rbp@investor.UUCP (Bob Peirce) (09/05/85)
Lately, as my add on hard disk began to fill up, I found I was
getting random errors. It turns out, when I installed the disk a
bit over a year ago, the instructions said to remove the
termination resistor on drive 0 and said nothing about a resistor
on drive 1. None was installed at the factory.
These are Quantum 8 in. 40 Mb. drives. Current thinking is that
there must be a 220 ohm terminating resistor on drive 0 and a 1k
ohm resistor on drive 1.
This problem is random and unpredictable, but it seems to have a
greater chance to appear as the disk fills. My drives are
running close to 90% of capacity. Beside the obvious HD error
messages to the console, other signs of the problem are messages
from tar that a file has changed size and lseek error messages
from the ps command.gnu@l5.uucp (John Gilmore) (09/07/85)
In article <235@investor.UUCP>, rbp@investor.UUCP (Bob Peirce) writes: > ...the instructions said to remove the > termination resistor on drive 0 and said nothing about a resistor > on drive 1. None was installed at the factory. > ...Current thinking is that > there must be a 220 ohm terminating resistor on drive 0 and a 1k > ohm resistor on drive 1. Terminating resistors are like the weight on the end of a rope, such that when you snap the rope (run a wave down it), the wave doesn't keep bouncing back and forth, but moves the weight a little (using up energy) and damps it down. Without them, the signals going to and from the disk bounce back and forth in the cable and make noise that corrupts the data. They always go on the END of the cable. If you have a cable that goes from your controller to drive 0 and then to drive 1, put them on drive 1. Note that most disks have TWO cables (plus power). One cable goes straight from the disk controller to the drive. The other (typically larger) goes to all the drives in series. That's the one you have to watch the termination on. The actual value of the termination resistors depends on what's going on. Traditional values are 220/330 ohms (220 between the cable and ground, and 330 between the cable and +5 volts ... or maybe it's the other way). These usually come as resistor paks (either a long skinny thing with little wires dripping from it -- a "sip", or single inline package; or a chip-looking object ("dip", dual inline package) often in a bright color like yellow). There's usually a socket next to where the cable plugs in, where one of these would fit (or is currently fitting). It's not always obvious which way the resistors plug in though.
fair@ucbvax.ARPA (Erik E. Fair) (09/08/85)
The Quantum Q500 series of disks have an ST506/419 standard interface. There may be up to four disks on a daisy chain. There should be exactly one terminator (resistor pack) on any daisy chain, on the last disk in the chain. computer |------+-------+-------+-------+ D D D D i i i i s s s s k k k k 0 1 2 3 In the above picture, the terminator pack should be on Disk3, and on NO other. Note, however, that the disks may be in any order on the chain, and the terminator must always be on the last disk in the chain. (one other thing; in ST506, there is a cable per disk drive, in addition to the daisy chained cable, which is not shown on the diagram above, since it is not relevant to this discussion). Erik E. Fair ucbvax!fair fair@ucbarpa.BERKELEY.EDU
jones@fortune.UUCP (Dan Jones) (09/24/85)
Apologies for posting to the net, but my mailer barfed on the address length. In article <235@investor.UUCP> you write: > > This problem is random and unpredictable, but it seems to have a > greater chance to appear as the disk fills. My drives are > running close to 90% of capacity. Beside the obvious HD error > messages to the console, other signs of the problem are messages > from tar that a file has changed size and lseek error messages > from the ps command. First off, there is nothing easy about diagnosing disk related failures. Given the above it is possible, but not likely, that you have a termination problem. I would try the termination first because it is damn simple. Since the flux densities are near maximum on the innermost tracks, I would guess that you have one of the following problems: insufficient shielding or contrariwise too much sensitivity to radiated noise, improper grounding, or a controller deficiency in the data recovery circuits. You should run a diagnostic program on a reasonable number of blocks (>1000) for a reasonable number of iterations (>10) with a random data pattern to verify the success of any changes you make to the system to correct your problem. Remember, innermost tracks equate to the highest block numbers on most systems. Start with the easy changes, like the termination, and go from there. It is quite possible that something as simple as re-seating the cable connectors may solve your problem. Good Luck, Dan Jones UUCP: {ihnp4,ucbvax!amd70,hpda,sri-unix,harpo}!fortune!jones "'Tis with our judgments as our watches, none go just alike, yet each believes his own." Pope, An Essay on Criticism -- Dan Jones UUCP: {ihnp4,ucbvax!amd70,hpda,sri-unix,harpo}!fortune!jones "'Tis with our judgments as our watches, none go just alike, yet each believes his own." Pope, An Essay on Criticism