I will be taking scientific7 down for imaging tonight around 10pm (I will wait if it is actively in use at that time) for about 25 minutes. I had intended to get to it with the other machines last night but problems with one of the host machines booting properly prevented it.
There appears to be an issue with the version of systemd on these machines. If it has only been operating for a while, it works correctly, but if it’s been up for a week or more, it often hangs during a run level change. I am seeing the same behavior on my workstation which is also Ubuntu 15.10.
Everything is back up except eskimo.com, which was not part of tonight’s maintenance. I’m too tired to make another trip tonight so it is going to have to stay down until I can make another trip tomorrow.
The mail spool server does not seem to be booting properly so I am leaving for the co-lo to go give it a kick.
The server which has home directories has been rebooted and is restored to service. Work on the server with the mail spool and client mail server is still in progress.
I will be rebooting some servers and taking others down to image them tonight. The reboots are necessary to load a new kernel. Imaging is necessary to have current backups in the event of a machine failure or corruption.
Servers that need to be rebooted include all the host machines so everything will be unavailable for a brief period during this interval but not everything at the same time.
Most outages will be brief (less than ten minutes) but the mail client server will need about a half hour of downtime to image it.
For older Unix’s and Linux’s that kept track of time in a 32-bit integer counting seconds since January 1, 1970, time would run out in 2038.
I have experimentally determined that I can set dates out to the year 2262 (but not 2263). It appears signed 64-bit nanoseconds runs out in 2262 so that is probably what the current kernels are using.
The old Eskimo.Com shell server will need to be retired by 2038 if it isn’t dust before that time since no updates to SunOS 4.1.4 are ever likely to be made available to resolve the time issue.
I had to say something. Wind is picking up now and we’ve just had three power hits here. My workstation survived, enough capacitor in the power supply to bridge the gap, but we may be in the dark yet.
So far the wind hasn’t happened in a big way, maybe 15 MPH gusts or so and 3-4 constant. There is eight hours left on the high wind warning though so might happen yet. The barometer is taking another big dive here, down to 28.18 so far. My sinuses are beginning to complain. But the telephones are still working.
There is a high wind warning for tomorrow predicting sustained winds of 40 MPH and gusts up to 60 MPH. Almost invariably when we have winds of that magnitude I lose power, Internet, and phones here (Internet and phones being provided by Comcast, and Comcast distribution amplifiers require power to run).
So if you call and the phone isn’t answered, please leave a message with your name, telephone number, user ID, and the nature of your call and I will return your call as soon as possible.
The equipment is in a co-location facility with a big UPS and diesel backup generators so service should not be impacted.
I applied some updates to one of the host machines and it caused a number of guest machines to reboot. There was no advanced warning to me else I would have done this at a low traffic time. I apologize for the interruption.