Crash!

Black Hole

May contain traces of nut
I was recording Quantum of Solace in HiDef and watching the Jubilee Concert in HiDef timeshifted, when the HDR crashed! Picture froze in Kylie's set, remote and front panel unresponsive, no response from the WebIF. I have no idea whether QofS is still recording, but I'm leaving it alone just in case.

I switched to live on the TV and found I missed whatever happened between Kylie and Stevie W. I expect there will be a re-run on 301.
 
I have had similar occurrences to this in the past, and the only thing to do was to switch off at the back. In my cases, when this has happened, the box had not recorded what it should have done.
 
So why didn't my box detect the freeze (timeout interrupt?) and reboot then?
Presumably in your case the Humax software didn't crash but something went wrong with the OS. Either a deadlock that caused it to hang or it crashed. I don't think they have configured a hardware watchdog (although it would be possible).

This has happened to me at least once in the last couple of years..
 
There are circumstances in which a timeout interrupt won't help, I never managed to devise a perfect system in all my design experience. The typical arrangement (if Humax went that far) is to have a hardware timer that will create merry hell if it times out, and gets poked periodically from software to restart the time period (obviously, more often than the timeout period).

The trouble is, the software has to be written to poke it from within the various process loops, and if the software gets stuck in one of those loops it just keeps poking the mechanism that is supposed to stop it getting stuck in a loop!

I don't know what the current thinking is, but at one time "proveably secure" computing systems were not allowed to use interrupts, because it is impossible to demonstrate that every single possible situation in which an interrupt could arise will result in proper operation. Secure systems use polling only, so that an input variable is only looked at at specific times and is completely controllable and testable.
 
Linux has a watchdog framework (Google linux watchdog), but it is pretty basic unless the various hardware drivers implement good support for it. The humax kernel was compiled without watchdog support. It was also compiled without kernel logging support so there's no way to find out what happened in your case.
 
Not the same problem but my TVonics HD500 (my other PVRs are Hummies [sp?]) carried on recording HiDef QoS until I stopped it 9 hours later. Curiously it also recorded a Std-Def version on ITV+1 which I had not programmed.

Martin
 
i've had four crashes since the euros started, two diff types

two were whilst watching something and just programme just stopped part the way thru, the scrambled message appeared, box responsive but no other recordings would play at all, power down failed, but once the box had been switched off at the back and left for a few mins all was well

the other two were part way thru an HD recording of the euros, the box was not being used either time and once noticed was completely frozen up, had to be switched off at rear, immediately unfrozen but recordings had ceased at the point, and a power failure message was displayed

both very odd, and out of character for an otherwise stable box with 270gb spare capacity
 
Froze overnight. Left on BBC Red Button as usual, fine last night, this morning totally unresponsive: blue ring, display scrolling frozen, no WebIF. Power cycle required.

Mind you, it was cold last night...
 
Not sure if this is linked to the freezes on this thread but it seems the closest to what's been happening to mine recently ...

HDR is on watching one channel and (usually) recording another, have switched over to watch something else on the TV, leaving the HDR switched on; when I go back to it later (eg an hour) there's nothing showing on the HDMI channel on the TV from the HDR. It hasn't crashed as such - the remote was still operational (could see from the display that changing channel on the HDR was still working) and the recording was still working (as checked later), but there's nothing getting from the HDR to the TV.

This has happened to me 3 times in the last week or so. The cure seems to be to switch off/on at the back of the box.

Any ideas what might be causing it?
 
I never managed to devise a perfect system in all my design experience.
Probably for no other reason than you've not really needed it? I bet you could if you had a real need.

It helps if you design from a failsafe perspective. There are methodologies for that. Mostly way over the top for something like a freeview box though.
The notion that software is always going to be buggy is a fallacy that exists to allow companies to get away with murder (sometimes literally)
I've yet to hear of a single commercial operating system developer using software proof technology for example. (such as Z or even a subset for instance)
You would think a widely used genereal purpose OS would warrant it. Or even require it by law.

As long as there is power on a system I'd suggest using discretes Its generally called a heartbeat watchdog I believe.
I'd be suprised if say the watchdog inside a PIC chip wasn't more than reliable though.
With something like a Humax box my guess is its just sloppy design.
After all linux is pretty rubbish anyway
As are all the 40 year old technologies - including google linux.
 
I never managed to devise a perfect system in all my design experience.

Ah, but at least you know it. After 30 years designing big business systems ( insurance and pensions rather than techy stuff ) I'm glad to be out of it. My impression is that those who now claim near perfection have merely reduced expectations to their miserable achievements. Most big IT projects are failures if measured against original cost, timescale and scope. And I don't just mean public sector.

Probably for no other reason than you've not really needed it? I bet you could if you had a real need.

I don't think so ( no reflection on BH ). Reminds me of an insurance company CEO who believed he could achieve perfection by insisting that every system algorithm was checked by an independent recalculation that did not duplicate any of the methods. Huge costs, no genuine errors detected ( though they existed ), many false alarms. Why? Because the best and most reliable methods we could devise were used in the system so the independent recalcs were inherently inferior. Took me years to convince him.

End of rant - nurse, tale me back to the ward.
 
soc... there is a difference between science and engineering
science is a controlled random negative feedback loop (hence seriosuly dangerous for say - climate modification)
engineering can be precise - mathematical provable.

This is a difference many journalists especially fail to grasp.
I suspect (but dont know of course) that your background is more science than engineering in flavour?
 
jack...... even worse, my degree was in philosophy so I'm really beyond help.

Whilst in principle I accept your distinction I would have to say that in practice most IT practitioners delude themselves when they talk about engineering systems, at least in the commercial world. There just isn't that culture of painstakingly building an established body of knowledge and proven techniques. Instead there is the naive belief that the latest technology makes all the old stuff redundant and today's bright new kids go and make all the same mistakes we did. Eventually the old discipline is relearned ( but with a shiny new name ) but it scarcely moves forward.

We're a bit off topic - maybe one for the Hummy Arms. Call it Old Gits unite.
 
jack...... even worse, my degree was in philosophy so I'm really beyond help.

We're a bit off topic - maybe one for the Hummy Arms. Call it Old Gits unite.
And there was me thinking you were: Sócrates Brasileiro Sampaio de Souza Vieira de Oliveira
 
And there was me thinking you were: Sócrates Brasileiro Sampaio de Souza Vieira de Oliveira

Shhhh - don't tell everyone. You've no idea how much it costs to fake your own death and start a new life. I never wanted to be a footballer........ I wanted to be.........


A LUMBERJACK!

All together now...
 
Back
Top