It is obvious if you have scripts that are like IF EXISTS: DROP: CREATE PROCEDURE, and run this on a production system that is calling those procedures at the same time, that some will fail when the hit that period between the DROP and the CREATE.
I was using sp_rename lately and came across this article that goes into the pitfalls.
Call me crazy but triggers are usually bad news. In my experience they tend to rise to the top of the queue for bug fixing more frequently than their relatively small SLOC would belie, and often have performance problems that don't become apparent until months or years down the track as record counts increase.
I recently wanted to know how given a table of user names (logins) and passwords how you could verify that list can connect to the local server from within a stored procedure. This is what I came up with, first the setup.
TL;DR version: If you encounter problems with GoToWebinar calendar entries on a Mac remove the ORGANIZER line from the file and try again. The full explanation follows.
I recently saw an issue where a large number of databases were consolidated onto a single server. Though they were backed up on a proper full -> differential -> log cycle, and the starting day for each cycle was roughly spread evenly over the week, the large variance of characteristics between each database resulted in backups running far too long on some days and terminating too early on others.
Updated 29 Sep 2016: Fixed a bug in IsNull where Bruno Martins pointed out it returned True for 0. This said… this post is super old… and should be ignored.
Stop me if this sounds familiar.
This is the accumulated hands-on knowledge of days of work and hours of technical Kerberos videos that I didn't (and still do not) understand.
I exported my bookmarks in HTML format from FireFox and wanted to do some processing on them, but parsers like the HTML Agility Pack could not handle it properly because the format does not include DT (Url blocks) or DD (Description blocks) end tags, and that in turn results in massively nested XPath lists.
Continuing on from earlier, after I discovered that FireFox includes far more bookmark information in its native format (JSON) I wanted to create a parser for that instead. Normally PowerShell provides you two simple ways of doing this: ConvertFrom-Json and ConvertTo-Json.
Today I needed to modify a function to remove some bad files that were causing an import to fail across a long list of servers (using Invoke-Command). Here's a sample I've constructed to show and test the removal part.
I've tried a few different options for FTP with PowerShell and the best one so far is WinSCP with its .NET assembly.
This is something to be very, very wary of when storing data blobs in SQL Server. If you compare varbinary fields they will match when one field has one or more extra zero bytes at the end, even if they are different lengths. Observe:
This is easy.
PowerShell error handling is great but becomes more complicated as your functions become more complicated and interdependent. For a while I used -ErrorAction Stop on everything because I was available to immediately deal with unforseen problems, but when functions call functions sometimes you don't really want them to just stop, you want to continue or possibly parse the output later.
I use one-way transactional SQL Server replication to ship data out from a source server to multiple destinations, and recently added a publication with a vertically filtered article; that is to say a table where I exclude a column from replication because I do not want it sent to the subscribers.
I had an error today where Business Intelligence Design Studio was running an SQL Server Integration Services package (I'm spelling it all out so search can find it) and failing with an error.
I encountered an error where an SQL CLR that had been in use for a long time was triggering this error from SQL Server: