=/gers/nurev/trc/temp
tar -cvf - $(find ${DATADIR} -type f -name &quot;*.err&quot; -mtime +1 -print) | compress -c > /tmp/DWMlogs_${DATE}.tar.Z

...now what I need is for it to be able to handle the filenames as a variable and then just define the variable as:

FILENAME=*.err:dwm*.* or sonething like that?

...then figure out how to delete the original files (*.err, etc) after it creates the archive.

 

[heprox]

OK this is working correctly:

#!/bin/ksh
DATE=$(date +%m%d%y)
DATADIR=/gers/nurev/trc/temp
FILENAME=&quot;dwm*.*&quot;
tar -cvf - $(find ${DATADIR} -type f -name ${FILENAME} -mtime +1 -print) | compress -c > /tmp/DWMlogs_${DATE}.tar.Z

...now how do you get the script to realize different directories nad filenames? I've tried:

FILENAME=&quot;dwm*.*&quot;:&quot;DWM*.*&quot;:&quot;log*.err&quot;

and

FILENAME=&quot;dwm*.*&quot;;&quot;DWM*.*&quot;;&quot;log*.err&quot;

and

FILENAME=&quot;dwm*.*&quot;,&quot;DWM*.*&quot;,&quot;log*.err&quot;

and

FILENAME=&quot;dwm*.*&quot; &quot;DWM*.*&quot; &quot;log*.err&quot;

...too no avail?

 

[danielhozac]

heprox,
Did you not read Annihilannic's post?
find won't understand -name filename1 filename2 etc. You need to use the brackets as per my example with -o -name between each filename. //Daniel

 

[heprox]

My fault, I didn't catch what he was trying to tell me. What I'm trying to accomplish is setting up a variable that I can add and subtract directories and filenames from at will without having to go and mess with the body of the script (easier that way...). Now I just have to find out how to remove the original files once they're archived. Thanks.

 

[heprox]

Well I settled on this:

#!/bin/ksh
DATE=$(date +%m%d%y)
tar cvf - $(find /gers/nurev/trc/temp /gers/genret/trc/temp \( -name &quot;dwm*.*&quot; -o -name &quot;DWM*.*&quot; \) -mtime +1 -print) | compress -c > /tmp/DWMlogs_${DATE}.tar.Z

...but I'm getting the following error:

/usr/bin/dwm_purge[3]: /usr/bin/tar: 0403-027 The parameter list is too long.

...any ideas?

 

[Annihilannic]

You might have to do it in two steps then; do the find on its own and redirect the output into a file, and then use tar -I tempfilename. Annihilannic.

 

[heprox]

Do you mean split the operations of the script? Conduct two seperate finds? I'm having difficulty getting &quot;find&quot; to lookl in different directories:

#!/bin/ksh
DATE=$(date +%m%d%y)
find /gers/nurev/trc/temp /gers/genret/trc/temp \( -name dwm.* -o -name &quot;DWManager.*&quot; \) -type f -mtime +1 -print

...finds files with both the &quot;dwm.* and &quot;DWManager.*&quot; names, but DOES NOT look in both the &quot;/gers/nurev/trc/temp&quot; and the &quot;/gers/genret/trc/temp&quot; directories. It only looks in the last directory named? Any help?

 

[Annihilannic]

Unfortunately I can't test on AIX as I don't have access to it anywhere.

I notice you're missing some quotes around &quot;dwm.*&quot;.

By two steps I mean something like this:

[tt]find a/ b/ \( -name &quot;dwm.*&quot; -o -name &quot;DWManager.*&quot; \) -type f -mtime +1 -print > /tmp/find.out.$$
tar cvf - -I /tmp/find.out.$$ | compress -c > /tmp/DWMlogs_${DATE}.tar.Z
rm /tmp/find.out.$$[/tt]

Specifying multiple directories like that works fine for me, maybe it's something peculiar to find on AIX?

Annihilannic.