08
Feb
2016

Virgin Media “SuperHub 2” not being very super

We’ve just been sent a replacement hub, and everything was working just fine when I set it up. Speeds were great, and WiFi worked just fine.

Then after a while, it stopped responding. It showed up as WiFi but it wouldn’t let devices connect. It got even weirder when I couldn’t even get an IP address for a directly attached PC. Strange…
I rebooted it, reset it, did all kinds of cable configurations to try and get it to work. Then, I noticed something:

I think its been dropped in shipping. Classic!

I might try opening it up to see if there is a loose cable, but I’d rather Virgin swapped the box for a new one. I don’t want this problem manifesting itself anymore. The box works just fine if it’s laid on its side!

Weird or what? I hate IT equipment sometimes.

07
Jan
2016

Windows 10 Crashing at Idle: Solved!

This is a weird one. Before, I posted about how Windows 10 after being left for about 10 minutes would bluescreen, with no idea why. Each bluescreen was slightly different, though they seemed to point to NTFS.sys.

wp-1450434508926.png

Strange. I had reinstalled the OS and it was still crashing. I reset the BIOS to it’s default values and they seemed to be fine, and Linux was perfectly stable. I went on the Google and found some tools for analysing the memory dumps that the bluescreens leave, the best tool that I used was WhoCrashed.

The tool told me that it was unlikely to be a hardware issue and more of a driver issue, like I had suspected before. I rebooted into Safe Mode with Networking and tried again, but this time the system did not crash after 10 minutes and seemed to be totally stable. Safe Mode runs Windows with a lot of services disabled, so I started looking at what services ran after 10 minutes of idle in a normal environment. Then, I found it.

Windows Defender.

Yep, Windows 10’s own built in anti-malware service was running after 10 minutes and using up a load of CPU and disk activity, before eventually crashing the system. Now, in Windows 10, you cannot turn off Windows Defender. You can temporarily disable it, but it gets turned back on automatically – so I can’t just use another (probably better) solution. I did some more research and worked out that indeed some power saving settings in the BIOS were causing interference. I’m not sure why, but having the C6 states enabled in the BIOS would cause Windows 10 to crash when Windows Defender needed disk access and CPU time.

tm1

So that was strange. Since changing that setting I’ve not had any problems with Windows 10, and Grand Theft Auto V doesn’t crash anymore. Maybe it was a hardware problem after all? Aw well, it’s fixed now.

20
Oct
2015

Cold boot problems: Gigabyte GA-78LMT-S2P

So I have been running my University PC system for about three years now, and it has been extremely reliable since day 1. But recently, I have been having a problem.

DSCF0243

Above: Inside the University PC with the Fractal Design Arc Mini case. The mainboard is the Gigabyte GA-78LMT-S2P revision 5.x.

The problem only occurs when the PC is cold (room temperature or below). It refuses to POST. I press the power button and it powers on and lets the drives spin up, before shutting off again. I find that pressing reset during this period helped for a while – the PC would then POST, but this has stopped working now and it just shuts off again. The power button is totally unresponsive at this point too – you have to turn the power supply off and wait for the capacitors to discharge enough.

So I’ve had to start using a hair dryer to heat up the middle of the motherboard and after about 15 seconds of heating the system boots up with no problems. I aim the heat right at the centre of the board, where the silver northbridge heatsink is. I think that is the problem – it is a BGA chip that could have some soldering issues that are manifesting over time.

Subsequent reboots are fine, the problem only occurs when the system is left to cool down. Putting the system to sleep is also fine – there are no problems waking up.

With the cold winter months rapidly approaching, I suspect I am going to have more and more problems with this board. I am looking to upgrade to the ASRock 970M Pro3. Then I can do some experiments with this Gigabyte board and figure out exactly where the problem is and maybe attempt a repair (not likely!).

05
Oct
2015

The Windows Registry: Vulnerabilities

So once again, I’ve had problems with Windows. I’m running good old faithful Windows 7 SP1, and whilst I was able to use the OS normally, I couldn’t open CCleaner. Or regedit.

Alarm bells ringing, I rebooted into safe mode as fast as I could. Some sketchy stuff was going on. I used Safe mode with networking to download Malwarebytes AntiMalware (I actually had it already installed, but I downloaded it again to be sure). Then something else happened.

I couldn’t run Malwarebytes. Or regedit. Or CCleaner. In safe mode. At this point, I let out a long sigh. Why, Windows, why do you drive me closer and closer to insanity. I run Linux as my main OS for many reasons, and this just got added to the ever expanding list.

So I was ready to download an ISO and reinstall from scratch. But I wasn’t done yet. Look at the screenshot below:

bs3

See what I have done? I renamed mbam.exe to notav.exe (any other name would have been fine, but I wanted to fool whatever mechanism was preventing me from running the exe). Luckily for me, Malwarebytes fired straight up under the different filename. I updated the database and ran a scan pronto:

bs1

bs2

Malwarebytes, you beauty! It found the offending registry keys and was able to remove them. You can see that ccleaner.exe, mbam.exe and regedit.exe, along with a load of other antivirus programs, are listed here to disallow execution (or shove into the debugger, I’m not sure why there are Debugger lines). It didn’t find anything else, so I don’t think any extra malicious exe’s were ran. I also ran an AVG scan and that came up clean, so I think I got away with something there.

I think I know where this came from. I was messing around with some dodgy files, and I think one of them bit. What really annoys me here though, is that at no point did I allow Administrator rights. I believe the registry changes were made from the user-level, which is pretty awful in terms of security. Now I think back, Malwarebytes no longer ran at startup but I didn’t think much of it. Now I know why.

So to sum up: I have been extremely lucky here. If this kind of manipulation can be done from the user-level, I could not imagine what else could be done. The Windows 7 install I use is just purely for gaming – I do no productive stuff and I don’t keep anything of importance on my Windows drives.

If you find that you can’t run your antivirus software, or CCleaner or regedit – do what I did, get Malwarebytes and rename mbam.exe to something else. It’ll be allowed to run then and clear out this registry crap.

You can avoid this by not being a plonker (like me) and not running slightly dodgy files… but that should be common sense by now. Stay safe everybody!

27
Sep
2015

Seagate Barracuda 7200.14 500GB: Bad sectors!

So its finally happened:

Screenshot_2015-09-26_17-23-54

One of my ST500DM002 (AKA 7200.14) 500GB drives has finally developed some bad sectors. I have already started a scrub on these drives just to be sure that my data is still consistent.

These drives have done very well in my opinion since they have done over 7,600 hours of runtime. I am running RAID-5, which should hopefully keep me protected in the event that one of these drives outright dies (which I find unlikely, these modern drives generally show some early warning signs, and I don’t class 8 bad sectors as a huge problem, yet).

Screenshot_2015-09-26_17-24-37

I am going to keep an eye on this drive and make sure that the bad sector count is not increasing rapidly, as this is a sign of a dying drive. I hope to be upgrading this array soon, since I am seriously running out of space!

13
Aug
2015

The dodgy micro-SD card: The Final Chapter

So I have been using a somewhat dodgy micro-SD card I bought from Rakuten (more on this here and here):

I have decided to do a complete analysis of what the card actually is. To Ubuntu, it is reported as so:

Screenshot_2015-08-11_14-57-32

“17GB”. Hmmmm. When I was using this for my Raspberry Pi 2, I partitioned it such that I only used half of the drive’s space (about 8GB) and it seemed to work just fine. So, lets get to the bottom of this.

I have partitioned the full space as ext4, and I am using a program called F3 (Fight Fake Flash) to test the drive. You can get F3 from here.

I compiled the software just by typing make into the unzipped directory, and ran the software with the following command:

./f3write /<wherever you drive is mounted>

This starts off the write phase of the testing. This fills the drive up at the filesystem level with test files, and does not overwrite existing files, so make sure you start with a blank drive when using this method. You can do this test at the block level using f3probe, but this is still experimental and not recommended.

This is the output I got for f3write:

./f3write /media/john/…/
Free space: 15.22 GB
Creating file 1.h2w … OK!
Creating file 2.h2w … OK!
Creating file 3.h2w … 14.90% — 2.45 MB/s — 57:37

And now we wait for the drive to fill up! This looks like it will take about an hour. Fake drives always report success on writing data, even though it could be just throwing data away when it reaches the actual physical limit of the internal storage. Also, this is meant to be a “Class 10” drive. 2.45MB/s is not “Class 10”. Its not even Class 6 or Class 4. Already smelling a rat here…

This is what you see when it has finished:

Creating file 14.h2w … OK!
Free space: 815.95 MB
Average writing speed: 2.64 MB/s

So, the test files have been written. Now, its time to verify, using f3read:

./f3read /<wherever you drive is mounted>

And you should see the files being analysed. Look in the corrupted column to see when these test files become corrupted!

It is finally time to see what this dodgy microSD card really is. Here is the output:

Screenshot_2015-08-11_17-12-15

Well well well. This microSD card only has 5.36GB of usable space. How that works I do not know, but I do know one thing. This card is going straight in the bin. This made me think – I had set the partitioning to use 8GB, when clearly this wasn’t small enough to prevent data corruption. My Raspberry Pi 2 seemed to work just fine, but it was probably because the Pi wasn’t using more than a couple of GB of space. If it went over 5.36GB (or wrote something into the corrupted area), then it would have been game over!

So make sure you get decent SD cards (or any type of flash storage) from a reputable manufacturer. If a 1TB USB drive for £20 seems too good to be true, it probably is.

08
Jun
2015

Minecraft Server Munin Plugin: Strange behaviour

I decided to use Munin to monitor the Minecraft Server but there is some strange behaviour:

image

image

image

image

As you can see, only the users plugin works! Strange, I will post the fix to this in the future once I figure out what’s going on!

28
May
2015

Dodgy SD card: more testing

Having used the cheapo “Class 10” microSD card for a bit with the Raspberry Pi 2, I decided to flash the card with another image, using “Disks”, and got the following screenshot:

Screenshot_2015-05-28_16-08-53

Firstly, 17GB is a bit of a dodgy number for a 16GB card, so something definitely isn’t right there. Also, for a “Class 10” card, 6MB/s makes me think it’s actually Class 6.

Strange, I will have to investigate further!

13
May
2015

ownCloud update to 8.0.3 Web GUI errors fix!

I logged into ownCloud as usual and found that the normal ownCloud Desktop Client worked just fine, but I could no longer access the Web GUI. All I got was a blank screen with no errors shown or anything.

I looked in the ownCloud Apache log file (mine was /var/log/apache2/owncloud.log) and found that errors were being thrown:

[Wed May 13 16:01:58.139858 2015] [:error] [pid 7879] [client x.x.x.x:xxxxx] PHP Fatal error:  Class OCA\\Contacts\\Share\\Addressbook contains 1 abstract method and must therefore be declared abstract or implement the remaining methods (OCP\\Share_Backend::isShareTypeAllowed) in /var/www/owncloud/apps/contacts/lib/share/addressbook.php on line 130

Weird. After some Googling I found that a lot of people had the same problem. It turns out its an error associated with the calender/contacts app, as I found here. You should disable these apps BEFORE you update to stop this from happening, but if you already updated, follow the steps below to get the Web GUI back:

  1. Open your mySQL database editor (I use Webmin) and edit the ownCloud database:
    Screenshot_2015-05-13_16-08-27
  2. Go into Edit table and select oc_appconfig.
    Screenshot_2015-05-13_16-09-24
  3. Press View Data at the bottom of the form.
    Screenshot_2015-05-13_16-10-19
  4. Select the rows containing Calender enabled and Contacts enabled and press Edit selected rows.
    Screenshot_2015-05-13_16-11-45
  5. Change the two configvalues to “no” and press Save.

That’s it! I got my Web GUI back!

02
May
2015

Follow up: Munin now showing Apache activity with WordPress

I posted how to fix Munin reporting for Apache when WordPress is installed a while back, and some data has been logged now:

image

All sorted! I’ve even added a monitor widget to my home screen:

image