Stuart Woodward's questions

I have found that a significant number of users of a system have set an email address addresses to an address with the top level domain .test. As a interim measure, I don't want to attempt to deliver these mails at all and I just want to silently dump them to the mail equivalent of /dev/null.

Is there way to configure exim to just junk these recipients?

(I need to educate people not to do this and not to choose some other random email address when QA testing but in the mean time I want to prevent delivery attempts for all recipients at this junk tld.)

Is there a standard way to know if a shell script is running under cron as opposed to running from the command line. i.e. I'm thinking of running with more verbose output when manually kicked from the command line and enabling some features that only run when under cron.

When a server (say Windows, Linux) is rebooted, it will take some time to respond to ping.

I'm assuming that the software firewall has to be up before pings will be returned as there might be a setting to disable ping responses.

Everyone knows that Windows and Linux have have totally different architectures so lets treat them separately.

The answer I'm looking for is "After XXX is running, pings will be returned." It would be helpful to know where in the boot order this is too. i.e. at the start or end.

I ask because we get questions from a customer about why it takes so long to respond to ping after creating a Virtual Machine. I'm sure this is just an artefact of the OS boot behaviour.

What is the strategy for detecting time drift in all linux based data centre? This is a more difficult problem than it seems at first.

Time drift can cause serious problems for certain applications and often, even though NTP is installed, it's possible to fail for the following (and many more) reasons:

• NTP was not correctly set up to automatically restart on reboot.
• The settings on a server are incorrect so the time server it points to is unreachable or inaccurate.
• The master time server is unreachable and all servers are syncing with it are now syncing to an unreliable source.

I would like a way to detect if all the individual servers are correct. Bear in mind that the server with the testing script/application may not be right.

I want to have limited access to a Linux server to be able to grab a certain set of logs. Ideally, I would like to be able to run rsync locally to grab the logs from the server and for security reasons I don't want to give the user full ssh login access.

I have two servers, a production server and a cloned image of the production server. There is some difference in the cloned image that is preventing it from working. I'd like to have a list of the differences.

The servers are on different networks so I can't easily compare them. What I need is a something which will create a database of the checksums on the production server and clone server and compare them.

Is there already something that can do this? Or is it easier to write a script to do this?

I have a virtual image of a WAS server but it can no longer access the LDAP/AD that holds the administration password as it has moved to a different network. As a result, I can't login to change the settings.

I want to change the IP of the address by hand or turn off security so that I can login and change it by hand.

I knew a way to do this once...

I have a deep and complex file system where some files have been accidently written by root. I want to change the ownership of those files back to the original owner in one go.

I am playing with commands like:

find /folder -type f | xargs ls -l | grep "root root" 

but there is a lot of garbage coming out too.

I want to make a list first and then change only the files in that list after confirmation.

Can the encryption option on Info-ZIP be used from the command line? The zip man page says:

-e --encrypt Encrypt the contents of the zip archive using a password which is entered on the terminal in response to a prompt (this will not be echoed; if standard error is not a tty, zip will exit with an error). The password prompt is repeated to save the user from typing errors.

Is there a way to create zipped files from an unattended script?

After a reboot, some partitions which were mentioned in fstab were not mounted as expected. The format of the line for the partitions which were mounted correctly and those which were not mounted correctly look the same so I am wondering whether some log exists of any problems which prevented the restoration of the missing partitions.

I am not able to see the console during a reboot but need to determine and fix the problem later.

On a server we have discovered that we need to add the FTP passive command in every users .netrc file when we create a new user.

Can this and other defaults (in .something files) be automatically created when a new user is created?

I'm migrating from a Solaris server with Perl 5.6.1 to a Redhat Linux server with Perl 5.8.8 .

My stance is that we keep the newer version of Perl and do not downgrade to the older version to avoid compatibility issues with any scripts as the compatibility issues with downgrading the OS default install would be worse.

My assumption is that Redhat Linux could be dependent on Perl somewhere and if the older version is necessary we should explicitly make a named binary perl561 so that users can choose to use the older version if they need it.

What options exist to compare the performance of a Solaris 5.9 and a Linux (Redhat RHEL 5.X) Server?

I have aging Solaris server and a replacement Linux server hosted on VMWare. A key success factor in the migration is improving the performance of a disk i/o intensive batch process. The process in question consists of multiple processes writing simultaneously to disk so this should be reflected in the test.

The tool required must be able to run from the command line on both severs as is.

Options under consideration:

IOMeter - NG. looks like it is a graphical application. Bonnie++ - looks like a candidate, but I'm not sure if it can be compiled on both systems.

Recently, I realized that a certain disk partition on a Linux Server that is only accessed by ssh was full. I tracked down the problem to a VNC log which had filled up the disk. Using lsof, I determined that even though the log was deleted and VNC was restarted the disk space was not returned because various processes related to the window manager had the files opened (nautilus, gnome-ses, gnome-key, metacity, gnome-pan, eggcups, bt-applet, nm-applet, pam-panel, escd, mapping-d, gnome-ter).

The suggestion from the sysadmin is to reboot the server. Actually, the GUI has only been used once to install the packaged software that the server runs so is not required for most of the time. I would rather fix the problem without rebooting if possible.

How can the GUI be restarted or even turned off gracefully from the command line to be restarted later when required?

I'm looking for a good web based portal to install on internal production servers to give information about the health of the server to people who have no login permission for that server at the command line. Typical information would be disk capacity left, processes running status, capacity graphs for network, disk and CPU.

Ideally, the portal would support installation on both Linux and Windows servers.

Occasionally, we run out of rollback segment space causing the query/update to fail. What are the best practices that I can expect from an Oracle DBA to ensure that this is monitored pro-actively and the appropriate measures taken before trouble occurs?

Why would a heavily disk intensive application run faster on a SAN than on a Physical Disk? I would have expected the Physical disk to be slightly faster but in fact the process ran 100 times faster when it's work drive was set to a partition on the SAN.

Our guess is that the SAN is optimised out of the box to be fast whereas the physical disk tuning settings are OS (Solaris) related and have not been touched or the OS patched.

During the highest activity the disk I/O was running at 100% and the time to complete a write was over 2 seconds as several processes were writing to the disk at the same time.

(FYI the application involved was Informatica PowerCenter)

I have the feeling that there are some difference between two large complicated Oracle schemas which should be identical, so decided to dump and diff them to investigate. I found an article (http://www.dba-oracle.com/art_builder_get_schema_syntax.htm) which gives the following commands to dump the table and index definitions.

SELECT DBMS_METADATA.GET_DDL('TABLE',u.table_name) FROM USER_TABLES u;
SELECT DBMS_METADATA.GET_DDL('INDEX',u.index_name) FROM USER_INDEXES u;

However, when I dumped and diffed spool of the output the schemas, there were trivial differences on almost every line making the difference worthless. For example the schema name is included in each table definition.

Is there better way to do this that will give the most important schema data in a format that can be compared easily using a diff tool. I am interested in tables, indexes and triggers.

I am using Oracle Database 10g Enterprise Edition Release 10.2.0.3.0

I have been a Linux user on and off for many years. Recently, I have been using it daily. I am looking for suggestions for media; commercial books, free information, online videos etc that will help me to understand the OS at a much deeper level. Your suggestions please.

I have two servers that have should have the same setup except for known differences.

By running:

find / \( -path /proc -o -path /sys -o -path /dev \) -prune -o -print | sort > allfiles.txt

I can find a list of all the files on one server and compare it against the list of files on the the other server. This will show me the differences in the names of the files that reside on the servers.

What I really want to do is run a checksum on all the files on both of the servers and compare them to also find where the contents are different. e.g

find / \( -path /proc -o -path /sys -o -path /dev \) -prune -o -print | xargs /usr/bin/sha1sum

Is this a sensible way to do this? I was thinking that rysnc already has most of this functionality but can it be used to provide the list of differences?