A Fine Slice Of SQL Server
Hugo Kornelis has a new video:
One of these two features is Lock After Qualification (LAQ). This feature avoids the scenario where a delete or update is blocked by a locked row that would not qualify. But, like any good thing, there is a price. This video shows the feature, explains how it works, and shows some of the potentially undesired side effects.
Click through for the video.
Leave a CommentReitse Eskens runs into a problem:
This warning felt benign; more like you can run into errors. Well, as I found out, this lets you run into an SQL Server installation that just fails.
The error logs fill up with stack dumps and a fatal error that makes no sense. The installation log, however, shows you a more meaningful error.
Click through for the error message, as well as one way to fix the problem.
Leave a CommentAK Gonzalez summaries some changes:
For years, database professionals faced a frustrating limitation. With Availability Groups, you could offload some workloads to secondary replicas—but not your real backup strategy.Yes, you could run copy-only backups on a secondary. But true full backups? Differential backups that maintain the LSN chain? Those had to run on the primary replica, until now. SQL Server 2025 introduces the ability to run true full backups and true differential backups on secondary replicas.
Read on to see what this means. AK also looks at the change to ZSTD for backup compression and immutable backups in Azure Storage.
One thing I would want to warn anybody (or any company) looking into using immutable storage for backups is, Microsoft’s not joking about that immutability. That includes deleting them when you see what the bill is going to be. You can set retention polices to delete these automatically, but that’s the only way you’re getting rid of those old backups. And just because they’re old doesn’t mean you get charged less for the privilege of storing them off-site.
Leave a CommentWe were setting up a brand-new SQL Server 2022 instance and couldn’t connect remotely using valid Windows credentials. Every attempt gave us the same error: “Login is from an untrusted domain and cannot be used with Windows authentication.” The client only has one domain. Permissions and firewalls checked out. Local connections with domain accounts worked fine. So why was the server rejecting us from every remote machine, no matter how we connected? The answer turned out to be a change in Windows Server 2025 that more and more DBAs are going to run into.
My advice for every DBA is as follows: make sure you have at least a rudimentary understanding of SPNs, including what they are, what can go wrong when they aren’t set correctly, what “set correctly” even means, and how to set them. You don’t need to be an expert on Kerberos, but I think you do need to be a technician who can note a specific error code and troubleshoot the issue from there.
If you ever had to deal with SSRS or SSIS double-hop issues, you’ve likely already dealt with SPNs in some fashion. Just bite the bullet and spend a few hours boning up on the topic.
Leave a CommentRob Sewell chains together some operations:
Last week I showed you how to work with workspaces — creating, updating, removing, assigning capacities. But we were doing each operation in isolation. Today I want to show you what happens when you connect those operations together using the PowerShell pipeline.
This is one of my favourite aspects of PowerShell and therefore it was imperative that Jess Pomfret B S L and I revamped the module to fully support pipeline operations. Every cmdlet that makes sense in a pipeline is built to work in one.
Click through for some examples of what this means.
Leave a CommentSean Gallardy troubleshoots an error:
The symptoms of this issue were interesting, every so often the instance would just kind of get “stuck” – at least that is how it was described to me. Nothing would work, cancelling queries, attempting to kill queries, submitting anything new, nothing seemed to really do anything except restarting the service. Once the service was restarted, the instance (and AG) would hum along nicely… until some random time later when submitting different queries would just grind to a halt. Fun.
The answer is just as dumb as you’d think. But I won’t spoil the punch line here.
Leave a Comment