[dba-SQLServer] How to troubleshoot a blue screen

jwcolby jwcolby at colbyconsulting.com
Tue Jun 19 15:27:42 CDT 2012


I have used Memtest64 dozens of times over the years.  Yes any single "can I access the memory at 
all" test only takes a "little while", however to do the full set of tests (adjacent bits, patterns 
and so forth) is not fast.

Furthermore the memory is ECC memory anyway.  I have no idea what Memtest86 is going to do with 
that.  If there is an occasional error the ECC circuitry is supposed to handle it automatically and 
not even report it to MemTest86.

I believe that the issue is probably a disk dying.  Now when it blue screens it fails to post, 
giving errors about talking to SATA channel 3.

It seems I was silly enough to try and use an SSD for my boot disk.  I have migrated the boot to a 
rotating disk and am back to work.  The only way I will know is if the blue screens stop occurring.

John W. Colby
Colby Consulting

Reality is what refuses to go away
when you do not believe in it


On 6/19/2012 2:02 PM, Francisco Tapia wrote:
> LOL,
>    well Ram is fast, and it doesn't take that long to run a test (be it 64gb
> or not) the cycle test is what can take some time, but a few hours would
> suffice for letting you know if there are any errors.  you only really need
> to do the full cycles if you want to ensure that it's fully burned in,
> something you've already taken care of.
>
> http://forum.notebookreview.com/notebook-dummy-guide-articles/142746-testing-your-ram-memtest86.html
> (steps for memtest+)
>
> Memtest86+ is very easy to use. All you need to do is:
>
>     1. Download the file. I'm going to walk through using the "Pre-Compiled
>     Bootable ISO (.zip)" file. You can get that from this
> page<http://www.memtest.org/#downiso>
>     .
>     2. Unzip the file. There's only 1 file in there, called
>     "memtest86+-X.YY.iso"
>     3. Burn the ISO file to CD. This is the potentially tricky part. An ISO
>     file is a disc image, and cannot be burned directly to a CD. Your CD
>     recording tool should have an option like "Burn image to CD" or something
>     like that. If you can't find it, you can try the free ISO
> Recorder<http://isorecorder.alexfeinman.com/isorecorder.htm>powertoy.
> After you burn the CD, if you put it back in the drive and see
>     the file "memtest86+-X.YY.iso" listed, you've done it wrong
>     4. Shut down your computer completely, then unplug it. *DO NOT DO THIS
>     IN SLEEP OR HIBERNATE MODE!!!*
>     5. I recommend testing the modules one at a time, so if one is bad, it's
>     easy to know which one! Take out all but one of the RAM modules.
>     6. Plug in the computer, put the CD in the drive and reboot. Make sure
>     it boots off the CD, not the hard drive. You should see a blue screen with
>     a green bar on the top that says: "Memtest86+". There are screen shots on
>     the memtest86+ web site <http://www.memtest.org/#screen> so you can get
>     an idea.
>     7. Now, you wait for the test! The test will *run forever* until you
>     stop it. You should run it for at least 1 full pass, more if you have the
>     time.
>     8. Once the test is done for that RAM module, shut down, unplug, take
>     out the module you just tested, put in the new one, and test again.
>
>
>
> so not a lot of time, and iirc the test will tell you wich bank failed if
> there is an error (but if you don't know your board well enough testing one
> at a time will save headaches).
>
>
>
> -Francisco
> http://bit.ly/sqlthis   | Tsql and More...
> Let Apple know you want turn by turn on your iPhone 4!
> https://plus.google.com/112803888444646122406/posts/HmQej9pHZQM
>
> <http://db.tt/JeXURAx>
>
>
>
>
> On Tue, Jun 19, 2012 at 9:54 AM, jwcolby <jwcolby at colbyconsulting.com>wrote:
>
>> Yea, and I am going to do that.  With 64 gigs of memory it will be more
>> than "a few hours"!  ;)
>>
>> It may be that the problem was a SSD which died Monday morning.
>>
>>
>> John W. Colby
>> Colby Consulting
>>
>> Reality is what refuses to go away
>> when you do not believe in it
>>
>>
>> On 6/19/2012 9:17 AM, Francisco Tapia wrote:
>>
>>> if you have some time, you can test the RAM easily by just downloading the
>>> memtest application to a cd rom boot to it let it run for a few hours, if
>>> it's ram related it'll show up in the monitor.
>>>
>>> http://www.memtest.org/
>>>
>>>
>>>
>>> -Francisco
>>> http://bit.ly/sqlthis   | Tsql and More...
>>> Let Apple know you want turn by turn on your iPhone 4!
>>> https://plus.google.com/**112803888444646122406/posts/**HmQej9pHZQM<https://plus.google.com/112803888444646122406/posts/HmQej9pHZQM>
>>>
>>> <http://db.tt/JeXURAx>
>>>
>>>
>>>
>>>
>>>
>>> On Sat, Jun 16, 2012 at 7:53 AM, jwcolby <jwcolby at colbyconsulting.com>**
>>> wrote:
>>>
>>>   My SQL Server blue screens about once a month or so.  It does not appear
>>>> to be heat related, or at least CPU heat related as I have top flight
>>>> HSFs
>>>> which hold the temps below 35 C.  The temps in the case stay around that
>>>> temperature or below.  The blue screens appear to be SQL Server related,
>>>> as
>>>> they seem to be triggered by my doing something in SSMS.  This last time
>>>> I
>>>> started an update query and it immediately blue screened, by which I
>>>> mean I
>>>> clicked start (the query) and the blue screen occurred.
>>>>
>>>> This is an AMD dual cpu server, both CPUs populated.  32 gigs ECC RAM on
>>>> each side.  Being a server there is no overclocking.  There is a lot
>>>> going
>>>> on hardware wise however.  I use an Areca 16 port RAID controller, RAID 6
>>>> arrays built from 1 tb WD black drives.  I am using SSDs to create a
>>>> Raid1
>>>> array hosting several of my Read Mostly databases.  The specific update
>>>> that caused the blue screen was not on SSD however, it was on the Areca
>>>> rotating media raid 6.
>>>>
>>>> I am at a loss on how to troubleshoot.  It "feels" like it must be a
>>>> memory problem, and yet the memory is ECC.  Perhaps I have a bad DIMM
>>>> which
>>>> just flakes out.
>>>>
>>>> --
>>>> John W. Colby
>>>> Colby Consulting
>>>>
>>>> Reality is what refuses to go away
>>>> when you do not believe in it
>>>>
>>>>
>>>> ______________________________****_________________
>>>> dba-SQLServer mailing list
>>>> dba-SQLServer@**databaseadviso**rs.com <http://databaseadvisors.com> <
>>>> dba-SQLServer@**databaseadvisors.com<dba-SQLServer at databaseadvisors.com>
>>>>>
>>>> http://databaseadvisors.com/****mailman/listinfo/dba-sqlserver<http://databaseadvisors.com/**mailman/listinfo/dba-sqlserver>
>>>> **<http://databaseadvisors.com/**mailman/listinfo/dba-sqlserver<http://databaseadvisors.com/mailman/listinfo/dba-sqlserver>
>>>> **>
>>>> http://www.databaseadvisors.****com <http://www.databaseadvisors.**com<http://www.databaseadvisors.com>
>>>>>
>>>>
>>>>
>>>>   ______________________________**_________________
>>> dba-SQLServer mailing list
>>> dba-SQLServer@**databaseadvisors.com <dba-SQLServer at databaseadvisors.com>
>>> http://databaseadvisors.com/**mailman/listinfo/dba-sqlserver<http://databaseadvisors.com/mailman/listinfo/dba-sqlserver>
>>> http://www.databaseadvisors.**com <http://www.databaseadvisors.com>
>>>
>>>
>>>
>>
>> ______________________________**_________________
>> dba-SQLServer mailing list
>> dba-SQLServer@**databaseadvisors.com <dba-SQLServer at databaseadvisors.com>
>> http://databaseadvisors.com/**mailman/listinfo/dba-sqlserver<http://databaseadvisors.com/mailman/listinfo/dba-sqlserver>
>> http://www.databaseadvisors.**com <http://www.databaseadvisors.com>
>>
>>
> _______________________________________________
> dba-SQLServer mailing list
> dba-SQLServer at databaseadvisors.com
> http://databaseadvisors.com/mailman/listinfo/dba-sqlserver
> http://www.databaseadvisors.com
>
>




More information about the dba-SQLServer mailing list