linuxMCE is still not ready for real life!

[totallymaxed]

When I have had lock ups - only since RC2 - it is probably just the interface locking which can make it difficult to get back in and so seem like a "hard" lock. But on at least one occasion I could absolutely verify that the core was still routing traffic normally between internal and external LANs so it definitely was not a real "hard" lock. The other occasions, I wasn't able to test this. But on at least 2-3 occasions the lock up seemed "harder" and the LEDs on my keyboard were flashing.....

Hard or soft, I definitely feel that RC2 is significantly more unstable even than RC1, and definitely more than b3/4 (which I spent more time on) Don't know why - gut says TV card, VDR or nVidia drivers, but my gut says a lot of things I don't listen to! There are many other reports I have read that suggest a connection with RC2, but nothing definitive.

BTW, yes I have rebuilt a few times. Started on AMD64 and have now moved to i386 for the last couple of builds...

If your Core is still routing traffic then its almost certainly an nVidia binary driver issue and X has crashed (try ssh'ing in from another machine)...see my earlier post in this thread where I mention this.

If you get 'two LED's on your keyboard' flashing then that is a real 'hard lock'... and i see this now and then on my Kubuntu 8.04 desktop machine.

We use a mix of i945GNT mobo's (we sometimes use these in bigger racked Cores where we need as many slots as we can get) and various AMD2 based mobo's... apart from the nVidia driver lockups (which are pretty rare) we see excellent stability (and we have a lot of systems out there... so believe me if we had the reliability issues being discussed here I'd 100% know about it!!

So my somewhat simplistic and less than scientific assertion is therefore that whatever is causing the problem being discussed here has to be specific to the hardware/configurations being used.

Andrew

[royw]

On my gentoo workstation/NFS file server, I've tickled a weird NVidia/JMicron bug. 

Basically the 8600GTS nvidia card and one of my Gigabyte X48-DQ6 motherboard's SATA controllers (JMicron 20360/20363) apparently share IRQ 16.  Something goes wrong about twice a week and the kernel (2.6.25) traps an error (irq 16: nobody cared), then disables IRQ 16.  This leaves the X server in a bad state (understatement) and disables the motherboard's ethernet ports.  The work-around I found googling (http://www.nvnews.net/vbulletin/showthread.php?t=115703) is to not use the JMicron controller.  I'm at +1 days trying this so don't know for sure if this is the solution/problem.

A quick check for this problem is to grep your /var/log/kern.log for "Disabling IRQ".

If you happen to notice what time a lockup occurred (after weeks the problem finally happened while I was using the system), then start examining the logs around that time and google any questionable entries.

HTH,
Roy

[colinjones]

Really quick post - just came home from the pub and had the keyboard LEDs flashing "hard" lock up issue again. Let me state with absolute certainty, I have not every had this issue with exactly the same hardware, no changes whatsoever, until RC2. I don't know how much more clear I can be. It never happened before. It happens "occasionally" now. That simple. Something has changed, and it is definitely Not overheating!


I realise that many of the experienced/devs here may not be experiencing this issue. But I'm not stupid, there are several others seeing the same thing, and it started between RC1 and RC2. At some point you have to accept the weight of averages suggest there possibly could be a connection...

BTW - I haven't installed any new nVidia drivers, and haven't for quite some time .... just the ones that come with 0710 (169). Also, my xorg is exactly as it has been for months...

The only thing I have changed is added the nVidia 7300GT card, but the lockups started before that.... not my biggest issue, but thought I should be clear about this...

[hari]

At some point you have to accept the weight of averages suggest there possibly could be a connection...

of course there is a possibility. But without hard facts/debugging there is no way to find out...

I'd start with removing every peripherals, tuners, whatever. Run the bare minimum and see if it is stable. Then add pieces and look where it breaks.

best regards,
Hari

[colinjones]

Hari - so I'm thinking given that it happened again, maybe I can catch it and get the dump files. I noticed the other day a stack of dump files... can't remember which folder, but I can probably find it again. I haven't turned anything specific to debugging on, are these the files we are talking about, or do I need to turn something else on beforehand to capture the debugging info?

[Zaerc]

Yeah yeah yeah, lmce is not ready because you guys have heat problems... 

Learn how to build a decent PC, or give it a rest.

Zaerc..  you give it a rest...  it's not a heat issue on mine..  otherwise I would be getting lockups when running xbmc/linux on the same hardware..  opengl 3d rendering for the gui and playing HD videos with no problems..  I'm getting lockups with lmce from sitting cold all night turned off, booting up and not even getting anywhere to generate any heat...

There are odd hardware issues that only relate to lmce even when using supposedly "supported" hardware..  for those of you lucky enough to have picked the "right" hardware and never experience these issue I'm sure it's hard to believe it's a lmce problem...  but it is...  it's too widespread to be denied....

I already gave it a rest, but you fuckups have to keep bringing this stupid thread back up.  And yeah I find it very hard to believe this is a lmce specific problem when only a few individuals are having symptoms that seem similar but probably aren't even related. 

If your system fails during boot, then you obviously have a hardware problem, but instead of asking for help and guidance you people choose to bitch about it in this thread blaming lmce for your own incompetence...

[Afkpuz]

Ok, I'm chasing the gpu temperature idea.  So far, looks promising.  Although, I might have changed too many options at once.  I took out the tv tuner which was massive and was in very close proximity to the video card.  I also have a box fan blowing air into the open case.  So, if this setup works, the problem could have been temp, or the tuner card.

[totallymaxed]

At some point you have to accept the weight of averages suggest there possibly could be a connection...

of course there is a possibility. But without hard facts/debugging there is no way to find out...

I'd start with removing every peripherals, tuners, whatever. Run the bare minimum and see if it is stable. Then add pieces and look where it breaks.

best regards,
Hari

I agree with Hari. Strip your system back to its most basic config and then run it like that and observer its stability. Then systematically add back single items and test again. Its what we do here when we're debugging a hardware config... it works if done rigorously.

Andrew

[colinjones]

Hari/Andrew - I don't have the luxury, the only system I have is in "production" otherwise I would do exactly that. I am in enough trouble at the best times from the other half I'm prepared to ride it out as it only happens once every couple of days or so (if I don't reboot) but I still need an answer to my previous question about getting dump files when it does happen.... do I need to enable anything, or are the files I mentioned sufficient to debug from? (the file names seem to mention process names from memory...)

[hari]

Hari/Andrew - I don't have the luxury, the only system I have is in "production" otherwise I would do exactly that. I am in enough trouble at the best times from the other half I'm prepared to ride it out as it only happens once every couple of days or so (if I don't reboot) but I still need an answer to my previous question about getting dump files when it does happen.... do I need to enable anything, or are the files I mentioned sufficient to debug from? (the file names seem to mention process names from memory...)

luxury? WTF, you put a broken system into production and now don't fix it but reboot it every other day?
I'm glad you don't run my infrastructure :-)

Yeah, you need to enable kernel crash dumps.

best regards,
Hari

[colinjones]

its only a little bit broken walking on eggshells here WAF-wise ... cool will look into how to enable those... thx

[los93sol]

Yeah yeah yeah, lmce is not ready because you guys have heat problems... 

Learn how to build a decent PC, or give it a rest.

Zaerc..  you give it a rest...  it's not a heat issue on mine..  otherwise I would be getting lockups when running xbmc/linux on the same hardware..  opengl 3d rendering for the gui and playing HD videos with no problems..  I'm getting lockups with lmce from sitting cold all night turned off, booting up and not even getting anywhere to generate any heat...

There are odd hardware issues that only relate to lmce even when using supposedly "supported" hardware..  for those of you lucky enough to have picked the "right" hardware and never experience these issue I'm sure it's hard to believe it's a lmce problem...  but it is...  it's too widespread to be denied....

I already gave it a rest, but you fuckups have to keep bringing this stupid thread back up.  And yeah I find it very hard to believe this is a lmce specific problem when only a few individuals are having symptoms that seem similar but probably aren't even related. 

If your system fails during boot, then you obviously have a hardware problem, but instead of asking for help and guidance you people choose to bitch about it in this thread blaming lmce for your own incompetence...

...That was certainly warranted...I guess you didn't get enough attention when you were little or maybe some adult male figure in your life used to play "doctor" with you.  Either way, back off and quit taking it out on the internet...nobody cares...

[niz23]

Yeah yeah yeah, lmce is not ready because you guys have heat problems... 

Learn how to build a decent PC, or give it a rest.

Zaerc..  you give it a rest...  it's not a heat issue on mine..  otherwise I would be getting lockups when running xbmc/linux on the same hardware..  opengl 3d rendering for the gui and playing HD videos with no problems..  I'm getting lockups with lmce from sitting cold all night turned off, booting up and not even getting anywhere to generate any heat...

There are odd hardware issues that only relate to lmce even when using supposedly "supported" hardware..  for those of you lucky enough to have picked the "right" hardware and never experience these issue I'm sure it's hard to believe it's a lmce problem...  but it is...  it's too widespread to be denied....

I already gave it a rest, but you fuckups have to keep bringing this stupid thread back up.  And yeah I find it very hard to believe this is a lmce specific problem when only a few individuals are having symptoms that seem similar but probably aren't even related. 

If your system fails during boot, then you obviously have a hardware problem, but instead of asking for help and guidance you people choose to bitch about it in this thread blaming lmce for your own incompetence...

...That was certainly warranted...I guess you didn't get enough attention when you were little or maybe some adult male figure in your life used to play "doctor" with you.  Either way, back off and quit taking it out on the internet...nobody cares...

ENOUGH!

This does not make the project getting forward.
I agree with both of you. Sort of.

In short:
Any hardware that is not listed as compatible may or may not work.
That mean your are on your own if you decide to use untested/unsupported hardware.
I most cases hardware similar to the supported ones will work out of the box but in other cases there is this "noname" manufacturer that does not know how to build decent hardware.
Have seen issues with Linux earlier were it just worked with Windows even though the chip/card was listed as working under linux.


Basic hardware fault searching:
Remove everything that you can remove except what is needed to make the machine running.
Enable kernel crasch dump.
Search on google or other forums if anyone else have run it on linux before or if there are known problems.


/niz23

[hari]

[...] or maybe some adult male figure in your life used to play "doctor" with you.

los93sol, you are a sick guy. That's really nothing to make fun about.

Hari

[golgoj4]

[...] or maybe some adult male figure in your life used to play "doctor" with you.

los93sol, you are a sick guy. That's really nothing to make fun about.

Hari

That is the problem when the thread gets derailed isnt it?

Lets move past all the BS (yes its BS) and try to get to the bottom of this issue. Do we need a clean (both in feelings and content) thread to start documenting all of this stuff?

Maybe every person who has this issue need to include in their post their system spec.

Mobo            -asus m2nvpn
Processor     -amd 4800
Video Card   -onboard 6150
Peripherals   -Hauppage PVR-500
                     -2nd Nic - realtek pci gigabit network adapter.
?

Lockup only occurs on my core/hybrid at the moment. After running TOP, xorg is @ 100% but all the other MD's, networking and orbiters work fine.


Maybe if we actually focus on something useful instead of putting down others we can get somewhere.