> On Jan 22, 2015, at 4:29 AM, René J.V. Bertin <rjvber...@gmail.com> wrote: > > On Thursday January 22 2015 08:56:25 James Linder wrote: > >> I cannot explain why a (normally) rational, sane thinking idiot did not make >> that his first port of call (beautifully synced) >> >> Jan 21 00:04:34 haycorn kernel[0]: disk0s2: I/O error. >> >> Thanks everybody, and sorry for being an idiot > > I wouldn't immediately think about disk i/o errors either from the symptoms > you described (not for short freezes in anyway). Not with an hdd anyway. > What's in your logs around those I/O error messages, and what do the > smartmontools (in MacPorts) tell about the disk's health? smartctl -a > /dev/disk0 and do run smartctl -t long /dev/disk0 (disabling disk spin down > for the duration of the test as that would interrupt it)? > > A bit too many reports of comparable symptoms in 10.9 somehow related to disk > I/O errors for my comfort zone. OS X wouldn't be doing something low level > that somehow stresses the disk hardware I hope?
Long ago in the days of Ultrix, there were some massive parts of the low-level disk-i/o that never saw the light of day. (Ultrix from DEC being a direct BSD clone.) None of the hardware level routines were addressed by any of the "accounting routines." Made performance look good, but debugging impossible. Performance was unexplaninedly lower than the new hardware predicted. After a tremendous amount of effort on the part of numerous kernel programmers at DEC, they discovered that the bottom level BSD I/O modules had not been "looked at" (literally) since PDP days. Disk I/O was being done in 128 byte blocks. The new hardware had 4096 byte tracks. Calculate how many I/Os were required to write a single track! Increasing the basic block size dramatically cut down on the number of I/Os and their consequent overhead - performance improvement was dramatic. One thing which I noticed immediately when I turned on iCloud disk in Yosemite -- the "lag" involved with launching any program which stored anything "in the cloud." Not unexpected, but significant none the less. Similarly, I had a problem where my internal hard drive would literally not spin-up. Took the iMac in to the Apple Store and they ran their diagnostics and pronounced nothing wrong -- the tests passed with flying colors! I finally convinced them that the drive was not spinning up and they got a tech to come out front who had a stethoscope and instantly verified that the drive was not spinning. I've seen too many cases related to both BSD (and later Mach, i.e. NeXT and OSX) where much of the hardware level "stuff" is completely ignored by any of the upper-level reporting software. -- one of the main reasons why Drive manufacturers developed S.M.A.R.T. -- the OS does not do the job. In my experience, by the time the OS flags a Disk error, you have been suffering constant performance degrading failures which are simply below the "reporting threshold", for quite some time. T.T.F.N. William H. Magill mag...@icloud.com mag...@mac.com whmag...@gmail.com _______________________________________________ macports-users mailing list macports-users@lists.macosforge.org https://lists.macosforge.org/mailman/listinfo/macports-users
