uninstall not clean

I tried to install Standard Server on a CentOS 5.5 machine. When running the setup it had a problem with clustering. I don't know why it tried to install clustering, but however after two howers I aborted that dialog box. The setup continued but the server did not run correctly (I could not connect to it). So I started the uninstaller. I removed manually PostgresPlus directory including all subdirs. I restarted the box.

I try to reinstall Postgres Plus Standard Server now, unfortunately with no success! The setup says that it found an existing installation in /opt/PostgresPlus/8.4SS!! Don't know where that information comes from, the directory not even exists nor is any process named postgres running! Consequently a bit later the setup fails with the message 'Error reading file /opt/PostgresPlus/8.4SS/data/postgresql.conf'.

The question is: Why does the installer say that there is an existing installation? Is there anywhere some information I need to delete?

Any help would be greatly appreciated.

Thomas

Hi Thomas,

This information is maintained in the registry (/etc/postgres-reg.ini in case of centos). For proper uninstallation you need to run the uninstaller executable and it will remove all related files and registry. Since you have manually removed all the files, now you need to remove your standard server related entry from the registry and then you can run installer to do the fresh installation.

Please let us know if you need further help in this regard.

Thanks a lot, after removing the registry file I could start the installation again.

However the first problem remains. Setup hangs with a dialog saying : Initialising the database cluster (see attached screenshot). It does not stay there forever. If I close the dialog setup continiues but looks like it is incomplete. This means I cannot install Standardserver on my CentOS box.

By the way: I did run the uninstaller but probably due to an inproper install it did not remove everything.

Any help would be greatly appreciated.

Thomas

Can you please send installation log which can be found under /tmp. Please remove other files before starting a fresh installation which might exist in /tmp to make sure that correct file is sent. File name would be install-postgresql.log.

OK please find enclosed the file. There are lots more files, I send only the one you requested.

If it is of interest for you:

I did NOT install Slony and PostGIS.

My CentOS runs as a VMWare virtual machine.

Thanks a lot for your help!

Thomas

log shows that installation could not proceed after executing:

Initialising the database cluster (this may take a few minutes)...
Executing /opt/PostgresPlus/8.4SS/installer/server/initcluster.sh "postgres" "postgres" "sNsWIGsy77p7" "/opt/PostgresPlus/8.4SS" "/opt/PostgresPlus/8.4SS/data" 5432 DEFAULT

can you please run following command in a shell and share the output:
/opt/PostgresPlus/8.4SS/installer/server/initcluster.sh "postgres" "postgres" "sNsWIGsy77p7" "/opt/PostgresPlus/8.4SS" "/opt/PostgresPlus/8.4SS/data" 5432 DEFAULT

At least on my disk is no directory called installer under 8.4SS!

OK, the result is:

/opt/PostgresPlus/8.4SS/installer/server/initcluster.sh: Befehl nicht gefunden.

I run a German CentOS, it means: Command not found.

Enclosed the content of my .../8.4SS directory.

At what stage of installation you are running the initcluster command.

Please run it in a shell when you are getting dialogue box related to initializing database cluster.

OK I could start it an enclose you the output. (I am now out of office for the next arount four hours and will read messages after that).

I removed the entire tree of directories before I started setup. I is possible, that the GUI installer already created the database file so that the second call to the shell script from the console produced this error message?

Thanks for your help!

initdb: directory "/opt/PostgresPlus/8.4SS/data" exists but is not empty
If you want to create a new database system, either remove or empty
the directory "/opt/PostgresPlus/8.4SS/data" or run initdb
with an argument other than "/opt/PostgresPlus/8.4SS/data".
Failed to initialise the database cluster with initdb

actually error message in the log suggests that data directory already exists and initdb is failing because of that. It may be because of installer's initdb process might have created that directory. So you may have to remove data directory before executing initcluster command if you want to check any other reason which might be causing your installation to fail.

This setup is really cumbersome. I removed everything inside the data directory. I called the shell script again. It completed without error (see attached file). However it seems missing a lot. there is no bin directory under 8.4SS so I cannot start the server. It seems the libraries are missing too. What is going wrong here? Is there any other way to install it? This 'one-click' installer seems not to work on CentOS at all!

Hi Thomas,

It is weird that initcluster command is running good when you are running it manually and fails when it is run by installer.

It is expected that there is nothing in bin and lib folder, when you cancel the installation after it hangs, component is uninstalled and it removes all the installed files as you can see from the log.

I will try to install the said installer at our end on centos and will check if it is failing.

Ok, thats great.

I am running CentOS 5.5
Kernel Linux 2.6.18-194.11.1.el5 on i686

latest updates on all packages.

It is 32-bits

2 GB RAM running on VMWare Server 2.0

Thomas

Are there any updates on this issue?