Category Archives: Windows Update

Recent Rising Reclaimable Counts

Recently, I’ve been noticing that Cumulative Updates (CUs) typically leave upwards of 10 or more reclaimable packages around, following installation. If my memory is any guide, numbers from 2-4 had been more typical until earlier this year. But recent rising reclaimable counts seem all too likely these days.

So I asked Copilot about it: “Why does dism /analyzecompo-nentstore find more reclaimable packages after CU updates recently?” Imagine my surprise when Copilot cited one of my very own blog posts in response as a potential source of “deeper insights.” Wow: that’s a shock to me!

Here’s what DISM told me after I updated Windows 11 24H2 to 26100.4202 last week: 10 packages reclaimable!

Why Rising Recent Reclaimable Counts?

When I check the component store using DISM /Get-Packages after recent CUs I see numerous staged and superseded items in the listing. These are what often gets cleaned up when a following DISM /StartComponentCleanup is run. Google AI says that  recent changes to “checkpoint cumulative updates” in 24H2 means that “Future updates then build incrementally from these checkpoints, leading to smaller downloads and potentially faster updates.” Looks like it means there are more and smaller items of this kind, rather than fewer and larger ones, as under the previous regime.

Thus, I think the phenomenon is real, and reflects a change in update approach and architecture. DISM dispatches these items pretty quickly (at least anecdotally more quickly than the older approach, in my own personal experience). So all in all, I’m inclined to find this a good thing.

It’s nice when incremental changes in Windows mean improvements in the general state of Windows-World itself. Today, that’s my story — and I’m sticking to it!

Note Added 6/7: A New Top Count!

I ran DISM /AnalyzeComponentStore on the X380 Yoga just now (running 24H2 Beta Channel Build 26120.4230). It reported a whopping 21 reclaimable packages, all of which vanish after running DISM /StartComponentCleanup. See this:

I’m inclined to take this as further validation of my thesis that CUs bring more (and smaller) packages for cleanup in their wake. What do you think?

Facebooklinkedin
Facebooklinkedin

Unsticking Lenovo System Update

From May 6 through 12, I had a Lenovo System update stick in WU. That is, it would attempt to install, fail, and then push a retry button at me. Alas, that meant WU wouldn’t show me any newer updates, either. At the end of this cycle KB5058496 came along. It didn’t show up in WU, either. That’s when I found myself unsticking Lenovo system update on the Lenovo ThinkPad Yoga X380 where it happened. How did I do that? I ran the:

Reset_Reregister_Windows_Update_Components_for_Windows11.bat

batch file from the Eleven Forums tutorial Reset Windows Update in Windows 11. As it so often does, the Lenovo System update worked the next time I tried after said reset operation had completed and I’d rebooted that PC (as per the batch file’s own instructions). The new CU installed, and went to Build 26120.3964.

More on Unsticking Lenovo System Update

I’m not sure why the WU version got stuck, nor why it stayed that way for some time. When I looked in the Windows-Update.log file I produced via the PowerShell Get-WindowsUpdateLog cmdlet, no supporting detail told me why it happened, either.

All that Copilot could tell me was that it must be a Lenovo servicing driver update of some kind. Google was willing to speculate it might be the driver for the Lenovo Intelligent Thermal Solution. Lenovo Vantage kind of confirms this in a back-handed way, in that its history shows the latest version dated March 2024 with version number 2.1.14.0, which certainly seems to follow in the general numbering track for the item that got stuck.

So I checked Device Manager > System devices > Lenovo Intelligent Thermal Solution properties. Sure enough, the currently installed version is 2.1.52.0. Interestingly the install date shows as 4/11/2025 (same as in WU update history). That leaves me glad this already-installed driver somehow got itself unstuck. I’m still wondering why WU offered it repeatedly from 5/6-12.

These meaningless mysteries never stop in Windows-World. I’m just glad this apparently unnecessary driver offer stopped when I reset WU. Now the machine is running Build 26120.3964 and the right Intelligent Thermal Solutions driver without further issues. I’m good for now, but sure something similar will pop up soon, on one or more of my mini-fleet of 12 PCs. Stay tuned!

Facebooklinkedin
Facebooklinkedin

KB5058379 Forces 8GadgetPack Manuevers

On Patch Tuesday (May 13) I installed the latest updates on my PC mini-fleet (12 units right now). I didn’t reboot the Windows 10 production PC until yesterday. As soon as it came up, so did 8GadgetPack. Alas, the latter promptly crashed — and kept crashing — as I tried to bring it up. Next I realized that KB5058379 forces 8GadgetPack maneuvers, as  I tried uninstalling, then reinstalling that application.

That approach worked nicely to bring 8GadgetPack back up. Next, I performed a couple of follow-up gadget shenanigans along the same lines. That is, I had to remove individual gadgets that didn’t render on screen properly, then replace them from the Add Gadget repository. (Screen 1 from that tool serves as the lead-in graphic.) The whole exercise took about 10 minutes. And now, it’s all good.

Apparently, that update caused more havoc on other PCs. Keep reading for some of those ugly details…

KB5058379 Forces 8GadgetPack Manuevers … and More

This morning, I’m reading online that KB5058379 apparently causes other, more serious problems, too. The title of this WindowsLatest item (dated May 15) captures much of what’s amiss: Windows 10 KB5058379 locks PCs, BitLocker Recovery triggered on boot, BSODs. I guess you can say that while I got irked by having to mess about with 8GadgetPack, I’m now relieved that I didn’t have to go into full-blown troubleshooting and recovery mode yesterday instead.

In one way or another, all of these issues appear tied in some way to BitLocker issues. In the afore-linked Windows Latest story, Mayank Pamar explains a demonstrated repair strategy on some PCs. See that story for the details. The TLDR; version is “Turn off Intel TXT  in the PC’s BIOS/UEFI (may show up as “Trusted Execution” or “OS Kernel DMA Support”). This turns off BitLocker, but lets the update finish. Then you can turn it back on again.

Uninstall/reinstall or turn off/turn on seems to be the underlying theme for today’s blog post. That’s why undoing and redoing stuff remains a tried-and-true troubleshooting technique here in Windows-World, I guess!

Note added 5/16: After the next Release Preview CU, the same thing happened again with 8GadgetPack. And once again, uninstall-reinstall returned it to working order. Sigh, and sigh again. It’s looking like this may become part of my post CU recovery process. I can’t say I find that a delightful prospect, but it is a tolerable one.

 

Facebooklinkedin
Facebooklinkedin

Correlating KB Items and DISM Package IDs

Here’s an interesting situation. I was reading on Neowin this morning that MS has fixed a Windows 10 issue that caused BSODs on some systems (not mine, thankfully). To find or uninstall such an item, one must use DISM. But DISM deals in “Package Identity” strings, not in KB article numbers (e.g. KB5057589, as in this case). Surprisingly, correlating KB items and DISM package IDs turns out to be vexing and tricky. Indeed, this SuperUser thread more or less confirms what I quickly figured out. That is: the only datum both items have in common (using Update History for KB items and DISM Get-Packages for PkgIDs) is their install date/time.

Fortunately, what I was seeking showed up dead last in the Get-Packages output in PowerShell/Windows Terminal. As you can see in the lead-in graphic, it’s the only item whose install date matches that for KB5057589. But there’s no inherent correspondence with its PkgID: Package_for_WinREServicing~31bf3856ad364e35
~amd64~~19041.5728.1.1. What to do?

More On Correlating KB Items and DISM Package IDs

I figured there might be a PowerShell script (or something similar) already available to establish this correlation. AFAIK, nope! I thought that Copilot might be able to write me such a script. Nope again: it wants to look for the KB item ID inside the PkgID. You can see from the foregoing item (or by looking at installed updates using DISM Get-Packages) that this just ain’t so.

It looks like the only way to put all this together is to install the PSWindowsUpdate module, then use its built-in Get-WuHistory cmdlet. By writing that to a file, and then doing likewise with output for DISM Get-Packages, it should be possible to use matching date strings for KBs from the former with the “Install Time” attribute value from the latter to find and document matches.

Another Project for My List

Now that I know what I must do, I need to figure out how to do it. That will make excellent fodder for another blog post. As soon as I find the proverbial “round tuit” I’ll put that together and post it here. In the meantime, it’s nice to see that the obvious path to success (looking for the KB item ID inside the DISM PkgID) isn’t the actual path to success. Here in Windows-World, that’s all too often the case. I’m glad it keeps me entertained. I hope you feel likewise.

Facebooklinkedin
Facebooklinkedin

Reinstall Now Builds Current Images

Last Wednesday, I blogged that a repair install for Windows 11 unsticks WU. As I think about what that really means, I want to emphasize that using Settings > System > Recovery > Reinstall now does something remarkable. That is, Reinstall Now builds current images for whatever version Windows Update is serving at the time. It used to be only UUPDump.net could do that, by slipstreaming all the latest updates into the base Windows image (24H2 in this case).

How To See That Reinstall Now Builds Current Images

If you look at the Settings > System > About info that appears in the lead-in graphic, it tells pretty much the whole story needed for evidence. You can see it shows version 24H2, Build 26100.3775 with an install data of 4/9/2025. That’s the very day I ran the repair install, and the build number matches what follows in the wake of the latest CU (KB 5055523 — see the parenthetical phrase at the end of that title).

What makes this facility remarkable is that UUPDump.net has to build a Windows image for the baseline release, then apply as many updates — the latest security, cumulative and servicing stack items — as it needs to bring the image current. This requires some time-consuming DISM manipulations that can take an hour to complete. Interestingly, the WU facility handled the entire repair in about 35 minutes.

I still recommend UUPDump.net as a way to create an ISO for some specific (and non-current) Windows Build. But if you need to repair a current version, it looks like built-in Windows 11 recovery really is your best choice. Good to know! That’s why I’m telling you…

Facebooklinkedin
Facebooklinkedin

Repair Install Unsticks WU

For the past 5 weeks or so, I’ve been working with the Lenovo ThinkPad T14s Gen5 laptop. For the last two weeks, updates have been stuck, with an error code that indicates file download issues. The usual repair techniques haven’t helped, either — namely run the troubleshooter or the reset & re-register Windows Update components. So this morning, with a new cumulative update out, I installed the latest Windows 11 24H2 repair version. That built-in repair install unsticks WU and catches me up with pending stuff, as you can see in the lead-in graphic.

Repair Install Unsticks WU Trades Time vs. Convenience

The problems with the afore-mentioned techiques (troubleshooter, reset&re-register) is that they take multiple steps and a bit of effort. Double that when, as often happens, remediation is also needed. It took a while to click Start > System > Recovery > Reinstall now and then work through that process. But the details took care of themselves and I didn’t have to do anything except fire it off to make it work.

In the end, this turned out to be easier and less vexing than the other techniques. Its results were also immediately apparent, and entirely positive, once completed — as you can see in the lead-in graphic. That said, Update History does become a little opaque when you conduct this repair. Here’s what it says now:

It doesn’t show the problem CU installed and running. It simply shows that “Windows 11, 24H2 (repair version)” got installed today. Of course, that means the installer used the latest version of the Windows image — including those problem CUs — as the install base. So really, it’s all fixed now. You just have to know what this reference means.

And ain’t that just the way things go here in Windows-World? The problem may be solved, but a hint of mystery — or is it confusion? — remains. Cheers!

Note Added 4 Hrs Later: Get-Hotfix Tells the Story

Reading through ElevenForum.com threads just now, I learned that running Get-Hotfix in PowerShell will shows installed KBs from a repair install image, to wit: This shows that various updates and security updates are indeed present in the newly repaired image. The current build number for that PC — 26100.3775 — also shows that KB5055523 has been applied. Good stuff…

Facebooklinkedin
Facebooklinkedin

KB5053643 Kills Mouse, Keyboard

Last Thursday, I downloaded and installed a new Preview CU for Windows 10 — namely KB55053634. After lunch Friday, I finally got around to rebooting to complete that process. Eventually, it succeeded. But first, just to make things incredibly exciting KB5053634 kills mouse, keyboard — my vital USB peripherals — dead. Here’s the story of what I had to do to bring those devices back to life, and actually log in to Windows 10.

KB5053643 Kills Mouse, Keyboard: Now What?

Because I couldn’t get past the lock screen without a valid input device, that turned out to be a little more vexing than one might guess. So first, of course, I rebooted again. Still no joy: the keyboard didn’t respond to keypresses (a good test on my Comfort Curve 4000 is to toggle the Function Lock or Scroll Lock keys because those also toggle handy little green indicator LEDs). Nor did a mouse click open the PIN input box as usual.

So I tried again. Still no dice. Then I thought: “maybe the device needs a cold, hard boot?” That means powering off the PSU, waiting 1-2 minutes, powering back up, and pressing the power button. And indeed, that did the trick. Once I went through that maneuver, the hardware got completely reset, reinitialized and enumerated. It was enough to restore my key USB peripherals to working order.

What (Would Have Been) Next?

If the cold boot or hard boot didn’t work, I’d have had to jump into UEFI, turn off Secure Boot, and then target bootable repair media to get something running on that PC. I’ve done it many times before and will no doubt do it again. But hey: I was glad not to have to do it this time. Cold, hard boot did the trick.

Makes me feel like I dodged a bullet. Remember to give that a try if you find yourself bereft of mouse and/or keyboard after installing a Windows update or upgrade. If you’re lucky like I was, that will bring the USB drivers back into play, and let your PC get back to work. Cheers!

Facebooklinkedin
Facebooklinkedin

Ghost in the Machine Needs Printout

I have to chuckle when I read about these kinds of things. For much of the week I’ve been reading online (see list of articles near the end) about printers waking up and printing stuff on their own. On  Windows 11 PCs running Build 2263.4825, it seems that those with specific printers may start printing garbage output spontaneously. ICYMI, “ghost in the machine” is a British philosopher’s shorthand phrase for Descartes mind-body dualism. In this case, I’m twisting that metaphor further to impute independent action to a Windows print driver gone wrong. That’s why I aver that the Ghost in the Machine needs printout.

Why the Ghost in the Machine Needs Printout…

Newer printers (mid-2010s and afterward) that support driverless printing technologies such as Mopria  (a printer maker alliance that includes Canon, HP, Samsun and Xerox) and AirPrint (an Apple technology widely used by printer makers, too) also support dual-mode printers. For the ghost to start printing on its own, such devices much support both USB print and IPP over USB protocols (IPP is the Internet Printing Protocol). After updating Windows via KB5050092 (release 1/29/2025) such printers may start spewing pages, no user print requests nor print spooler files needed.

You can read about this specral phenomenon from a plethora of sources including:

BleepingComputer Recent Windows updates make USB printers print random text (March 12)

Windows Forum Windows 11 Printing Glitch (March 13)

PC Gamer Haunted printers turning on by themselves and printing nonsense (March 12)

I’m not the only industry follower who’s picked up on the “ghost in the machine” metaphor, apparently. And you thought Windows was a brute and soulless beast, I’ll bet…NOT! Anybody who works with the OS for any length of time knows full well it’s possessed of a host of spirits that range all the way from the most angelic to the deepest of deviltry. I’ll let you decide how magnificent or malefic this particular haunt might be for yourself.

One more thing: the uber-cutesy graphic that starts off this blog post is Copilot’s response to a prompt that reads “show a PC printer possessed by a ghost.” Another clear case of you get what you pay for, IMO.

 

 

Facebooklinkedin
Facebooklinkedin

Installing Build 27802 Throws Memory Error

Here’s a new one on me. Last Friday, as I was installing the latest Canary Channel upgrade, the installer threw an error code that I’d not seen before. That code is 0x8007000e; its output from the Microsoft Error Lookup Tool (err_6.4.5.exe) appears as the lead-in graphic above. That error occurred during the GUI portion of the install. And it occurs to me that while installing Build 27802 throws memory error, it might have been because I was running WinGet in parallel, installing other stuff at the same time. I’m guessing was a self-inflicted thing…let me explain.

Self-Inflicted: Installing Build 27802 Throws Memory Error

The recommendation that comes with this error, is to restart the PC and try again. As soon as I did that — without added activity on the side — the upgrade installed successfully with no further errors along the way. As I look back on what got updated during my first botched attempt, I see that some fairly intense items were involved. Most notably, it included Visual Studio, for which a typical install is usually around 50GB in size. I can see where trying to juggle both on a 2021 vintage laptop (Lenovo ThinkPad X12 Detachable Tablet with 16GB RAM) might cause resource issues.

Anyway, the proof’s in the observation that a second attempt worked. That’s probably because I didn’t try to multi-task while the GUI install was underway. The only reason I haven’t done this to myself before is that you can’t do anything to the PC except let the installer run, during the post-GUI phase!

27802 Takes a While to Complete, Too

I couldn’t help but notice — because I perforce went through the process twice on the X12 — that the upgrade process to this latest build takes some time to complete. Normally, a Canary Channel upgrade finishes in under half an hour. This time around, the whole process — including download, GUI install, and post-GUI install — took about 75 minutes to complete from desktop to desktop.

At least I now know I should leave my PCs (mostly) alone while the GUI phase of a Windows upgrade is underway. I wonder what my next creative abuse of the runtime and installer will teach me? There’s always something new and interesting to learn, here in Windows -World!

 

Facebooklinkedin
Facebooklinkedin

WU: How Long Is Too Long?

Here’s a Windows road I’ve been down many times. Indeed, it’s the kind of road, as in Robert Earl Keen’s excellent song, that “goes on forever…” It’s the road you walk on when WU hangs during download, GUI install, or post-GUI install. I read with amazement this morning in an ElevenForum thread that some poor soul waited THREE HOURS on a stuck install before asking for help. Ouch! Of course this raises the question with WU: How long is too long when things get stuck?

For me, the TLDR; answer is “10-15 minutes.” I just don’t have the patience to wait much longer. And FWIW, I’ve only seldom seen something that’s been stuck that long succeed after such a delay.

In WU, How Long Is Too Long Depends on You

At some point, the stuckee realizes that nothing is going to change, no matter how much longer one waits. That’s the point at which one must bite the bullet, and restart the stuck PC. Holding down the power button for 10 or more seconds until the PC shuts down will usually do it. Sometimes, however, one must either power off the PSU (desktops) or take more drastic steps (e.g. disconnect battery or wait for it to drain completely on a laptop).

Surprisingly, in the dozens of times I’ve had to do this when stuck in the past 5 years or so, the aftermath has mostly been positive. Often, Windows will simply pick up where the stuck update left off and finish up from there. Sometimes, it will roll back to the pre-install state instead.

Only in a handful of cases has the affected PC refused to boot correctly. When that happens, it’s time to pull out your rescue media and perform an image restore to your last known,good, working image backup. You have one of those, right? I’ve learned the answer to that question had better be “Heck, yeah. Let’ s go!”

Overcoming The Worst Case Scenario

No image backup and no working PC can be problematic. Hopefully, you’ve got at least some important stuff backed up someway, somehow (OneDrive, maybe?). You’ll either find a way to run a repair install (works sometimes) or you’ll have to choose between a clean install or a factory reset. Hopefully, it won’t come to that. I haven’t had to go there but once or twice in the 30-plus years I’ve been running Windows. Hopefully, your odds and experience will be the same. Good luck!

Facebooklinkedin
Facebooklinkedin