Date: Wed, 27 Dec 1995 01:33:35 -0800 From: Pete Delaney <pete@RockyMountain.rahul.net> To: rgrimes@GndRsh.aac.dev.com Cc: karl@mcs.com, Peter.Delaney@eng.sun.com, jbryant@argus.iadfw.net, fnf@amigalib.com, bardhan@corp.megatest.com, hackers@freefall.FreeBSD.org, pete@RockyMountain.rahul.net, freebsd-hackers@freebsd.org, karl@mcs.net Subject: Re: 4GB Seagate Barracuda (Hawk Replacement) goes nuts after a while Message-ID: <199512270933.AA01962@RockyMountain.rahul.net>
next in thread | raw e-mail | index | archive | help
> > Hi Rodney (et. al.): > > Hello. Not sure what lead you to me on this, but I suppose I am a bit > of a known source for information on these types of things. We discussed my 4GB Hawk drive when I cooked it about four months ago. > > > Seagate replaced the 4GB Hawk with a repaired 4GB Barracuda but > > I seem to have a problem after Solaris 2.4 (without any patches) has > > been running for a while but not doing anything. I suppose the > > nfsd is scanning the disk once per night but other than that it's > > wasn't doing much. > > Nice of seagate to replace your drive with a faster one except that the > Barracuda series is pretty much known as a lemon do to overheating > problems and media that likes to flake :-(. So I would be better off with a Hawk, good to know. > > > > > It seems to get in a state where some SCSI reads are taking a few > > seconds. > > This is a good indication that a new error has developed and the drive > goes to every possible measure to recover the data then it spends a bit > more time doing an automatic bad sector allocation to map it out. Most > drives make a lot of noise during this operation as it is doing recalls > and head offsets trying to recover the data. The strange thing is that when I reboot the system the problem goes away, as if a SCSI Reset clears the problem. It's as if a timming track is only read during a SCSI reset and their is internal slippage between the platters. > > > I shutdown openwin (took about 5 mins) and then ran format > > and scan'ed the disk. It was stopping at some sectors for as long as > > perhaps 20 seconds and then moving on. I can hear the drive makeing a > > lot of noise, kind of like the Hawk before it died. > > And a strong indication that your Barracuda is about to die :-(. But why does it go away when I reboot the system? A reboot does a system reset on the sparc and consequently does a few SCSI resets. I think the drive may have experienced this even a few months ago when I got it and I thought it was a Solaris 2.4 NFS bug. Now it's becoming clear that the problem is with the drive (or perhaps very low SCSI or disk driver processing that is common for raw and cooked I/O). > > > I suppose it's time to back up the drive, I fear this is going to > > require my replacing the drive yet again. For years I haven't had > > any trouble with my Micropolus 1.7GB and 1GB drives. My experience > > with the Seagate is leading me to believe I should avoid Seagate > > in the future. > > I have been avoiding Seagate drives for about the past 5 years. Some times > I have used them due to the fact that they made the only drive in the right > capacity/performance/price point. Infact I sold 2 and 4G hawks for a short > period of time earlier this year since I was having a very hard time getting > any Quantum or Micropolis drives in that size. I seem to have no problems with the Quantum and Micropolis drives. Seagate had a bad reputation and apparently still lives up to it's poor reputation. Returning drives to Seagate every 3 months for the next five years is going to be a painfull experience. > > > The drive is very well cooled, so it's not a heat problem. Any thoughts? > > Do you have any tools to look at the grown defect list for the drive, if > it has more than a few sectors in it then the drive is defanitly going > south on you and you need to get your data backed up ASAP. Sure, Sun's format utiltiy is pretty good fro this kind of stuff. I suspect the defect list is not growing. I havn't found a single defective sector while scaning the complete drive a few times this evening. > If there are > not any grown defects then there is probably something wrong on the scsi > bus casing retries, this could be bad termination or bad cables, or even > bad or noisy power. I detached the scsi cable and it didn't get rid of the problem, the power supply is the one that supplies the CPU. I suppose I could try adding a loading resistor if their isn't one in there now. > > > -- > Rod Grimes rgrimes@gndrsh.aac.dev.com > Accurate Automation Company Reliable computers for FreeBSD Nice to hear back from ya. -pete
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199512270933.AA01962>