Removing old version of cloverleaf

[Gary Atkinson]

We upgraded to 5.5 some time ago.  We still have previous version of 5.2 directory structure on the server and I would like to regain the disk space.  What is the safest way to remove the 5.2 version?

thx,

Gary

Topic

[Robert Kersemakers]

Hi Gary,

You have followed the same upgrade path as we did.

If you are certain no 5.2 processes are running, then just delete the complete …/qdx5.2 directory. Of course you made one or more backups just in case. I did, but never needed them.

Can’t see any danger in this. Oh: make sure none of your 5.5 processes is using a 5.2 directory.

Zuyderland Medisch Centrum; Heerlen/Sittard; The Netherlands

[Gary Atkinson]

Thanks Robert.  I checked in the process directory and I did not find any pid or cmd_port files.  We are on AIX.  What is the command to backup a directory and then delete it?

[Robert Kersemakers]

We are on HP-UX, so I can’t really help you with the commands.

I always use a script (from our reseller) to make a backup of a site. It first makes a .tar file of all the important directories of that site with ‘tar’, and then that .tar file is compressed with ‘compress’.

Deleting a directory (and all its subdirectories) is easy: rm -R .  But be careful with this command (if it works at all on AIX).

Maybe someone with AIX can help you further.

Zuyderland Medisch Centrum; Heerlen/Sittard; The Netherlands

[Russ Ross]

Before I get to the point of doing the actual delete of an entire directory tree of what I believe is an obsolete version of cloverleaf running on an active server, I make the directory logically invisible to anything that might still be using it.  Most of the time I discover something is still using it.

For example:

mv   /quovadx/qdx5.2   /quovadx/qdx5.2_hide

Russ Ross
RussRoss318@gmail.com

[Gary Atkinson]

I forgot to add that one point Ross  😳 While back when we upgraded we changed the path from /qdx/quovadx/qdx5.2 to /qdx/quovadx/oldqdx5.2.  Yesterday we copied the /oldqdx5.2 directory tree to an archive directory; we had to use cp -R, because we found a fifo file.  I watched the system for awhile and no problems occurred.  We then did an rm -r on the directory.  We gain almost 2 gig space back!  This may not seem like a lot, but hard drive space is “hard” to come by here, in fact our production server is only 24 gig  😯 Right we are sitting at about 59% free space.  One of the other programmers here has about 8 gig on the qdx mount point and he is going to make a symbolic link to our archive directory.  After that I hope we are more around 50% free space.  My issue has also been, they give me more interfaces to build, but not more space  😈  

Thanks all for the help!!

[Robert Kersemakers]

I must remember that: always rename the old directory first.

And we only have 7 GB on our production (and test) server… We had a change some months ago, where this space should have been enlarged to 30 GB. But still nothing. And too busy too make much work of it now. Need to have a word with a colleague about that (and then file a call…)

We are on 64% space in use; will drop down to 57% when recycling SMAT files in the weekend.

Zuyderland Medisch Centrum; Heerlen/Sittard; The Netherlands

[Gary Atkinson]

Yes, SMAT files are the biggest use of space on our servers.  Right now I keep them for 18 days and then purge them.  I also, got rid of some xlates I was not using anymore, but that didn’t even make a dent  😆

[Russ Ross]

I don’t have tight space anymore with our current hardware but there was a time it was so tight an upgrade was a real challenge.

Once upon a time I had maybe 8 gigabytes for all the cloverleaf stuff.

Fortunately, these days I have at least 200+ GB’s on all our cloverleaf servers.

Here is a script I wrote and used to make an exact copy of an entire directory tree.

I initially worte this script to help a system owner migrate everyhting form an old disk drive to a new disk drive which included an entire oracle data base.

The oracle data base was huge and oddly enough the system admins installed the new hard drive but refused to help migrate any of the file systems.

The system owner came to me desparate for anyone that might be able to help since he heard I administer my own servers.

He was so desparate he didn’t even hold me responsible if I screwed up so what a great learning opportunity.

Now the system admins use this script I wrote all the time after they saw it did the job they were afraid or unwilling to attempt.

The solution I found somewhere on the internet I believe, turned out to be an interesting solution because the script did not use much memory to run even though what we were moving was huge.

Ever since I’ve been using tar_tar_dir.ksh for myself to copy cloverleaf sites and various other directory trees.

It assumes the source and destination directory roots both exist so you may have to do a mkdir to create the destination directory first.

I called the script tar_tar_dir.ksh and here it is:

Code:

#!/usr/bin/ksh

# Begin Module Header ==============================================================================
#
#——
# Name:
#——
#
# tar_tar_dir
#
#———
# Purpose:
#———
#
# Copy an entire directory tree with all files and links to another base directory
# using the tar command
#
# Notes:
#——-
#
# This script is nice because it pipes the output of one tar command to another,
# which requires a little bit of temporary memory.
#
# Example of normal usage:
#
# cd $HCIROOT
# tar_tar_dir ./test_picis /3.5.2P_picis
#
#———
# History:
#———
#
# 1999.10.29 Russ Ross
#          – wrote initial version
#
# End of Module Header =============================================================================

(cd $1 && tar -cf – .) | (cd $2 && tar -xBf -)

Russ Ross
RussRoss318@gmail.com

[Author]

Replies