One more day
May. 26th, 2011 09:28 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
altivoAnd then a long weekend. Not that it's going to be much fun, as I should be weaving every free moment of it, and the weather isn't all that promising anyway. I should learn not to take on these projects that outlast the fun part.
Anyway, yesterday's Alpha glitch is now fixed. A replacement lithium cell took care of it. Turns out that the behavior that first mystified me is a well documented foible of the small Alphas at that time. The firmware designers made a zero value in a test byte into the signal to autoboot Windows NT rather than thinking about the fact that a CMOS battery failure would put a zero there, and defaulting to the NT loader would short circuit all the diagnostic facilities in the machine (of which there are many, even without loading anything from disk.) Fortunately it is easy to fix the problem, IF you recognize it.
I don't think my other Alpha, which is running at the library, will be prone to that as it is a later model and was never intended to run Windows NT.
Once the Alpha was running normally, I tried something I always meant to test. Brought up the MicroVAX emulation on the Linux PC, booted the Alpha, and checked to see if the two OpenVMS systems would recognize each other. They do. No issues there, the real machine and the fake one will be happy to form a cluster and share workloads. Of course, I don't have nearly that much work for them to know, but it's nice to know that the emulator (SIMH) is that good. The operating system is the real, undoctored installation intended for genuine DEC iron, and it loads and runs without complaint. The Alpha, running the next higher version of OpenVMS, is perfectly convinced that the VAX is genuine. DEC/Compaq/HP's software for clustering is flexible enough to allow a mixed group (both in terms of hardware and software version) to work together.
Gloom and rain much of the day, unseasonably cold. Possible frost tonight even. I sure hope it warms up a little for the weekend. Some sun would be nice too.
Anyway, yesterday's Alpha glitch is now fixed. A replacement lithium cell took care of it. Turns out that the behavior that first mystified me is a well documented foible of the small Alphas at that time. The firmware designers made a zero value in a test byte into the signal to autoboot Windows NT rather than thinking about the fact that a CMOS battery failure would put a zero there, and defaulting to the NT loader would short circuit all the diagnostic facilities in the machine (of which there are many, even without loading anything from disk.) Fortunately it is easy to fix the problem, IF you recognize it.
I don't think my other Alpha, which is running at the library, will be prone to that as it is a later model and was never intended to run Windows NT.
Once the Alpha was running normally, I tried something I always meant to test. Brought up the MicroVAX emulation on the Linux PC, booted the Alpha, and checked to see if the two OpenVMS systems would recognize each other. They do. No issues there, the real machine and the fake one will be happy to form a cluster and share workloads. Of course, I don't have nearly that much work for them to know, but it's nice to know that the emulator (SIMH) is that good. The operating system is the real, undoctored installation intended for genuine DEC iron, and it loads and runs without complaint. The Alpha, running the next higher version of OpenVMS, is perfectly convinced that the VAX is genuine. DEC/Compaq/HP's software for clustering is flexible enough to allow a mixed group (both in terms of hardware and software version) to work together.
Gloom and rain much of the day, unseasonably cold. Possible frost tonight even. I sure hope it warms up a little for the weekend. Some sun would be nice too.
