Bug #40448 MYSQL 5 bluescreens on any command.
Submitted: 30 Oct 2008 22:33 Modified: 18 Nov 2008 17:03
Reporter: Colin S Email Updates:
Status: Can't repeat Impact on me:
None 
Category:MySQL Server: Installing Severity:S2 (Serious)
Version:5.0.67 OS:Windows (XP)
Assigned to: CPU Architecture:Any
Tags: MYSQL Blue Screen

[30 Oct 2008 22:33] Colin S
Description:
I am a new user of MYSQL and also of these forums. Im unsure how to report this issue or get assistance. However MSQL is bluescreening my PC on entering the very first command after installing it. 

While I can accept that a bluescreen crash is a MS OS issue, the cause is clearly MYSQL and I would like to think it would at least terminate with some kind of explanatory message rather than simply crashing the PC ! So as this is not an error message or issue I can investigate myself, I find myself looking for help here.

Here is the synopsis of the problem :

I have no previous versions of MYSQL (or any database) on my PC. In fact Ive never used MYSQL before.

But a bluescreen is a tough one for a first problem ...

Im using a Toshiba M500 portege (Intel 2 CPU T5600@1.83GHz, 1GB RAM) running Windows XP Professional Version 2002 SP2.

I have installed MYSQL 5.0.67 from the msql-essential-5.067.msi (and several other versions - see later) download using the default install for general use etc. I have added a password for user root.

I can run MYSQL as either a service or via the commandline (mysqld-nt.exe). mysqld.exe --console reports InnoDB Started, port 3306 version 5.0.67 as expected.

The post installation tests run as predicted on http://dev.mysql.com/doc/refman/5.1/en/windows-post-installation.html excepting for :

C:\mysql\bin\mysql -uroot -psecretpass -e "SELECT Host,Db,User FROM db" mysql
which runs OK but returns nothing.

The tests on http://dev.mysql.com/doc/refman/5.1/en/windows-testing.html all run as expected.

I then add a new user as follows :
mysql -h localhost -u root
> create user 'user1'@'localhost' identified by 'user1';
> grant select, insert, update, delete, create, drop, references, execute on *.* to 'user1'@'localhost;
> exit

And that's about as much as I can do. I then reaccess MSQL as follows :
mysql -h localhost -u 'user1' -p

I then type two commands as follows. On the second command the PC bluescreens (something it rarely does) consistently (ie on 100% of attempts) :
> DROP DATABASE IF EXISTS books;
> CREATE DATABASE books; (PC bluescreens at this point)

Clearly I have done something fundamentally wrong since I can even get a simple command into MYSQL.

I cant find any obvious posts or outstanding bugs on this issue, which makes me more suspicious Ive done something fundamental.

Please can someone kindly point me in the right direction ? Or should I be logging this as a bug ?

PS
Additional info :
In trying to resolve this issue I have tried version 5.1 and 6, which exhibit identical bahaviour. I have uninstalled each via windows before trying a different version. I do note some of the files are still present since it remembers my root password and user id so I mention this for completeness.

How to repeat:
Unfortunately I accept that this may be environment - specific to my PC.

However I do believe the product should be enhanced to handle gracefully whatever is causing it to fail.

Or, failing that, could I at least get some pointers to what I may have done during the install that could cause this - or what other settings or products may cause this ? 

Certainly there are no other instances or programs that have ever bluescreened my PC in the 12 months since I purchased it.
[30 Oct 2008 23:02] Colin S
After the investigations over the last couple of days I have discovered that if I install MYSQL as part of Ruby on Rails (that is if I let Ruby on Rails do the MYSQL and Apache installs etc for me) then I am able to go into the MYSQL command prompt and type CREATE DATABASE test; and the command (appears to) succeed. Certainly there is no bluescreen.

Ruby on Rails is not what I want on my server, I just want MYSQL, but I thought this information might be of some use.

For this test I used InstantRails 2.0 which appears to intsall MYSQL 2.0.27-community.
[31 Oct 2008 4:19] Valeriy Kravchuk
Thank you for a problem report. Please, upload your my.ini file (it should be in the MySQL install directory) and entire error log (usually it is a <hostname>.err file in the data subdirectory).

Is there anything related to MySQL in the Event Viewer of Windows for the moments before and after blue screens?
[31 Oct 2008 6:40] Colin S
Thanksyou very much for responding to my report.

Files are attached as requested. In the install directory I saw only .bak files, presumably because I uninstalled the bluescreening version of MYSQL. Since I didnt make any changes to MYSQL save the commands I listed, the BAK file should be accurate. Ruby, as you probably know, installs MYSQL in a separate directory which in my case is on a different partition. 

I have also attached the .err file.

Before the crashes there is nothing obvious in the system log. For example, immediately prior to the last crash I see an information message logged by Service Control Manager (event 7036) saying it has entered the running state. 31 seconds later I see the kernel crashevent id 1003 reporting :

Error code 1000007f, parameter1 00000008, parameter2 f786fd70, parameters 3&4 0.

There are several similar information messages prior to the crash, and also an error event from W3SVC id 115 noting "the service could not bind instance 1".

I dont know if any of these are relevant.

I have today run several commands through the Ruby installed MYSQL without issue.

Is there any more information I can give ?
[1 Nov 2008 20:44] Sveta Smirnova
Thank you for the feedback.

Please check windows logs.
[1 Nov 2008 22:16] Colin S
WHich windows logs do you mean please ? Could you be more specific. 

I have referred to the Windows event log in my post, is there another one you would like me to examine ?
[1 Nov 2008 22:37] Sveta Smirnova
Thank you for the feedback.

I meant if you have more information in pop-up with details.

But "Error code 1000007f, parameter1 00000008, parameter2 f786fd70, parameters 3&4 0." says you can experience hardware problems (0x0000007F: UNEXPECTED_KERNEL_MODE_TRAP). Please check your hardware and RAM in first turn.
[1 Nov 2008 22:39] Sveta Smirnova
Error "the service could not bind instance 1" means port is busy.
[2 Nov 2008 21:14] Colin S
Tanks for the info.

In fact when I ran up MYSQL as installed by Rubyonrails it initially did inform me the port was busy, which I was able to easily rectify.

Would I expect the stand-alone version of MYSQL for which I have reported the bug to blue screen if the port is busy ? 

It seems a rather harsh response for a busy port !!

Or perhaps you are not suggesting this is the cause of the problem ?

Re the comment on checking my hardware - what checks should I perform ? Is there something hardware-related that would cause the stand-alone version to bluescreen and the Ruby version of MYSQL to behave perfectly normally ?
[3 Nov 2008 7:57] Sveta Smirnova
Thank you for the feedback.

> Would I expect the stand-alone version of MYSQL for which I have reported the bug to blue
screen if the port is busy ? 

Of course not. This error either is not directly related to the blue screen, or shows  "the service could not bind instance 1" for some unexpected reason than expected problem with busy port.

> Re the comment on checking my hardware - what checks should I perform ? Is there something
hardware-related that would cause the stand-alone version to bluescreen and the Ruby
version of MYSQL to behave perfectly normally ?

Yes. More likely RAM, but worth checking hard drive as well.
[14 Nov 2008 23:14] Colin S
Being totally new to MYSQL, I didnt want to speak out of place. However having spent several days researching, I have learned a lot. 

Ive seen a high number of blue screen crash problems blamed on the end user, not the product.

Ive seen other problems where people have written extensive dialogues backing their anaylsis of a problem, and their frustration that the problem isnt being acknowledged. 

But this isnt enough for me to have an opinion - Im no expert. And they are just dialoges on the internet.

However it did make me confident in my own assessment that, although technically possible, a bluescreen crash caused by a single piece of software and no other software (out of around 100 installed programs) on a heavily used PC is not the top of the list of possible causes - given its not accessing any new or seldom used hardware .... just the same disk and memory as everything else. 

But it did allow me to ignore the suggestion of a hardware problem -although I did run a disk check and do memory swap out. But really - its not top of the list of possible causes is it ? Come on !

How about something a miss with the install process ? That might explain why Ruby on Rails can install MYSQL just fine on the exact same PC, because the install is automated.

No ? Well a bit of a detailed check around with some of the /bin utilities turned up some errors about missing files. A complete removal and re-install didnt work. But a removal and re-install with all manual configuration options did (and a lot of experimenting). 

No blue screen crash any more.

But I guess putting the onus onto the end user and their hardware, while quite justified in some cases, really didnt make sense and was a red herring from the start. There Ive said what I wanted to all along !

So I can see where we're going here . . . the re-install has somehow bypassed the faulty hardware on my PC. The fault's still there waitng to bluescreen. Its just that its not the installer or MYSQL thats causing the problem - in fact the installer is my friend because it allowed be the flexibility to configure MYSQL so it doesnt upset the faulty hardware on my PC. 

And I suspect at that we will reach an impass. I say (well I know) its the installer that can under some conditions allow an invalid install that will bluescreen the PC 100% of the time. 

Others will say its my faulty hardware.  

So, lets call it stale mate and close the case. It may at least help others with the same issue.

My thanks (genuine) for resopnding to my bug report - I could only wish for perhaps something a little more likely as a first-pass diagnosis. Perhaps a little Kepner Tregoe training would go a long way here (Kepner Tregoe - what's that all about then ?)

regards

Colin
[18 Nov 2008 17:03] Sveta Smirnova
Thank you for the feedback.

I think we can close this report as "Can't repeat".