Everyone’s INBOX should be restored.  There were two customers whose INBOX was not just moved to a BOGUS file and I had to restore your e-mail from backups made on the 23rd.  I will e-mail you individually.

For those who are interested in what technically went wrong, NFS version 4 uses a program called idmapd to map username to UID across an NFS connection so that files owned by a user on a server are owned by the same user on a client machine.

Until recently this daemon was able to determine the correct domain on it’s own, however, with a recent update, unless it was configured in /etc/idmapd.conf, it didn’t know what domain it belonged to and id mapping broke down.

The mail software, if it doesn’t know who a file belongs to, won’t append or otherwise modify it, instead it moves it to BOGUS.filename.xxxx where xxxx are four characters.  When id mapping went away, as either mail came in or people accessed mail with pop-3 or imapd, or from the shell server, or webmail, when that program attempted to modify your INBOX and couldn’t determine who owned it, it moved it to BOGUS.filename.xxxx and started a new box.

About 160 accounts got affected this way.  To correct this I had to take all the BOGUS files, put them back together into one spool file in the correct order and owing to the random characters, I couldn’t easily script it and had to do by hand which is why it took so long.

I apologize for this interruption.  I’m sure there was probably a note associated with one of these updates, the problem is there were between 300-900 updates to apply per server so no way I could read all that.

Updates to CentOS today and last night caused major problems.

There is a change in id mapping that is causing problem with mail, an update also overwrote my 2.4.6 Apache installation with an ancient version, and it caused some other issues.

The web server has been restored, I need to do an emergency reboot to get the
id mapping issues back in sync, and then will have to move mailboxes back in place by hand.  Please be patient, I am working on correcting this as fast as possible.

There will be maintenance activity between midnight and 4AM this coming Saturday to install a great number of updates to various servers and to image them after successfully updating.

Hopefully this will fix some things, in some cases it’s broken some things, dates, a calendar manager is no longer present.  Orang is still available.  I like dates better and I’ll see if I can chase down a working copy from a third party source.

I recently sent out a note to people still using the old ‘eskimo.com’ server to ask what they were using there that was not available or did not work properly on shellx.eskimo.com.

One thing was some of the BSD games like boggle.

The full suite of BSD games has been installed on shellx.  Most are under /usr/bin but fish, which for some reason was not included in the Fedora RPM, is in /usr/games.

I’ve added the necessary permissions for everyone who has a mysql database here so that Joomla should work without any modification to your permissions if you already have a database.

Anyone who would like to use Joomla or needs a mysql database for other reasons, please e-mail support and one will be created for you.  Please include your login and a desired password in your e-mail.

If you setup Joomla, one thing to be aware of, after you complete the install and remove the installation directory, your browser will still try to run install when you try to go to your Joomla home page.  The fix for this is to clear your browser cache.

It is possible to use Joomla for a website here.  However, those of you who already have databases, the permissions are not adequate, so if you want to run Joomla, contact support here and let them know you need LOCK TABLE added to your database permissions.

I’ve got a copy installed on joomla.eskimo.com that I’m playing with just to learn it.  I am considering using it for some projects and possibly our website.

I’ve set logrotate up on the Apache web logs but I’m running into a problem.  I wanted to compress the rotated logs with the highest possible compression using xz so I added:

compress

compresscmd /usr/bin/xz

compressoptions -e9

compressext .xz

delaycompress

It “almost” works, it does the compression with xz but it is ignoring the compressext command and instead giving it a ‘.gz’ extension which is wrong since it wasn’t compressed with gzip.

Any suggestions how to correct this?  Leave comments or e-mail nanook@eskimo.com please.

If you’d like to have a WordPress blog here on Eskimo, it’s available to you at no extra charge.  Send an e-mail to support@eskimo.com if you need assistance setting it up.

I’m going to be phasing out eskimo-announce list.  This will replace it.  You can have notifications of posts mailed to you by registering here.

It is possible to link your WordPress blog to Status.net so that when you post to your WordPress blog, a notification is automatically posted to Status.net.  This will provide one location where people can follow all your blogs as well as other social activities and it will drive additional traffic to your blog.

To set it up please see:

        http://www.eskimo.com/bbs/showthread.php?tid=89&pid=106#pid106