The client mail server is totally wedged and I can not figure out why so I am restoring it from backups.  I have saved the spool prior so no mail will be lost.

     Last night I became aware that our outbound mail server mail.eskimo.com had been added to the SORBS real time black hole list.

     I received no warnings or notifications or spam complaints prior so I had no reason to believe we had an active spammer.

     Afterwards, I was able to locate enough information in their database to identify the sender and to determine his machine was infected rather than a case of compromised password.  I have disabled the account pending some confirmation that the customers machine has been cleaned.

     Two carriers I am aware of rely solely on SORBS for spam filtering (an extremely bad practice), those are Yahoo.com and Cox.com.  As a consequence of being on SORBS, more than 7000 messages backed in queue to these two domains.

     In an effort to try to force some of the backlog out I reduced the timeouts so that cox and yahoo were more appropriately hammered, but was unaware there were limits to the allowed parameters (they were NOT documented in the Postfix documentation AND postfix check which is supposed to check for a valid configuration did not complain) but it ended up breaking the outgoing server after I went to bed which by this time was about 6AM.

     I finally figured out there was a problem around 11AM and fixed it.

     I applied to be delisted at SORBS but being a volunteer run organization it does not happen fast and so we are still listed today and Yahoo and Cox are still refusing our e-mail.

     Please understand I have no control over the poor policies at Cox or Yahoo, please contact their support staff regarding your mail not going through to these sites.

     I rarely write or post about things which are not highly relevant to our service here, but I came across this and what he is saying here is so relevant to life in general, I thought it worth sharing:

     I am going to be rebooting all Intel based servers here shortly to install a 5.5 kernel.  This kernel should provide some performance improvement, it certainly has done so for my workstation.

     The user count in ruptime was wrong, that has been corrected.

     Backups of zorin are completed.  It has been restored to service.

     Backups are completed on Mint.  It has been restored to service.

     Zorin and Mint are now going down to be imaged.  Debian, JuLinux, MxLinux, and Ubuntu are alternate Debian derived machines available while these are being backed up.

     Reboots are completed.

     I am still checking to make sure NIS bound and NFS mounted on all servers.

     Shortly I will be taking down some servers for imaging, specifically Ubuntu will be first.

     I will be rebooting almost all the servers tonight (all of the Intel based machines).  If everything goes well, downtime should be minimal perhaps 15-20 minutes but there are no guarantees until it’s finished.