Film-Tech Forum ARCHIVE: DSL100 SAS card gotcha

This is topic DSL100 SAS card gotcha in forum Digital Cinema Forum at Film-Tech Forum ARCHIVE.

To visit this topic, use this URL:
https://ft-forum.com/ft/cgi-bin/ubb/ultimatebb.cgi?ubb=get_topic;f=16;t=003623

Posted by Leo Enticknap (Member # 534) on 03-14-2019, 08:00 AM:

Made an interesting discovery that I thought would be worth writing up.

I have been refurbishing a DSL100 for delivery to a customer with the new Dolby TMS software on it. As the only install ISO I have is an upgrader (i.e. Show Manager has to be installed first - you can't install it on a clean RAID), after I had got the hardware side of things done (through internal clean, replaced thermal grease on the processor/heatsink mating, fans, RAID drives and the CMOS battery), I installed Show Manager (4.9.1.22). At that point it occurred to me that as there would be no external SAS RAID used in this installation, I could pull the SAS card and file it in our spares inventory, in case we ever had another customer whose SAS card died.

After I pulled the card I rebooted into Show Manager, just to confirm that all was OK. The SATA RAID card's BIOS appeared OK during boot, and displayed the list of connected drives. But Show Manager would not see the SATA RAID. After I put the SAS card back, it did.

Thinking that this could be to do with the installation process - that if a SAS card was present during installation, it had to be present thereafter - I pulled the SAS card, nuked the RAID, and attempted a reinstall of Show Manager. This time it said that it was running on a DSS200, and aborted the install!

The same thing happened when I ran the Show Manager to TMS 4.3 upgrade DVD - it said "This is a DSS200 - invalid hardware - bugger off, pal!" When I replaced the SAS card, the upgrade ran without problems or issues.

So the moral of the story is that even if you're not using it, that SAS card has to be present in a DSL100 in order for both Show Manager and the new TMS software to see the SATA RAID, as well.

I'm guessing that the DSL100 uses the same model of motherboard and SATA card as the DSS200, that the presence of absence of the SAS card is how the installer determines which of the two models it's running on.

Posted by Steve Guttag (Member # 268) on 03-14-2019, 09:50 AM:

It is a different motherboard. Dolby has used the absence or presence of things to decide what model server the software is running on. Perhaps, since the SAS is unique to the DSL100, they triggered off of that.

I recently converted a DSS200 to a DSL200 that had an X7 motherboard. I don't know if the DSL200 ever shipped with the X7 or not (I've only seen them with the X8). The key for the DSL200 is the presence of the 9750-8 RAID card (not the earlier 9650-8). Once it saw that, it was a DSL200 and has performed well in its new capacity (as a DSS200 it died due to my first failed/non-repairable CAT862 so I "recycled it" to replace a troublesome DSL100...the speed/reliability improvement was dramatic).

Posted by Leo Enticknap (Member # 534) on 03-14-2019, 10:18 AM:

The SATA RAID card in this DSL100 is a 9650 8-port, so that makes sense.

Posted by Marcel Birgelen (Member # 6801) on 03-14-2019, 01:20 PM:

I guess what Steve said... They're probably looking at the presence of the SAS RAID controller to identify the type of system.

There are much more clever (and reliable) ways to do it, but... software + time + money, don't always pave the way to good decisions (just look at the unfolding Boeing 737 Max debacle...).

Another problem that could arise from removing devices, is that Linux sometimes tends to attach the remaining devices under another device label, which can also mess-up stuff in badly written scripts and software.

Posted by Mattias Mattsson (Member # 4308) on 03-18-2019, 07:00 PM:

As already mentioned the DSL/DSS startup scripts uses the RAID card (and IDE flash) for platform detection. Look into /etc/rc.d/platform_detection.sh for details. Here is an excerpt of the most relevant part:

if [ ! -z "$TW9550" ]; then
TMP_PLATFORM=dss100
elif [[ ! -z "$TW9650" ]] && [ -z "$TW9690" ] && [ -z "$TW8006" ]; then
if [ -d /proc/ide/hda ]; then
TMP_PLATFORM=dss200
else
TMP_PLATFORM=dss200-noflash
fi
elif [[ ! -z "$TW97504I" ]] && [ -z "$TW9690" ] && [ -z "$TW8006" ]; then
if [ -d /proc/ide/hda ]; then
TMP_PLATFORM=dss200
else
TMP_PLATFORM=dss200-noflash
fi
VARIANT=SAS
elif [[ ! -z "$TW9650" ]] && [[ ! -z "$TW9690" || ! -z "$TW8006" ]]; then
TMP_PLATFORM=dsl100
elif [ ! -z "$TW9750" ]; then
TMP_PLATFORM=dsl200
elif [ ! -z "$NOHWRAID" ]; then
TMP_PLATFORM=dss220
else
echo "WARN: Platform detection failed, assuming DSS100"
TMP_PLATFORM=dss100
fi