As some of you may be aware, I’ve been building kernels for Eskimo’s servers and for my own workstation tickless for some time and making them available to the general public.  Until now I have not made realtime kernels available because building realtime used to conflict with virtual machines, virtual GPUs, and some other generally useful features.

     When the realtime kernel was first integrated into the mainline kernel, UHD630 Intel graphics did not work and since that is what I have on my workstation I did not pursue it.  However, it is now working as are virtual machines under a realtime kernel. I have not yet tested virtual gpu passthru so I do not know if it is working at present but it is such a rare use case I felt it worth while to make this available regardless.

     You can download these kernels from https://www.eskimo.com/kernel/, or by ftp at ftp://ftp.eskimo.com://pub/kernel/, chose the appropriate version (usually only the most current) and the flavor client, server, or realtime.  Then if you have a debian based system, download the *.deb files and install with dpkg -i *.deb, or if you have Redhat based system, download the *.rpm files and install with rpm -i *.rpm.  Note you may have to remove conflicting packages like the previous header package.  To do so without having to remove half your operating system, specify the –nodeps option to rpm -e, for example rpm -e package, you may have to do this also for previous develop packages.

     These kernels are provided without warrantee, however, I would like to hear of something does not work.  If you encounter problems, please e-mail support@eskimo.com.

     The server Inuvik that provides all of these services has been down for a while.  I apologize this but I spent some time in the hospital with a serious infection and another month on home IV and working on computers will tethered to an IV wasn’t very tenable.

      I had ordered a new CPU but that CPU turned out to be defective, would not even post, and then yet another Motherboard also died.  I received a new motherboard, but it has a dead memory channel so it’s going back.  I have yet another I am trying to install now.  Yet another new CPU should arrive between this Wednesday and next Monday so hopefully I can get this b east back online soon.

The grub subsystem on mail.eskimo.com has been repaired. For some reason the Ubuntu update system installed some packages needed for a signed kernel onto a system that is running an unsigned kernel AND installed some arm64 packages on an x86_64 system. I’ve deleted the inappropriate packages, installed the correct and necessary packages and test booted to make sure it comes up cleanly.

I have to take the mail subsystem down for perhaps 45 minutes or so in order to fix an issue with grub that was broken by a Ubuntu update that left it in a half-secure boot and half non-secure boot state.  This could, if not fixed, cause a failure to reboot at a more critical time.  It’s best that I address this now while usage is low.

Wanted to update folks, the new CPU arrived, I installed it and the machine would not even post. Have not had time to troubleshoot further. However, I have another machine that I recently build that is essentially the same architecture, 10-core rather than 18-cores, but the 18 core machine was only running at less than 4% CPU occupancy so this should be fine. Over the next few days I’m going to move these services over to that machine since the existing machine seems to be determined to give me no end in grief.

The estimated time of return to service is around December 20th.  I am waiting upon the arrival of a new cpu.

     Ubuntu is currently down because it is in the process of being moved to new faster hardware.  It should be back up about 12:30AM Dec 11th.

The mail server is taking longer to copy than I anticipated.  This is because the RAID array is still synchronizing disk and so competing with the copy for disk I/O.

     The client mail server, mail.eskimo.com, is in the process of being moved from the Igloo server to the Ice server.  This is being done because the new Ice has 2-1/2 times as many CPU cores as the old machine and 4x the amount of RAM and mail needs more RAM to handle list serve deliveries better.

     During this time, login to shell servers will hang because they are waiting on the mail server spool mounted via NFS which isn’t responding because it is down while being moved.  i expect this to be completed around 3:30AM.