Manual Installation Error

[drakeu]

Hello,

When running the 1-prereqs script I receive an error part way through the installation which says in part:

Warning: 3.0.x versions of yum would erroneously match against filenames.
You can use "*/perl(DBD::Sybase)" and/or "*bin/perl(DBD::Sybase)" to get that behaviour
Loaded plugins: protectbase, rhnplugin, security
Error: Need to pass a list of pkgs to install
usage: yum [options] COMMAND

This occurs during the package "perl-Net-SMTP-TLS.noarch 0:0.12-1.el5.rf" install while installing the dependency packages "perl-IO-Socket-SSL.noarch 0:1.01-1.fc6" and "perl-Net-SSLeay.x86_64 0:1.30-4.fc6".

I should point out that we are running the yum portectbase plugin to keep the core RHEL packages from being overwritten. Could this be the cause of the error (a certain package not being installed earlier due to a protected repository)? We would like to keep the protectbase plugin in use if possible, as it is our policy to keep core RHEL packages from being overwritten. We are running RHEL 5.4. Thanks!

[mmestnik]

This simply indicates that DBD::Sybase is not packaged in any of your repositories. Is DAG configured and working?

You can install this package any way you see fit, it's only essential for the MS-SQL check commands so you can continue your install without it.

[admin]

I would recommend installing XI on a 32-bit system if possible. We've seen some situations where 64-bit packages are not available to meet all the requirements of XI.

[drakeu]

We reloaded our RHEL 5.4 server to the 32-bit version. When running the 1-prereqs script we receive the following errors fairly early in the script.

no package perl-CPAN available ... Nothing to do
no package php-mcrypt ... Nothing to do
no package php-mssql ... Nothing to do

This leads to several other errors later in the installation script due to missing dependencies. Incidentally, these are the exact same errors we were receiving when trying to install on the 64-bit version of RHEL 5.4.

[mmestnik]

Firstly I'd like to know more about what is the same when trying to install on the 64-bit version of RHEL 5.4. You mentioned portectbase and that you were having a problem installing perl(DBD::Sybase). perl(Net::SMTP::TLS) and perl(IO::Socket::SSL) are installed one after the other with separate calls to yum.

Now the problem you are having is with a previous step where a call to yum installs at least a dozen pkgs in one go.

Once again I question your configuration of DAG, did you forget to run ./0-yum?

As for portectbase I'm not going to say that you can't use it but we don't know anything about it, so it's up to you to ensure it's not a problem. All that's needed from 1-prereqs is the rpm's installed the perl modules and if I'm not mistaken there is one python task that needs doing. Once these are satisfied you should be able to run the rest of the install.

[drakeu]

I manually download the php-mcrypt and php-mssql RPMs from pbone.net and installed them. I then turned protectbase off to ensure all the latest packages from DAG would be downloaded. Re-running the 1-prereqs scripts now yielded the following errors:

No package perl-CPAN available.
Nothing to do

Transaction Check Error:
file /usr/share/man/man3/Getopt::Long.3pm.gz from install of perl-Getopt-Long-2.38-1.el5.rf.noarch conflicts with file from package perl-5.8.8-27.el5.i386

Setting up Install Process
Package perl-Mail-POP3Client-2.18-1.el5.rf.noarch already installed and latest version
Nothing to do
Loaded plugins: protectbase, rhnplugin, security
Error: Need to pass a list of pkgs to install
usage: yum [options] COMMAND

I don't think any of these are fatal. I have continued with the installation scripts and made it to the B-installxi script. Initially when I ran it I received the following errors:

Creating product directory...
Adding password for user nagiosadmin
-bash: /tmp/xi/nagiosxi/create_postgres_db.sh: /bin/sh^M: bad interpreter: No such file or directory
Stopping postgresql service: [ OK ]
Starting postgresql service: [ OK ]
psql: FATAL: database "nagiosxi" does not exist
psql: FATAL: database "nagiosxi" does not exist
chown: cannot access `/usr/local/nagios/etc/hosts/*.cfg': No such file or directory
chown: cannot access `/usr/local/nagios/etc/services/*.cfg': No such file or directory

Turns out the create_postgres_db.sh script had dos eols so I ran it through dos2unix and reran the B-installxi script. Now I am receiving the following:

Creating product directory...
Adding password for user nagiosadmin
CREATE ROLE
CREATE DATABASE
Stopping postgresql service: [ OK ]
Starting postgresql service: [ OK ]
SET
ERROR: parameter "standard_conforming_strings" cannot be changed
SET
SET
SET
ERROR: must be owner of schema public
SET
CREATE SEQUENCE
ALTER TABLE
CREATE SEQUENCE
ALTER TABLE
CREATE SEQUENCE
ALTER TABLE
CREATE SEQUENCE
ALTER TABLE
CREATE SEQUENCE
ALTER TABLE
CREATE SEQUENCE
ALTER TABLE
CREATE SEQUENCE
ALTER TABLE
SET
SET
CREATE TABLE
ALTER TABLE
CREATE TABLE
ALTER TABLE
CREATE SEQUENCE
ALTER TABLE
ERROR: syntax error at or near "OWNED" at character 39
LINE 1: ALTER SEQUENCE xi_events_event_id_seq OWNED BY xi_events.eve...
^
CREATE SEQUENCE
ALTER TABLE
CREATE TABLE
ALTER TABLE
CREATE SEQUENCE
ALTER TABLE
CREATE TABLE
ALTER TABLE
CREATE SEQUENCE
ALTER TABLE
CREATE TABLE
ALTER TABLE
CREATE SEQUENCE
ALTER TABLE
CREATE TABLE
ALTER TABLE
CREATE SEQUENCE
ALTER TABLE
CREATE TABLE
ALTER TABLE
ALTER TABLE
ALTER TABLE
ALTER TABLE
ALTER TABLE
ALTER TABLE
ALTER TABLE
ALTER TABLE
ALTER TABLE
ALTER TABLE
ALTER TABLE
CREATE INDEX
CREATE INDEX
CREATE INDEX
CREATE INDEX
CREATE INDEX
CREATE INDEX
CREATE INDEX
CREATE INDEX
WARNING: no privileges could be revoked
REVOKE
WARNING: no privileges could be revoked
REVOKE
WARNING: no privileges were granted
GRANT
WARNING: no privileges were granted
GRANT
SET
ERROR: parameter "standard_conforming_strings" cannot be changed
SET
SET
SET
SET
setval
--------
20
(1 row)

setval
--------
16
(1 row)

setval
--------
25
(1 row)

setval
--------
15
(1 row)

setval
--------
14
(1 row)

setval
--------
142
(1 row)

setval
--------
53
(1 row)

setval
--------
21
(1 row)

setval
--------
33
(1 row)

setval
--------
28
(1 row)

setval
--------
170
(1 row)

setval
--------
25
(1 row)

INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
INSERT 0 1
chown: cannot access `/usr/local/nagios/etc/hosts/*.cfg': No such file or directory
chown: cannot access `/usr/local/nagios/etc/services/*.cfg': No such file or directory

[mmestnik]

-bash: /tmp/xi/nagiosxi/create_postgres_db.sh: /bin/sh^M: bad interpreter: No such file or directory

This indicates this file was mangled by a text translation... It needs to be copied as a binary file, well not now. Now this file needs to be run through 'dos2unix'.

I'm curious of the history of this file. Was it unpacked under Windows and then copied over? It looks like the Windows untar program converted the file to a Native text document and then SFTP'ed over as a binary file.

[drakeu]

As I mentioned in my last post, I already ran create_postgres_db.sh through dos2unix and then reran the B-installxi script. What I am interested in getting an answer to now is the error messages I received when the script was setting up the database. I am also curious about the "file missing" messages at the end.

FYI - I downloaded the xi tarball with wget on the Linux box to which I am attempting an install. I untarred the file with the tar command. None of the files have been transferred to or from any other machine.

[mmestnik]

That's likely because these scripts arn't completely idempotent and once fail they may fail successive calls. Though this could be a larger dos2unix issue.

Using this to track the issue.
http://tracker.nagios.com/view.php?id=24

[drakeu]

Not seeing any movement on the bug report in Mantis, I went ahead and ran the rest of the installation scripts to complete the XI setup. I then imported all of the configs from our Nagios Core installation into XI. No errors were reported during the import process. Several users have been setup to correspond with the users we had in Core. So far we have not seen any error messages or functions that do not work. However, we are still concerned about the stability of our installation given the errors we received when setting up the database. Is there anything we can do to just look at the database configuration and correct any errors?