I had issues with large numbers of parity errors not too long ago and thought I solved the issue since I discovered a bad ram chip.

See this thread: http://lime-technology.com/forum/index.php?topic=2727.0;topicseen

 

I no longer get the parity errors but the system stops responding after a few days of running.  I cannot access it via network nor directly connected.  Nothing on the screen at all.  No response to pings.

I created a log last time I rebooted it but i don't think that will help much since my understanding is that it overwrites the log everytime it boots.

 

How can I troubleshoot this if I can't get to a see a log since the system freezes up?

 

My config is:  It was built with all parts in the recomended list, Gigabyte GA-EP35-DS3R , Intel pro 1000 net adapter, SuperMicro AOC-SAT2-MV8,  Cooler master Centurion case with 2 ICY Dock 5in3 backplanes, celeron CPU and 1GB of Crucial Ballistic RAM.  I had 2 - 1GB chips installed but one tested bad with the mem test on Unraid boot so i removed one which leaves a single 1GB Ballistic chip installed.  System is ALWAYS very cool to the touch since it has quite a few fans in it.

 

Any help would be greatly appreciated.

 

Thanks!

Nothing on the screen at all.

 

Can you clarify what you mean by this?  There is usually always a screen full of messages viewable.  Do you mean that at the point of freezing up, the screen is suddenly cleared, completely blank?  Does the NumLock key toggle the NumLock keyboard light?

There are no messages on the screen of any kind.  It is COMPLETELY blank.

It just happened again.  I had rebooted a few hours ago and it started a parity check.  It was about 75% done and around 140 parity errors when I first posted this issue.  It appears that I am back at square one.  last parity check was only 3 errors and before that there were no errors.

I powered off before I saw your comment to check the numlock light.  I can check during the next freeze.

 

This will get much worse now.  Everytime I power it on, it starts a parity check and each time the number of errors grow substantially.  It will become unresponsive during the parity check causing me to power it off again and perpetuating this cycle.  This is exactly what happend last time and then i did the memory test which indicated a bad memory module.  perhaps it is not the memory and instead is the motherboard?

 

If I cancel the parity check, I can usually play a basic DVD but if I play too much or copy too much data to the array, it will become unresponsive again.

 

Thanks for any help you folks can offer.

Since the parity check is essentially not working correctly, there's no point in running it.  You might as well cancel it each time it starts, until this problem is resolved.

 

Would you mind clarifying what you do see on the screen, *before* it freezes or becomes unresponsive.  Are you seeing the regular messages there?  Is it not allowing you to capture your syslog then?  And does the screen suddenly go blank?  Does the monitor's power-on light change to the standby color (usually amber), or is it still green?

Update:  let the 2nd memory chip run through the test overnight and more memory errors reported.

I highly doubt that both memory modules went bad unless some sort of issue with motherboard or power.  Perhaps compatibility?

I have two replacement 1gb chips that I will install tonight to test.  Different brand.

 

 

I did not realize that the parity check was not working correctly.  Is there a post I can read up on this?  i will search.

 

I periodically checked the screen while it was doing the parity check and the screen was in power saving mode, yellow light.  Tapping the keyboard powered up the screen and it was resting at the Login prompt ready to go.  I could still login at this time.  Came back an hour later and no response from keyboard and monitor was black background but the power light was green so it did not seem that it was in standby.

I just saw on the bottom of the hardware compatibility WIKI that Western Digital AAKS drives might cause this issue if using the power down mode which I am.

I believe that I have at least two of these and perhaps 3 in my system.  i will have to check tonight.

 

Update:  let the 2nd memory chip run through the test overnight and more memory errors reported.

I highly doubt that both memory modules went bad unless some sort of issue with motherboard or power.  Perhaps compatibility?

I have two replacement 1gb chips that I will install tonight to test.  Different brand.

Is it possible your motherboard is not set up for the voltage ant timings needed by your "Crucial Ballistic RAM" ??

 

That would explain why both RAM strips are failing..  From the Crutial site, it seems most of the "Ballistix" RAM takes 2.0Volts.  Did you check if your BIOS is set correctly... It might not even have the correct timings.

 

Incorrectly set voltages and/or timing would make the RAM look bad, even if it is not.  That would make anything run on the MB look bad.  If your memory test is telling you you have errors, no sense in doing anything else till they are resolved.

 

Joe L.

I was wrong on the hard drives.  There are no WD HDs at all.  3 Hitachi 1TB's and several 400GB of Samsung and Seagate.

I changed out the RAM to different brands and rerunning the MEMtest86.

I did not realize that the parity check was not working correctly.  Is there a post I can read up on this?  i will search.

 

I'm sorry I was not clearer.  There are no issues with the unRAID parity check, as to how it works, and its results.  I was only commenting on your system's inability to run the parity check reliably, whether from memory or motherboard or other cause.  Since it was reporting a new set of many parity errors on each run, something was obviously very wrong, and those results were completely untrustworthy.  There was no point in running it, until you have achieved system stability.  If the system is running correctly, you should NEVER get parity errors on repeated runs.  If you do, then there is bad memory or bad motherboard or bad cabling or a drive is going bad or something is way too hot.  I can't think of any other reasons.

 

It does sound as if you are on the right track to narrowing down the problem.  If this new memory tests well, and you still get more parity errors on a second run of the parity check, then you will have to suspect the motherboard.  That unfortunately has been at fault in one or 2 other cases.

 

The reason I say 'on the second run', is that because of the issues you have had, with memory or whatever, it is very probable that the previous parity checks WRONGLY 'corrected' the parity info, so the next parity check you do will have to correct the 'corrections'.  But after it does, and the system is finally running correctly, then a followup parity check should be perfect with NO errors.

Would the command ...

 

tail -f /var/log/syslog>/boot/syslogtail.txt

 

catch the last log messages of a crashing server?

another option would be to use a syslog server like kiwi and tell unraid to pipe syslogs over the network real time to your pc

Update:  Swapping out the memory resulted in no memory errors after nearly 3 hours of testing.

I was able to copy around 500GB worth of stuff over to the array without incdent. 

Parity check is OK with 0 errors.

Should I be concered that I copied over nearly 4TB of data with memory that was not correctly set up????  Where would this bite me?  Most of the data are my DVD's transferred to disk for my HTPC to play.

 

Joe L. : Initial thoughts are that you were on to something with the voltage.  i did not realize that this was "performance" memory that required a higher voltage than standard.  I did not see an option for upping the voltage in the MB BIOS.  It was clearly set lower than 2V.  I did see plenty of timing options but I was not about to play with the settings as I am not familiar with those options.  I really doubt that 2 sticks of Crucial memory was bad.  Must have been the voltage.