[Info-vax] VMS startup - building LLA devices

[Marc Van Dyck]

on 17/03/2023, Ian Miller supposed :
> On Friday, March 17, 2023 at 3:03:32 PM UTC, Marc Van Dyck wrote:
>> Ian Miller used his keyboard to write :
>>> On Wednesday, March 15, 2023 at 4:18:37 PM UTC, Marc Van Dyck wrote: 
>>>> Very early in the OpenVMS startup sequence, VMS builds the LLA (Lan 
>>>> failover) devices. Can someone tell me which file(s) the startup code 
>>>> gets the information from to do that ? 
>>>> 
>>>> -- 
>>>> Marc Van Dyck 
>>> 
>>> SYS$SYSTEM:LAN$DEVICE_DATABASE.DAT
>> That's also what I was thinking. But no. Let me explain. 
>> 
>> I have a system disk residing on a SAN that I must be able to boot 
>> from two different hardwares. One is a BL870, the other is an RX2800. 
>> So I built two different LAN$DEVICE_DATABASE.DAT files, one for each 
>> config, and before reboot, we put in place the one that will be in 
>> effect after the reboot. Should work, right ? Well, it doesn't. The 
>> RX2800 has only EIxx devices, the BL870 has EIxx and EWxx ones. If 
>> I'm on the blade, rename files to reboot on the RX, I still see EXxx 
>> devices in the LAN config after the reboot. 
>> 
>> Also, in the doc, it is said that this file can be moved to an other 
>> location by defining a logical name. But how can that work when the 
>> LLA formation is done so early in the boot sequence, way before DCL 
>> even exists ? 
>> 
>> -- 
>> Marc Van Dyck
>
> A quick bit of research shows LANACP started from VMS$DEVICE_STARTUP after 
> SYCONFIG and SYLOGICALS.COM have been executed. As far as I know LANACP is 
> the process that reads the LAN$DEVICE_DATABASE and creates the LL devices.

I don't think so. As far as I can see, it happens much earlier, way
before SYS$SYSTEM:STARTUP.COM begins execution. I'm not at work today,
so I can't get to my systems, but next monday I will post a console
log showing that.

-- 
Marc Van Dyck