This week I am looking for some examples of error messages that you have seen from Microsoft products. SQL error messages would be good, but I am not being picky. The idea is for me to see if I can perform different types of root cause analysis on the error messages so I will also need some general context on how you came across the error.
In my own experience I have found that many times the error message has nothing to do with the actual issue. I like to call them “ghosts”, as in “I feel like I am chasing ghosts.” I have lost track of the number of times (and hours spent) focusing on the context of the error message only to find that the message was an end result and nowhere close to the root cause.
If you have any to share with me please provide the info in the comments or drop me an email sqlrockstar@gmail.com.
You’ve hit an interesting topic. In a complex system, you’re never really sure which component delivers an error message – it could be a discrete component (such as the SQL Server Engine Locking mechanism) or the system just above it (the Query Processor) or even an interaction between the two. For that matter, most errors you never see – the SQL Server base code intercepts them and deals with it. After all, it’s not that we want you just to know what went wrong – if we know that, we can often just fix it for you. It’s that complex interaction that makes the messages difficult to interpret.
thanks for the info Buck. I know the system is quite complex and it would be difficult to have perfect errors messages for every possible scenario. the point of my question is to gather some information on error messages so that I can make an effort to perform some RCA testing. i would love to take some examples from around here but everything seems to be working perfectly as of late…
Well, I’m assuming you’ve already opened sysmessages….
And of course you can always peruse various forums. A lot of users post the full error messages they are getting in certain circumstances there.
Recently I’ve come to deal with (again) the dreaded “Cannot generate SSPI context” message from client. Nothing changed and the SQL instance (2000 SP4+HF) was running under a domain account that is made for running as a service. But after a recent restart of service I got the error message. Only fix I found was to change the service to run as Local System. I’m still investigating as to why that account had issues. What made it stranger is that the account was a local admin on that box (I know, I know not best practice but this one was legacy and getting upgraded soon anyhow).