Webserver misbehaving

Jun 10

On Monday, May 29th, there were no more mails coming in. A look into the logfiles showed a lot of strange errormessages, about no more memory allocatable, but neither the swap-partition nor any of the other partitions were anything near full.

"top" was reporting pretty heavy system load, and sendmail was holding unusually much meOn Monday, May 29th, there were no more mails coming in. A look into the logfiles showed a lot of strange errormessages, about no more memory allocatable, but neither the swap-partition nor any of the other partitions were anything near full.

"top" was reporting pretty heavy system load, and sendmail was holding unusually much memory, strange though I couldn’t stop (kill) the sendmail processes. So I tried to stop the webserver (apache), which worked, but when I tried to start it again this failed.

So finally I decided to restart (init 6) the whole server, but it wouldn’t shut down. So I filled out the service-form of my hoster with a software reboot-request. Now the server shut down, but apparently it wouldn’t come up again.

So next reboot request-form filled out, now hardware reset. Still the server didn’t come up again, but instead there was all of a sudden a reinstall-request pending, which I didn’t send. And since I was out of the country I couldn’t reach the service-hotline (0-900-…).

Then about 4 hours later the server was up again, but what was that? A totally new and "virgin" machine:

new (and better) hardware : cool

newer distribution of the linux-system : cool, ’cause the old one wasn’t supported anymore anyway

But now the problems just started:

Now apache2 was running on the new server, and the config-files (which I fortunately backed up – like almost everything else – onto another server reachable over the internet) were not 100% compatible, so I had to do some adjustments there.

After reinstalling the typo3-sources (version 3.8.1) I found out that installing extensions globally without backing up at least the extensions directory isn’t the best idea, so I had also the install the missing extensions again, hoping not to hit the problem of installing a too new version (not downward compatible)

So finally I got the webservers basically up-and-running again.

Now I had to reactivate the email-stuff.

But whoa, this turned out to be a bit too much for me.

The old server was running with sendmail and qpopper, the new server now uses postfix and courier. So the old configuration was good for nothing.

After some fiddeling I got it so far that it was receiving mails again, but now they landed in Maildir of the email-user instead of the mailbox-file in /var/spool/mail, and even worse

But worst of all neiter could I log into the system to retrieve the mails via pop3 nor could I get the system to accept mails from my mailclients for delivery

Well, I thought, so if I modify the records in the configuration tool confixx, it might rewrite the correct configuration files, but whatever I tried nothing happend. Finally I found out, that even though all the changes were written into the database, the scripts which should take the information out of the database and write the corresponding data into the configuration files weren’t running, because when reinstalling my server it was forgotten to (re)activate the confixx-license.

So, in the meantime it was Friday, June 2nd I wrote a complaint-ticket about that. But until Tuesday, June 6th, when I was back in the country and could call the expensive 0-900-number nothing happened. When I finally called them, the licence was reactivated within minutes. And mail was almost running fine, except the home directories of the mail-users were different than I installed them. So all my procmail-setup which actived two spam-filters was ignored, when I moved the setup over into the new home-directories, the mails were again delived into the /var/spool/mail/ – mailboxes, instead of the new Maildir so courier in turn couldn’t deliver any mails. Finally I also figured this out too…

Now I realized that my webservers weren’t running anymore. After short investigation I found out that the confixx-scripts had overwritten my apache2-vhosts configuration with the – for my setups – unusable confixx-standard-configuration-settings. So I had to find a way around that as well.

Now since Wednesday, June 7th, it seems everything is working normally again.mory, strange though I couldn’t stop (kill) the sendmail processes. So I tried to stop the webserver (apache), which worked, but when I tried to start it again this failed.

Then about 4 hours later the server was up again, but what was that? A totally new and "virgin" machine:

new (and better) hardware : cool

newer distribution of the linux-system : cool, ’cause the old one wasn’t supported anymore anyway

But now the problems just started:

So finally I got the webservers basically up-and-running again.