Dovecot terminates unexpectedly

Recently my dovecot IMAP server died regularly. In my log files I found the following message:

...
dovecot: 2008-08-31 09:12:50 Fatal: Time just moved backwards by 11 seconds. This might cause a lot of problems, so I'll just kill myself now. http://wiki.dovecot.org/TimeMovedBackwards
...

The wiki describes the cause - the clock of my server is not in sync - but does not offer much of help. Usage of ntp is suggested.

Openntpd syncs the system clock but does not slow it down (effectively causing the same problem). ntp offers to slow down the clock with side effects to database and CMS.

Examination of the logfile reveals that webmin runs a time sync just before dovecots unnatural death. OK, here's the cause.

...
Aug 31 09:13:01 myserver /USR/SBIN/CRON[12345]: (root) CMD (/etc/webmin/time/sync.pl)
dovecot: 2008-08-31 09:12:50 Fatal: Time just moved backwards by 11 seconds. This might cause a lot of problems, so I'll just kill myself now. http://wiki.dovecot.org/TimeMovedBackwards
...

Restarting dovecot right after the time sync will solve the problem. You can do that from the webmin user interface:

System/Scheduled Cron Job/Create a new scheduled cron job

enter

/etc/init.d/dovecot restart

select time right after /etc/webmin/time/sync.pl
and you are done.

Updating VMWare Server, MUI and Console

Updating VMware Server 1.0.5 to 1.0.6 works fine. However, installing VMware MUI afterwards breaks with VMware MUI 1.0.6.

VMware Server must be installed on this machine for the VMware Management Interface to work

Installation aborts.

In order to fix this I had to remove the library libgcc_s.so.1 in /usr/lib/vmware/lib/libgcc_s.so.1:

mv libgcc_s.so.1 libgcc_s.so.1_orig

Installation of MUI worked fine afterwards.

In order to start the management interface during the next reboot one has to add a few lines at the beginning of /etc/init.d/httpd.vmware:

RUNDIR="/var/run/vmware/httpd"
OWNER="www-data"
GROUP="www-data"

/usr/bin/test -d "$RUNDIR" || \
/bin/mkdir -p "$RUNDIR" && /bin/chown "$OWNER:$GROUP" "$RUNDIR"

Now, the management interface survives a reboot.

This workaround fixed the installation problem of the server console on the workstation as well.

Linux week in Vienna

May 15th to 17th, the Austrian chamber of commerce held their annual Linux weeks in Vienna.

Besides giving interested visitors the opportunity for hands on experiencing Linux, topics focused on

• Beginners and Switchers
• Linux in an educational environment
• Business and Government
  

I held a presentation together with my daughter covering our experience with (Ubuntu-) Linux and educational software. I did the theoretical part, while Paula gave a hands on demonstration of

• Tux4kids (TuxPaint, TuxType, TuxMath)
• GCompris

Here are some pictures of the event.

Overall, the Linux week was a great success.

Interesting Ubuntu blog

Searching the web for solutions to my problems I ran into this interesting blog.

I also would like to refer you to a blog that is maintained by a co-member of Ubuntu-Austria, Martin Lettner.

May the force be with you.

April 30th = Independence Day

On April 30th, 2008 I celebrated Independence Day.

I had switched my last application from Windows to Linux.

Free at last.

Distribution Upgrade 7.10 -> 8.04

As with the upgrade from 7.04 to 7.10:

1. sudo do-distribution-upgrade
2. keep the configuration files
3. accept deletion of outdated files
4. reboot

For a complete list of prerequisites visit my old article here.

There is an error in Webmin 1.410. It shows Apache2 server as being stopped. This is not the case as one can verify with ps afx | grep apache2. I will post a fix as soon as I found one.

Configuring VMware Server

In Kernel 2.6.23, a change was introduced that breaks vmware-server (a complete explanation can be found here):

Another solution is to get the vmware-any-any-patch116.tar.gz. I unpacked the tar and copied vmmon.tar and vmnet.tar into /usr/lib/vmware/modules/source.

vmware-config.pl configured the server (ignore the warnings, the server starts fine), and ran vmware-config-mui.pl.

Server is now up and running.

UMTS Modem in Ubuntu 8.04

Dial up internet connections are easy to set up, one might think. I wanted to connect to my ISP using a PCMCIA UMTS card. Searching the internet I found three different approaches:

1. use wvdial to set up a ppp connection
2. use gnome-ppp to set up a ppp connection
3. use a specific software (like umtsmon or vodaphone-mobile-link)that installs all required drivers

I run Ubuntu 64 on my notebook, so vodaphone does not work. There is an ancient x64 .deb package that refuses to install. The tar files don't compile due to missing twisted python modules. and finally (after installing twisted) it refused to start.

I found a much simpler solution: Gnomes network-manager.

network-manager allows you to configure 1 dial up connection. It requires some tweaking of a configuration file in order to reap the maximum speed from your modem but other than this, it is an easy solution without any bells and whistles.

Configuration of network manager

1. Left click your network icon in the top right of your Gnome panel. Select "Manual configuration..."
2. Unlock the dialog
3. Select "Point to Point connection" and edit its "Properties" (this brings up a 3-tabbed dialog)
4. On the General-Tab tag "Enable this connection" and adjust the connection type, access point name and account data.
5. Change connection type to "serial modem" to enter the phone number. Change back to GPRS/UMTS. The phone number gets stored in the configuration file
6. On the Modem-Tab type the modem port. In my case I hat to overwrite what was offered by the dialog ("/dev/ttyUSB0"). I found this information in the gnome-device-manager.
7. The Options-Tab allows to set the modem to provide the default route (which is helpful on the road) and to accept the DNS servers from your provider. I did not select "Retry if the ..." reconnect option here, but rather in the configuration file (as this sets it to retry permanently)
8. OK and close the configuration dialog and you are done.

Insert the card, left click the network icon in the panel. You can "Dial Up Connections>/Connect to ppp0 via Modem...". It takes a few seconds, then the connect is made.

There is not much of an indication as the line being up. On my modem, the LED is on permanently. Network-manager does not change its icon to reflect connection status.

Configuration myths and files

There are lots of rumors out. Here is what I found out:

• wvdial and other command line tools use /etc/wvdial.conf to configure dialup connection.
• gnome-ppp and network manager use ~/.wvdial.conf (in your home directory)

Both files use similar syntax but Gnome does not read /etc/wvdial.conf or setup configuration in /etc/ppp...

Here is my .wvdial.conf:

[Dialer Defaults]
Modem = /dev/ttyUSB0
ISDN = off
Modem Type = USB Modem
Baud = 460800
Init = ATZ
Init2 = ATQ0 V1 E1 S0=0 &C1 &D2 +FCLASS=0
Init3 = at+cgdcont=1,"IP","mynetwork"
Init4 =
Init5 =
Init6 =
Init7 =
Init8 =
Init9 =
Phone = telefonnumber_from_provider
Phone1 =
Phone2 =
Phone3 =
Phone4 =
Dial Prefix =
Dial Attempts = 1
Dial Command = ATM1L1DT
Ask Password = off
Password = mypassword
Username = myusername@mynetwork.domain
Auto Reconnect = off
Abort on Busy = off
Carrier Check = on
Check Def Route = on
Abort on No Dialtone = on
Stupid Mode = off
Idle Seconds = 180
Auto DNS = on
;Minimize = off
;Dock = on
;Do NOT edit this file by hand!

Despite the last line, the file can be edited by hand. network-manager honors changes in the file. I changed Baud from 57600 to 460800. This provides significant increase in performance.