disk0s2: I/O error even after new disk fitted. Ideas?

[craig1410]

Hi,

Anyone got any experience of an iMac throwing "kernel[0]: disk0s2: I/O error." errors even after the internal drive has been replaced? Could it be the SATA cable or logic board or perhaps just some errant software?

I have a 2007 Aluminium iMac which I've owned since new. I upgraded the hard drive from 320GB to 1TB 2 years ago and it has been running faultlessly for nearly 5 years. I've been running Mountain Lion since the first previews were released and it's been pretty good with only a few little glitches here and there.

A few days ago the machine started running really badly with lock ups and beach balls etc and a quick look in the log files showed "kernel[0]: disk0s2: I/O error." messages scattered around. I made the reasonable assumption that my disk was failing and sourced a new one and fitted it. I restored from Time Machine, going back to before the machine became unstable to avoid any corrupted files. The TM restore took ages and the result was that the disk I/O errors were still showing up and the machine was still hanging. I decided to do a fresh install so downloaded the newly released GM seed of ML and tried to install that but the downloaded DMG file could not be verified prior to install, suggesting corruption during download. Sure enough there were a few disk I/O errors in the log files during the download. I ended up doing the Cmd+R recovery HD thingy and installed ML from Apple's servers directly. It took a while but worked. Note that this was a fresh install over the top of the existing TM restore so is probably more correctly called an upgrade but if I understand correctly, this does replace the System folder and kernel etc.

Now the machine has been a lot more stable but just now I tried to download the XCode 4.4 GM Seed from Apple and it has become corrupted during download with some more disk I/O errors in the logs.

I should mention that I removed VMWare Fusion and some other bits of software which do low-level hardware manipulation and I've upgraded as many applications as possible to try to make sure everything is supported by ML but clearly I still have a problem. My previous drive was getting very full (about 85%) and I know that Macs don't like full drives but I doubt that is the issue.

So, what's my next step? Replace the SATA cable? It felt very positively engaged when I replaced the hard drive but I have heard of at least one report of someone having a bad SATA cable on a MBPro. Maybe a clean install onto a freshly formatted drive while waiting on a cable arriving?

Any ideas would be very welcome - most of the advice on the web just says that the drive is dying and to get a new one but I've already done that. By the way, the old drive is now in an enclosure and I still see disk I/O errors when accessing it via USB which either means it is detecting errors caused during a failed write operation or this issue doesn't just affect the SATA interface which may point to software or logic board. Yes the new hard drive could be faulty but it seems unlikely that I would see the exact same issue.

Oh, one other thing to mention - I've already done the Apple Hardware Tests (both short and extended tests) and they came up clean.

Thanks,
Craig.

 

[Intell]

The new drive could be bad as well. It isn't uncommon for a brand new drive to be faulty. Have it replaced under the warranty or send it back. The SATA cables in iMacs are much more robust then the ones in Macbooks. Because they are real SATA cables and not a flimsy ribbon cable.

 

[craig1410]

The new drive could be bad as well. It isn't uncommon for a brand new drive to be faulty. Have it replaced under the warranty or send it back. The SATA cables in iMacs are much more robust then the ones in Macbooks. Because they are real SATA cables and not a flimsy ribbon cable.

I don't disagree that it could be the new drive but I've been building PCs and latterly Macs for 30 years and have never actually seen a new hard drive fail in this subtle way. I've seen hard drives which are completely DOA, usually due to bad handling in transit but never a case of data corruption.

Just FYI, the new drive is a retail boxed WD Caviar Green 2TB drive from PC World. I decided to go for this drive because it is cooler running and quieter than the bloody noisy Seagate Barracuda which was in it before and I was able to source it locally the same day. The Seagate was a 7200RPM drive and was quite hot running. The WD is 5400RPM but actually has faster sequential rates than the smaller Barracuda and since it is less than half full, I get the extra benefit of using the faster outer edge of the disk.

I've actually just found a brand new SATA cable in a box beside me but it's a straight connection rather than right angled so might not fit. I'll go take a look on iFixit and see if there appears to be enough room for the straight cable.

My thoughts at the moment are:

1. SATA cable replacement
2. Clean Install
3. <anything else suggested here>
4. Get another new drive

Thanks,
Craig.

 

[craig1410]

Hi, just a quick update.

I decided to do a clean install onto my original (supposedly faulty) 1TB drive which is now mounted in a USB external enclosure. I erased it with Disk Utility and installed a fresh copy of the GM seed of 10.8 onto it. All went well.

Something I have noticed is that disk I/O errors seem to only be detected during READ operations, not during writes. I discovered this because I had been downloading the OS X 10.8 install package from Apple using my Macbook Pro and then transferring it over ethernet to the iMac and when I did this it seemed to copy fine but when I tried to use the install package it would fail verification (checksum failure). I tried this several times when booted from the "non-clean" install and I would always get a disk I/O error and the package would fail to verify. However, now I'm booted from a clean installation I can copy the 4GB+ install package to the internal SATA disk and I can test the checksum using the md5 utility and it copies correctly. What this suggests to me is that my hardware may well be okay and this may indeed be a software/kext problem or low level filesystem corruption. By doing a clean install on a freshly formatted partition it seems to avoid this.

So, my plan is now to migrate all my data manually to the 1TB external drive to the point where I can do everything I need to do (email, docs etc.). I will then clone this using CCC or SuperDuper to the 2TB new internal drive and see if that works or not. If I get I/O errors I will change the SATA cable and failing that I will just condemn the internal SATA interface and buy a FW800 external drive and use that instead.

A clean install is probably not a bad thing as I've been doing archive/install's since Tiger so god knows what crap has been sitting there...

C.

 

[craig1410]

Just another update:

I'm now typing this on my iMac running on the new 2TB internal SATA drive. It seems that disk I/O errors can lurk around long after whatever event caused them and continue to give the impression that the machine is still generating them. It also only takes a few bad files to cause severe slowdown of the machine, especially when things like spotlight is indexing and/or the app store is downloading stuff. Basically all I/O seems to stop for several seconds which causes things to queue up and get exponentially worse, That is my current understanding at least...

I don't know why or when the disk I/O errors originally happened (and why they weren't detected) but I managed to identify the bad files by using the SuperDuper cloner to clone from the internal drive to an external drive. As part of the cloning operation it will detect if files have I/O errors on them and stop the cloning operation. I ended up having to delete some iTunes content (which I can get back through iTunes match anyway) and a few emails. Something like 6 files in total. That allowed the clone to complete. I then had another problem with some stale app store downloads but I flushed the corresponding cache files (not cloned by SD so not detected) and cancelled and restarted the downloads.

I'm now transferring over the last few bits and pieces of data and running permissions repair etc. but it's looking much better now I think (hope)!

This has been a massive pain in the neck and if only I had known that "disk I/O error" in the log file doesn't mean that an I/O error has just occurred and that it may have occurred at some point in the past, then I would have taken a few different directions during the recovery. For one, I would have tried detecting and cleaning up the bad files to and try to work out from Time Machine when they occurred to help determine if it was a single event (e.g. power surge) or a gradual problem. I would probably have replaced the drive as a cautionary step anyway but I wouldn't have freaked out when the new drive was still detecting errors...and would have tried to clean them up first.

Anyway, I've now got a shiny new 2TB (cool and quiet) internal drive and I also bought a 3TB external FW800 drive which I will use for cloning and other things. I've also bought a disk integrity/performance testing suite (http://www.speedtools.com/TestSuite.html) and a full featured copy of SuperDuper (http://www.shirt-pocket.com/SuperDuper/SuperDuperDescription.html) which I shall use to take regular clones of my main drive in future.

After this experience I can understand why many in the Mac world were excited about possibly bringing ZFS to the Mac. I will now add my voice to that request!!

 

[mas90guru]

I'm experiencing a nearly identical issue - except I'm not seeing file errors -- only I/O error. in the console.

Curious if your changes completely eliminated the issue.

I took my iMac (similar 2007 model as yours) into the Genius bar and they didn't do much I couldn't have done (run the standard MAC disk utility which came back clean).

They advised to reinstall the OS which I'm probably going to do. I've been perplexed because the only symptom I'm seeing is sporadic beach ball and 15-30 second freezing.

I'm a little afraid that there could be more to come if there's a failing disk inside.

 

[craig1410]

Hi,
Sorry I meant to update this thread earlier.

My issues are now completely resolved but there was another twist in the story after my last post...

The brand new Western Digital Caviar Green 2TB drive was faulty!

What happened was that my system continued to throw Disk I/O errors and things started to get worse and worse. Then eventually one day I noticed that the S.M.A.R.T. status in Disk Utility showed "FAILING" instead of "PASSED". When I looked at the SMART details, I found that the Reallocated Sectors Count metric was the cause of the SMART failure and was at a high level. There is a threshold set in the SMART params and when the number of sectors exceeds this threshold then it triggers the FAILING message.

When this happened, I was in the process of trying a replacement SATA cable. I managed to rig up a brand new SATA cable which fitted well enough when the iMac was partially disassembled and if this had worked then I would have ordered a proper iMac SATA cable. I also tried running the 2TB drive in an external USB enclosure and it was continuing to throw I/O errors and generally perform very poorly.

So, although I knew all along that it was possible for a brand new drive to be faulty, I was pretty unlucky that it was faulty in an almost identical way to the old drive which I took out. If it had been DOA then I would have realised that very quickly but the fact it worked but slowly corrupted my data was a bit unlucky. I've also never before experienced a faulty brand new drive and in all likelihood, never will again.

I now have a Seagate Barracuda Green 2TB drive which is sold through PC World in the UK as cnMemory brand. These drives spin at 5900 instead of 5400 and have quite a bit higher sustained data rates than the WD Caviar Green drives while maintaining low power consumption and lower noise than 7200 RPM drives. I'm very happy with my Barracuda and had no trouble getting PC World to do the exchange. I even got £7 as a partial refund!

So, get yourself a copy of this app: http://www.volitans-software.com/smart_utility.php and check your SMART status. You can get a lot of this information from Disk Utility but this app is easier to use. It is not a free app but works for 30 days free. You can also run short and long tests on the drive. Note that the long tests can take a number of hours so start with small. Look at the SMART stats and in particular the ones related to reallocated sectors. Take a note of the stats and see if they get worse over time. Don't worry about things like raw error rate as these are expected to show large values. For reference, my 1 month old WD Caviar has something like 1300 reallocated sectors whereas my Barracuda replacement drive still has zero reallocated sectors after a couple of weeks of use.

Feel free to get back in touch if you need any further help. Having gone through this recent pain, I'd be happy to help someone else avoid it if I can.

Good luck,
Craig.

 

[mas90guru]

Thanks - I appreciate the detail of your response as well as the link. I'm looking at the disk utility right now.

 

[craig1410]

Thanks - I appreciate the detail of your response as well as the link. I'm looking at the disk utility right now.

Ha ha, as you may have noticed from my other posts, "detail" is my middle name....

By the way, if you are still on your original disk from 2007 then it could well be the disk as they were notorious for failing just after the 3 year Applecare warranty period. I replaced my 320G drive with the 1TB drive a week before my warranty expired due to these reports. If yours is still going after 5 years then you've probably run out of luck.

Hopefully the SMART utility will shed some light.

Craig.

 

[mcgroa]

me too similar issue

i cant believe i just found this thread - very useful.

i have a late 09 imac with a 1tb seagate barracuda that was beachballing and throwing occasional I/O errors.

so i bought a new "similar" seagate and replaced (seagate only to try and avoid fan/temperature sensor issues which i had to solve in the end with HDD Fan utility)

I booted off OSX DVD and then used Timemachine backup and got everything back to normal and felt quite happy with myself.

2wks vacation with IMAC off, and i come back boot my IMAC and go straight to clicking hard disk, that takes 45 mins to boot throwing many I/O errors.

New HD only showing 27hrs old but SMART utility says "FAILING" with Uncorrectable errors of READ DMA EXT and READ FPDMA QUEUED

MAC Disk Utility says disk is fine with SMART status of "VERIFIED"

Any thing I can do? Prev posts talk about deleting files. Did my restore mean I brought over some junk that could be causing these errors that is non hardware related?

Send the suck new Seagate back to Amazon and get something else sounds like best bet.?

any help apprec. cheers