Operating systems can be… fun. I recall getting involved with a Warp Server case that was as wonky as anything I’ve seen.
Groupe Desjardins in Montreal bought a bunch of ProLiant ML370 G1 servers to try out - they were an IBM shop, but the IT boys wanted to get out of lock-in to a single vendor. They ran OS/2 Warp Server on their systems, and I don’t doubt they were seeing the writing on the wall. Naturally enough, they ran French Warp Server.
Problem was, when they went to use the OS/2 utility to make their recovery floppies (yes, this was in the first years of the millennium), the 370s would lock up tighter than a drum. Not something you’d expect from a relatively simple disk utility on a fairly solid server model running a protected mode O/S. (The ML370s were really the next generation of 1600s, and 1600s were probably the best workhorses we made back then.)
The case gets booted up to my department, because front line support and field service are having no luck at all. There were three of us in the Hull office, and because it was close to home, so to speak, we got the case rather than Houston (same team, two locations, handling all of North America). Unfortunately for Desjardins, they got the BS artist. He sets up one of our office 370s with Warp Server, gets to the same patch levels (I forget IBMspeak for that), and can’t duplicate the problem. He does the same thing with a captured server from the customer, so essentially tells the customer to update everything, which they’ve already done…
Desjardins starts to get antsy. The IT boys there want to pitch our servers to management, but can’t very well while this problem is hanging fire, and they are having absolutely no difficulty locking up their systems at will: it happens every time they use the utility. The case gets booted to me. I get them to send me their media. I can duplicate the problem on the captured server; I can duplicate it on our own server. The difference? Our copy of Warp was English. The English version never freezes.
Clean boot? No difference, so I start swapping out French system files for English. This is easier to do with Warp than an NT-based O/S - the configuration files are all text, so you get an easily perused list of drivers and so forth. Eventually I get a French-localised setup where all of the system files are from the English distro - kernel, drivers, everything. Boots up and works fine… until I run the wretched floppy utility (which I also swapped). I’m tearing out my hair. I even end up heading to Complexe Desjardins to see if I can get some clues from the lads there. Nice visit to my old hometown, but I really didn’t have much opportunity to paint the town red.
Eventually IBM put out a patch roll-up, and our firmware engineers put out an updated BIOS at roughly the same time, and the combination of the two “automagically” fixed the problem. That’s the kicker - you needed both to fix the problem, and I know damned well that neither Big Blue nor Houston were looking at fixing an obscure problem involving the floppy utility in a specific localisation of Warp Server on this particular server. IBM wasn’t about to worry about French Warp Server on a non-Blue model, Houston wasn’t going to worry particularly about French Warp (which was a real niche market by then), and neither patch fixed the problem on its own anyway.
I wrote up an advisory.
As for the customer, they were happy enough - the G2 model came out while this was going on, and it had no problem whatsoever… <sigh>
