Atomic I/O letters column #154Originally published 2014, in PC & Tech Authority
(in which Atomic magazine is now a section)
Reprinted here December 12, 2014 Last modified 16-Jan-2015.
A couple of months ago I knocked over my 2Tb Seagate Expansion USB external hard drive - I had it standing vertically and elbowed it over onto the desk with a sickening thump. It seemed fine afterwards though, I did an error scan and there were no problems. This morning though, it's stone dead, might as well not be plugged in at all as far as the computer's concerned.
As you have said on many previous occasions, data you do not have backed up is data you do not particularly want. Imagine a crazy parallel universe in which there was data that I DID want on this drive. What chance do I have of recovering it, and how?
A standard physically-small external hard drive will contain a 2.5-inch laptop drive, which is quite resistant to physical damage from falls. Larger external drives that're cheaper per gigabyte contain ordinary 3.5-inch desktop drives, though, and those are more vulnerable to drop damage. I say this for the benefit of other readers, though, not you, because if a fall's going to break a hard drive, it'll almost always break it immediately. Not even an hour later, let alone two months.
A large proportion - possibly a majority, actually - of external-drive failures are electronic, not mechanical. And, better news yet, they're usually not a failure of the hard drive's own electronics, but of either the drive box's power supply or its "bridge" hardware, which is what translates the hard drive's native ATA into to USB or FireWire or whatever. (ESATA external drives don't have any bridge hardware to fail, though.)
So if you open the box and plug the drive inside into new bridge hardware, or just straight into a PC, you'll often be in business again.
The new bridge hardware can be in another external box, or it can be one of the super-cheap USB-to-PATA/SATA adapter cables that're all over eBay for a few bucks delivered.
Note that the cheapest kits don't come with a power supply for the drive, but if the broken external box's PSU still works you can use that. A full USB-to-PATA-and-SATA kit with a power supply and cables for a SATA drive (if you want to hook up an old Parallel ATA drive you'll have to bring your own ribbon cable) should still cost you less than $US15 on eBay, but note that separate power supplies just like the ones that come in these kits cost less than four US dollars delivered, and in my experience are generally worth about that much. They probably won't kill you or your drive, but may die with a pop the moment you plug one in. Running a drive from a PC power supply is a good alternative; if you've got a spare PC PSU you can get it to turn on in the absence of a PC by using a paper clip to bridge the only green wire on the big ATX connector to any of the black-wired grounds.
My dad has one in his all-singing-all-dancing home PC now, and persists in running scheduled defrags of it once a week.
SSDs wear out when you write to them, right? Doesn't defragging write to the whole disk, so it wears the thing out as fast as anything ever could? Is there any possible reason to do this?
I'm hesitant to say that there's never any possible reason to defrag an SSD, because filesystems are pretty deep magic and there are some extremely perverse possible drive use cases.
SSDs also use "wear levelling" to spread write operations over the entire drive, so if you defrag a 50%-full SSD, moving most of the data on it, you'll not unduly damage half of the drive and leave the other half alone.
But the principal purpose of defragging a mechanical drive is to reduce its "latency", or access time - the "seek speed", the time it takes to get the heads to the right location on the platters, and then the "rotational latency", the time it takes for the right data to spin around under those heads. Modern defraggers also move frequently-used and associated data to the most quickly-accessed part of the platters.
All parts of an SSD's storage are as fast as each other, and even a slow SSD has random seek times of well under one millisecond. (That's about a tenth of the fastest mechanical drives, and about a twentieth of what many drives manage.)
So even if you ran a disk fragmenter, software specifically designed to split files into as many widely-separated parts as possible (here's an old defragger with an option to do that!), the effect on system performance would probably not be noticeable.
Perhaps this will convince your dad. I'm not putting any money on that, though.
When I use IrfanView to view images in a folder that contains MP3s, it insists on PLAYING those MP3s when it gets to them in the file order. I don't have IrfanView associated with those file types, so it doesn't open them if I double-click one, but I can't stop it playing them when it gets to one. Oh, and it opens PDFs too.
How do I stop this?!
In the IrfanView Properties/Settings interface, click the lyrically-named "Load only associated types while moving through directory" radio button.
Sometimes, the Chrome browser starts locking up all of one CPU of my computer - I've got a hyperthreaded 4-core i7, so that means 12.5% utilisation in Task Manager. There doesn't seem to be any particular site that does this; if I've got a lot of tabs open, Chrome just sometimes decides that one page, or if I'm unlucky more than one, needs some really deep thought.
I can see in Task Manager or Process Explorer that this is happening, and if I quit the whole browser and re-run it to re-load all of the tabs then the problem may be solved. I also, however, notice that it often is one particular tab doing it, and when I close that tab the extra CPU usage goes away. But I only find this at random, because there's no way for me to see which tab has the problem in advance!
Is there a finer-grained task-monitor thing that will show me each tab's resource use?
Yes there is, and it's built into Chrome. Just press shift-Escape to bring up Chrome's own inventively-titled Task Manager, which can monitor lots of different things, with CPU usage one of the default columns.
(It's more comprehensive than you probably need in a browser, but the control-Escape Task Manager comes into its own in Chrome OS.)
Certain Web sites give me an "unable to access the network" error in Chrome and analogous ones in other browsers. I just ran Internet Explorer and it says "Internet Explorer cannot display the webpage" with the connection-diagnosing thing saying "the device or resource [URL-of-site] is not responding", and Firefox says "The proxy server is refusing connections", "Firefox is configured to use a proxy server that is refusing connections"... but I don't have a proxy server!
Or maybe I do, that'd be something that malware that wants to display its own special ads or whatever might do, right? Supporting evidence for the malware theory is that my phone, connected to the same home network as the computer, can see the Internet fine. And most Web sites work for both the PC and the phone (as far as normal Web sites work on a phone anyway). But some don't work on the PC.
Among the blocked sites are the City Council site for Penrith NSW, where I live (some other random NSW council sites are blocked too...), and also buchmann.ca, webcom.com, lawlink.nsw.gov.au, patft.uspto.gov and lightbulbs-direct.com.
What malware/other thing is this? I'm at my wits' end!
A bit of back-and-forth with Olav turned up the culprit: PeerBlock, a utility for blocking certain hacking attacks, and self-appointed copyright cops, and various other things. (Olav actually installed Peerblock after I mentioned it in this column a year ago.)
The default PeerBlock setup is extremely paranoid, blocking connections with all sorts of blameless governmental and educational institutions just to make sure it also catches the maximum number of bad actors. And PeerBlock, by default, also blocks HTTP, which is to say normal Web-browsing, traffic along with everything else. The effect of this is a semi-random scattershot blocking of all sorts of useful Web sites.
Can you map a network drive on... your own drive? Like, bind a drive letter to a folder or other drive on your computer, so you can run a program that expects all its stuff to be on e:\programname from a folder called c:\programname, without changing any configuration stuff for the program?
I ask this because I dumped my ancient Eudora e-mail program to this almost as ancient WinXP laptop so I can use it while my proper computer's being fixed, but now it's not in e:\Eudora any more so when I run it it complains and it seems to work but I'm worried I'm messing something up. I suppose I could edit the setup files, but it strikes me that binding E: to C: somehow would be a useful thing to be able to do anyway. Is it possible?
it certainly is, using the good old SUBST command, which became part of MS-DOS getting on for thirty years ago now.
If you open a DOS prompt (just run "cmd" in the Start-menu box in Windows versions pre-8, or sacrifice a rabbit and then type "cmd" in the Start screen of Win8), and type...
subst e: c:\somedirectory
...then that folder on C will now be e: as far as the system is concerned. You'll have to re-run the command every time you reboot, but that's probably the way you want it for a temporary setup like this.
When I go to Sharing in the properties for a folder (Win7 64 bit), the list of people the folder can be shared with when I click "Share" includes "Administrators", my own account, and "Everyone", and "<Unknown Contact="Contact">".
That last one apparently has "Read/Write" access, by default. It doesn't appear in "Manage [user] Accounts".
What IS it?!
There are a couple of ways this can happen, both involving previous Windows installs.
If you installed your current version of Windows on a computer that had some earlier Windows version on it, a user account from that previous Windows can appear as an "Unknown Contact" on the new one.
Also, if you've moved an NTFS-formatted drive from a previous computer to a new one, folders that were network-shared on the old one - which you're likely to be re-sharing on the new one - will have the Unknown Contact on their access permissions list, while other folders won't.
If you've managed to encounter a folder that only Unknown Contact is allowed to access, then you'll have to take ownership of it via Properties -> Security -> Advanced -> Owner. This shouldn't happen, but given the semi-random behaviour of Windows file sharing and permissions, I'd never say never.
Otherwise, you can safely delete the Unknown Contact from the access list right now, or ignore it.