SQL Server 2019 CU7 didn’t go so well, and Microsoft pulled it. Old news. That’s so 30 days ago.
Microsoft pulled CU7 due to a bug involving snapshots, and they’ve fixed it in today’s Cumulative Update 8.
Or at least…I think maybe they have.
Because CU8’s release notes don’t actually mention anything about snapshots (other than a replication bug).
So I’m gonna be honest, dear reader: I have no idea if CU8 fixed CU7’s bug or not. See, Microsoft has stopped bothering to create KB articles explaining what a Cumulative Update is fixing. They want you to go on faith, and based on what’s happened so far with 2019’s CUs, I’m personally out of faith.
I don’t recommend that you install further updates to SQL Server 2019 until we can get a straight answer as to what the sam hell is inside these updates.
Knowing how the world works, somebody from Microsoft is going to read this, and say, “Whoa, we forgot to include the part in the KB article that says we fixed the CU 7 bug.” They shouldn’t be relying on some blogger somewhere to point out that they have blatantly obvious problems in the update notes.
Microsoft, c’mon. This is getting ridiculous.
Update Later in the Day
After Microsoft folks saw this post, they added this to the top of the KB article:
Still no link to a KB article on what the bug was, why it’s so important for people to uninstall CU7, or what steps they should take to avoid problems while they’re waiting for the business outage window to apply CU8. Just flat out disappointing.
Hi Brent, if it had not been fixed in CU8, there would be no CU8. And we did state the fix would be in CU8, in the CU7 blog post. The release CU8 blog post has that note too: https://techcommunity.microsoft.com/t5/sql-server/cumulative-update-8-for-sql-server-2019-rtm/ba-p/1736647
For your readers clarity, Microsoft hasn’t stopped creating KBs. Just not child KBs for *everything*, replacing short worded KBs with the same information that would be there but just inline in the main CU KB, as announced and explained here: https://techcommunity.microsoft.com/t5/sql-server/changes-to-kb-articles-used-for-cumulative-updates/ba-p/1597145
Pedro – so in the KB article for CU8, which one fixes the issues caused by CHECKDB snapshots?
Ok, so no KB, but its not even listed with a bug reference number. Many of those listed have no KB. But why is this fix no associated with anything referenceable?
Can’t see it mentioned / fixed in CU8. Can’t even work out which “fix” in CU7 introduced the issue.
This is getting as bad as the PBI Report Server releases. The last 3 major releases have had 2 or 3 release each. And we installed one version that wasn’t even listed in the the release notes.
The KB article has been updated…
Yellow box near the top
Curtis – thanks for the heads up. Looks like my prediction was right: Microsoft’s relying on the community to proof their CU release notes. What a disappointing mess.
But no bug reference….I guess it wasn’t a bug?
Nope. Not worth documenting. Important enough that they had to ask everyone to UNINSTALL THE FREAKIN CU, but not important enough to write a single page about the damage or risk. This is seriously bullshit.
Tons of memory dumps / access violations and production outages since upgrading to 2019… had to revert everything back to 2017. Have added a lot of new, strange, unpredictably odd test cases to our list to catch future issues. Logical page corruption in query store was one of my favorite things to fix over the summer. We test CU’s for a minimum of 30 days before promoting to a user acceptance testing environment… and we keep finding things in production to add to the list… unlike every before… some random weird one off scenario will trigger a massive SQL 2019 fail. We are running out of beer money to donate to our phone support team as well as clever things to tell our stakeholders when we are down. We are holding where we are at until we can trust again. It’s sad that I’m relieved when there is an Azure outage (Monday 9/28) that spans for hours instead some kind of db issue. Do these CU’s come with stress management courses?
Same. Tons of memory dumps and access violations since upgrading to 2019. On CU6. 2017 was fine. Not to the point of reverting yet but MS isn’t answering any of the uservoice reports about the access violations (there’s many, new and old) so I’m running out of hope. Historically SQL Server was the most stable piece of software I ever experienced until these recent versions 🙁
Apologies on reviving an old thread but I just came across this today. We recently upgraded from 2017 Ent to 2019 Ent and are seeing a myriad of issues that seem to follow the same pattern as what you mentioned. Memory trouble at off-peak hours, failovers, login fails we have been attributing to resource issues, etc. Did you also upgrade to Enterprise? Mainly asking because we have not seen the same on 2019 Standard as of yet. Also curious if you discovered something secret that resolved some of these issues?
For personal support, your best bet is to contact Microsoft.
>>We test CU’s for a minimum of 30 days
we do the same
and the day MS revoked CU7 was exactly the day I was going to promote it to UAT…
luckily I have not
Here is the release announcement
[…] Case in point, just look at how buggy Cumulative Updates for SQL Server 2019 have been: Microsoft pushed out the 2019 CU7 only to soon remove it and push CU8 as a fix what CU7 had broken. […]