Kernel panics from cabling?

[RChapman]

I am now getting kernel panics reproducibly after having replaced an internal hard drive, but only with two drives connected, as in the previous working configuration. My system: 400 MHz AGP G4. It had a 10-GB Maxtor internal ATA + 40-GB Seagate internal ATA (the boot drive with System 10.3.9).

The Seagate was showing frequent errors, so I decided to replace it with a 60-GB Western Digital drive I had on hand. First, I temporarily removed the 10-GB Maxtor, installed the 60-GB WD in its place on the terminal connector of the ribbon cable, with the 40-GB Seagate on the middle connector (the other terminus to the motherboard). Disk Utility initialized the new drive, erased it, and verified that it looked OK. I then used CMS Products' BounceBack Pro to copy the entire old 40-GB Seagate's contents to the new 60-GB WD. That copying seemed to go fine.

When I temporarily removed the new 60-GB WD, replaced its position in the bay with the 10-GB Seagate that I want to keep, and placed the new 60-GB WD into the top slot of the bay (middle cable connector), the system gave kernel panics even when booting from the OS X 10.3 installation disk. Sometimes those came before anything booted, sometimes after Disk Utility was started from the CD though ti could never get far. I checked the four RAM modules (one each 64, 128, 256, 512 MB) and removing each did not prevent the kernel panic.

What did prevent the kernel panic was removing the last drive on the cable, at this time the 10-GB Maxtor. With that connector removed, everything booted fine. Next, I could place the cable's middle connector on the 10-GB Maxtor, and it would boot from that drive fine. Whenever two drives were on the cable, kernel panic. So, bad cable? I bought a new ATA/IDE cable for dual internal drives, connected it onto the two drives, and same result: reproducible kernel panics when two drives are attached! Next, I tried a single-drive cable I had on hand: connecting only the new 60-GB WD gave a kernel panic. Replacing the cable with the original dual-drive cable, but connecting only the new 60-GB WD on the middle connector: everything OK.

TechTool Pro 4.0.3 says all the hardware is fine. Disk Utility and Norton Disk Doctor say the software on the drives is OK. So what happened that I can no longer have two drives on the cable without getting a kernel panic? I'd like to use the 10-GB Maxtor if possible, though it's not critical because I backed up all its files to the 40-GB Seagate before copying that to the new 60-GB WD.

 

[joker2]

Are both of the drives set to "Cable Select" or is one set as "Master" and one "Slave" (or both one or the other?) This would be the first thing I'd check...

 

[RChapman]

Drive settings

The new 60-GB WD initially came with a jumper for Cable Select, and I did not change it for the first installation (when the 40-GB Seagate was copied to it). After the kernel panics started, I removed the jumper, but that caused no change. (In that configuration, it works fine as the sole drive.) I don't think the jumpers on the 10-GB Maxtor have ever been changed, either when it was my original sole drive or later after I added the 40-GB Seagate as the new startup disk, and I'm not sure what its jumper setting means. Do the jumper settings matter in a Mac, since the startup disk is controlled by software?

 

[joker2]

Yes, the startup disk is controlled by software, but the Master/Slave/Cable Select designation works at the hardware level, the priority of interrupts... if two claim "master" designation, it can confuse the machine.

The boot drive doesn't come into play until Open Firmware loads... (I'm guessing it loads the OS with that drive, and then while polling the other drive it comes back with the conflict, sending the machine into kernal panic)

Apple Article on Master/Slave configurations for G3/G4 machines...

 

[frankblundt]

i'd try replacing the cable too, to see if that helps - they seem to be quite a common point of mechanical failure, especially the intermittent kind when the wires are broken but still in contact. Especially after changing drives over (when they tend to get the most stress).
And make sure they're not too long or too twisted either (can cause data errors).
And if you're using cable-select the cable needs to be compatible (usually indicated by a black connector for master and a grey one for slave) - if in doubt set the end drive's jumpers to master and the middle to slave.

 

[RChapman]

Problem solved

The problem apparently was the jumper setting on the new WD drive. (Thanks, joker2, for the Apple URL on the topic.) I had tried a newly bought dual-drive cable but had the same result as with the old cable. Looking up the specs on the various drives, I found that the original 10-GB Maxtor was indeed set as a Master drive. A slight mystery is why the 60-GB WD didn't work with the 10-GB Maxtor, when the former came set as Cable Select and was on the middle connector. I believe (from recent reading) that one needs a Cable Select cable to use that setting, but my cable has a black terminal connector and gray middle connector, which is the description I read about Cable Select cables. But maybe that doesn't ensure that it is one? (So if you don't have a Cable Select cable, does that setting behave as Master? This drive did.) Anyway, by setting the jumper on the new WD to Slave, the combination works fine. The Mac initially booted on the old 10-GB Maxtor (Master), but "Startup Disk" changed that to the new WD.

 

[joker2]

I'm glad you got it solved, and that I was able to help.

In order for the "cable select" to work, both of the drives must be jumpered to the "cable select" setting... (and even then it isn't guaranteed to work, as demonstrated by a Western Digital and Seagate combination on my windows machine that is now manually set as "Master" & "Slave".)