zfs-windows-2.3.1rc9 #461
lundman
started this conversation in
Development
zfs-windows-2.3.1rc9
#461
Replies: 37 comments 48 replies
-
|
are these "ghost installs" or corrrect system devices? 
|
Beta Was this translation helpful? Give feedback.
0 replies
-
|
It seems there are remaining ghost installs in 2.3.1? maybe the reason or the result of my bsod after running the 2.3.1 installer pnputil output
|
Beta Was this translation helpful? Give feedback.
1 reply
-
|
Neat, does |
Beta Was this translation helpful? Give feedback.
-
|
tried to deinstall openzfs but failed 
|
Beta Was this translation helpful? Give feedback.
1 reply
-
|
Yeah, try the zfsinstaller way, it has new code we need to test to make sure it can uninstall. |
Beta Was this translation helpful? Give feedback.
-
|
when I try the rc1+ installer, I get |
Beta Was this translation helpful? Give feedback.
2 replies
-
|
Hmm, not stopping zed perhaps? |
Beta Was this translation helpful? Give feedback.
-
|
re-installed a few times without issues here |
Beta Was this translation helpful? Give feedback.
-
|
The sudden need to adjust ACL to rename drivers is also new in my setup. Win11 Pro 23H2 |
Beta Was this translation helpful? Give feedback.
0 replies
-
|
update From a list of services I deleted the OpenZFS ones on some drivers I was asked after the deleted message to reboot what I did problems: (driverdate see above thread) on oem139.inf I got a BSOD on oem164.inf I got a BSOD after this I could deinstall OpenZFS and run the 2.3.1+ installer without BSOD I tried a rerun of the installer. This ended in a BSOD (pagefault in nonpaged area) After reboot I tried again and could reinstall (tried twice) without further problem. |
Beta Was this translation helpful? Give feedback.
1 reply
-
|
I think I need to find a way to make ghost installs on my VM, so I can test the new cleanup code. Sounds like it would be better to figure out how not to unload, and just remove, then force a reboot. |
Beta Was this translation helpful? Give feedback.
-
|
OK, I had not yet cleaned up my VM snapshots, and some older had ghost installs, which enabled me to test the code properly. It was not right. Now it cleans up after uninstall: |
Beta Was this translation helpful? Give feedback.
0 replies
-
|
I have pushed a |
Beta Was this translation helpful? Give feedback.
3 replies
-
|
Ah hmm, we don't run Uninstall before installing a new version? Hmmm |
Beta Was this translation helpful? Give feedback.
-
That would be pretty brutal. :) |
Beta Was this translation helpful? Give feedback.
-
|
It appears Inno isn't expected to, if AppID doesn't change, it handles the updating automatically. But makes the new cleanup ghost code in uninstall not be run - might need to ask users who have run OpenZFS for a while to run the uninstall once. |
Beta Was this translation helpful? Give feedback.
-
The URL seems to be truncated (incomplete). Probably this: https://github.com/openzfsonwindows/openzfs/releases/tag/zfswin-2.3.1rc1 |
Beta Was this translation helpful? Give feedback.
3 replies
-
|
Found a BSOD early on, so want to fix that first. |
Beta Was this translation helpful? Give feedback.
-
|
What a stupid issue, basically, you can tell OpenZVOL - the Storport driver part, to Unload at any time, including when it has disks up and running. I am unsure what you are supposed to do in Storport to stop it from unloading while it is in use. |
Beta Was this translation helpful? Give feedback.
-
|
OK That was stupidly difficult, but rc2 is up to test. |
Beta Was this translation helpful? Give feedback.
-
|
Where can i download rc2? |
Beta Was this translation helpful? Give feedback.
1 reply
-
|
Got one more thing to fix, so I had to pull it |
Beta Was this translation helpful? Give feedback.
-
|
pool import problem in 2.3.1rc1 (id mismatch) |
Beta Was this translation helpful? Give feedback.
3 replies
-
|
Should have 1ac22ba in it? Still an issue then? |
Beta Was this translation helpful? Give feedback.
-
|
seems the case |
Beta Was this translation helpful? Give feedback.
-
|
Ah yeah, so created a pool and it mismatches, amusing. The cpuid remains the same, but something else is up |
Beta Was this translation helpful? Give feedback.
-
|
Updated from rc1 to rc2: bsod (irq not less..) pool import: id mismatch remains |
Beta Was this translation helpful? Give feedback.
3 replies
-
|
Yeah not looked at cpuid yet. |
Beta Was this translation helpful? Give feedback.
-
|
Oh hah ok so it's now |
Beta Was this translation helpful? Give feedback.
-
And maybe the description on the wiki page needs to be updated as well? |
Beta Was this translation helpful? Give feedback.
-
|
Copy problem still there 
|
Beta Was this translation helpful? Give feedback.
1 reply
-
|
I have no problems copying (copy/paste) and dragging files, including one named |
Beta Was this translation helpful? Give feedback.
-
|
A performance test with Crystaldiskmark results in a BSOD |
Beta Was this translation helpful? Give feedback.
1 reply
-
So did FileObject go NULL during the |
Beta Was this translation helpful? Give feedback.
-
|
Pushed out zfs-windows-2.3.1rc3 - minor changes, mostly hostid. |
Beta Was this translation helpful? Give feedback.
0 replies
-
|
During some work in Adobe apps, Windows suddenly crashed with a BSOD (openzfs.sys) without any zfs actions memory.dmp |
Beta Was this translation helpful? Give feedback.
0 replies
-
|
I downloaded filespy.exe and copied a file to ZFS |
Beta Was this translation helpful? Give feedback.
1 reply
-
|
Hmm yes. I wonder if there is a use of DELETE_ON_CLOSE in use and we fail to trigger it, or TEMPORARY usage. I see the ENCRYPTION requests, not that they play into it here. But we could look at a better reply eventually. The trick with FileSpy is to try to capture as small amount as possible, when the issue happens. But I think we can see here. It creates it, then complains it exists (so it expected it to be gone?) |
Beta Was this translation helpful? Give feedback.
-
|
@lundman short status update: rc4 didn't bsod on my even a single time, the previous troublesome operations that could by sometimes didn't are not a problem (import/export, mount/unmount, heavy load on async operations - for example using qbittorrent either bsod windows or cripples the queue so far that the software freezes and process will zombify- exiting but never closing handles) multi vdev pools spread across several devices/interfaces (ahci sta combined with pcie/nvme) don't have any issues with any load i throwed at them so far- earlier a heavy load would cripple and bsod the os, mount/unmount import/export work fine, if some process zombifies (see below) sometime, when trying unmount => export on a multi-vdev pool, it seems the process will zombify during termination - i will try to track down which op exactly causes this condition, sometimes I can force-close handles on a zombified (hanging on termination) process with sysinformer by using the kernel-driver and forcefully closing handles of the process manually (which isn't a healthy, windows can bsod, but in itself this never seem to damage/cripple/error any zfs pool/dataset (sometimes the process terminates without bsod, after manually closing handles, and finally and releases all resources). From my hillbilly observation (correct me if im wrong) it seems that either the driver waits for something or the process waits for the driver during termination - this then stales and produces a zombified process that is in termination state, but cant.. well.. terminate (i dont even know how to call this state, this is windows internals lingo). TLDR: end result is, dataset won't unmount and pool won't export, and zpool command returns busy state (earlier, zfs unmount returned an unknown error) im not sure if this occurs because because the driver waits for the process to close the handles or do "something" and fails to finish/propagate this to the driver, or the process waits for the driver to finish/propagate "something". I can however track it down more efficiently now and deliver more details, let me know if you want to me to prep details with dumps and exact location which op seem to be the cause or you have enough on tha platter for now. Finally, I was unable to BSOD windows with anything that I throw at it - great work, i will (with high probability) recommend zfs only backup for our internal systems at work (woohoo, one ring to rule them all!) |
Beta Was this translation helpful? Give feedback.
1 reply
-
|
@lundman correction, not a single bsod on 2.3.1 on win10 (19045.3757) or win11 (several last 27xxx to 27823.1000 canary) |
Beta Was this translation helpful? Give feedback.
-
|
Question about Is this mandatory now to allow programm execution from ZFS? when set to off, I get btw if I set to off at pool level vhd and to ntfs for vhd/data exxecution is also not possible for an app in vhd/data |
Beta Was this translation helpful? Give feedback.
2 replies
-
|
Sadly, both Windows and Apps on Windows will just check the name, and if it is "ntfs" they behave normally. Anything else, they break. Nothing we can do about that, the default could be NTFS, but that creates differences with imported pools. |
Beta Was this translation helpful? Give feedback.
-
|
This issue occurred so many times, that I wonder if it should get a dedicated issue :) |
Beta Was this translation helpful? Give feedback.
-
|
This means, for a "it just works" experience even with imported older or Linux pools, mimic should be always ntfs on Windows as default unless manually forced to off for whatever reason. |
Beta Was this translation helpful? Give feedback.
1 reply
-
|
Yeha, but that means we would change a setting just because you imported it on Windows, what if you set it a specific way |
Beta Was this translation helpful? Give feedback.
-
|
Do not change the onpool property, just treat it like this when imported on Windows |
Beta Was this translation helpful? Give feedback.
2 replies
-
|
Change it temporarily? |
Beta Was this translation helpful? Give feedback.
-
|
We could also make it "if unset, say NTFS. If set to ZFS, say ZFS." |
Beta Was this translation helpful? Give feedback.
-
|
yes, ntfs unless otherwise defined so it works when importing older or Linux pools where this is not a property |
Beta Was this translation helpful? Give feedback.
0 replies
-
|
I was bouncing of this problem as well without actually understanding why it works and sometimes it doesn't, I always assumed this has to be some Defender or Policy stuff I don't know about. Guess I need to RTFM again. |
Beta Was this translation helpful? Give feedback.
0 replies
-
|
As far as I understand it, an app can ask for the filesystem type to behave different example between fat and ntfs, |
Beta Was this translation helpful? Give feedback.
0 replies
-
|
bluescreen on 2.3.1.rc6, drag&drop of four files in explorer from a non-ZFS volume. Should have been the first write operation on volume after having mounted it (and very first use attempt after just having updated from rc5 and rebooted). Haven't determined if it's repeatable yet. |
Beta Was this translation helpful? Give feedback.
6 replies
-
|
Hmm so something is rotten in the state of vdev |
Beta Was this translation helpful? Give feedback.
-
|
because zfs_vdev_async_write_min_active and zfs_vdev_async_write_max_active are 0, from Registry. The bug will be to find why Registry goes bad |
Beta Was this translation helpful? Give feedback.
-
This is one of the of first windows |
Beta Was this translation helpful? Give feedback.
-
|
I'm no longer able to reproduce this under rc7; was that /0 fix expected to affect this? Haven't tried the rc7shim build as I'm not sure what that's intended to address. |
Beta Was this translation helpful? Give feedback.
-
|
divide fix was indeed the correct patch. You can ignore shim |
Beta Was this translation helpful? Give feedback.
-
|
This very common ms embrace oss pattern could change if ms would actually care about user and the fact those users actually use different components that are outside of the |
Beta Was this translation helpful? Give feedback.
5 replies
-
|
maybe some ms staffer will feel bad enough to pick this up and push it so it gets the attention of the higher powers unreachable to those ntfs-deniers and heretics (us), so zfs (or any other fs actually) gets added to the list, instead of gawking at unknown or uninitialized putting hours into inventing totally industry recognized and very valued experience of how-to-ntfs this or that - because there is no way to add zfs as an option to the operating system list of fs identifiers without ntfs sudoku or being within the 5 people on the planet with access to ntoskrnl internals |
Beta Was this translation helpful? Give feedback.
-
|
@inoperable, it's possible to patch native M$ code on the fly but that would require a lot of reverse engineering effort and would be pretty unstable with regard to Windows Updates. I still support your call! :) |
Beta Was this translation helpful? Give feedback.
-
We can go down the rabbit hole and try to use parts of Efiguard , at least that is the only thing I can think of. I discovered though early on that disabling DSE and patching additional parts of ntsoskrnl has the unfortunate side effect of randomness. I tried this out to solve another pain point: turing off DWM composition and enabling back the capability to turn it off :] hehe |
Beta Was this translation helpful? Give feedback.
-
|
So disabling patchguard/dse works, other arcane parts of the so-called security tend to notice that and you got yourself a nice introduction to windows quantum effects ;> |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for your practical insights. I wonder though what would it take for @reactos to enable an additionally installed FS to work and be registered/integrated in a seamless way (without relying on hacks and workarounds from FSD side). |
Beta Was this translation helpful? Give feedback.
-
|
I want to say a thank you from the bottom of my heart for this amazing software! I have been trying to get very important personal files of mine from a zfs pool in ubuntu desktop to my windows installation for 4 days now, and every single time i attempted to copy no matter how i did it, either though a vm or baremetal, the destination drive would get corrupted files which then caused my OS to completely lock up every single time. Now i am copying for 2 hours already and so far so good, so i think this is going to save my files by the looks of it. I barely use github to ever comment on anything but again.. thank you! <3 EDIT(~2 days later): |
Beta Was this translation helpful? Give feedback.
0 replies
-
|
Back from my Vacation to Canada (feel free to watch the youtube videos of the trip) - so it is time to get back into development. |
Beta Was this translation helpful? Give feedback.
1 reply
-
|
Cheers! |
Beta Was this translation helpful? Give feedback.
-
Wow is rocket science included in this repo? |
Beta Was this translation helpful? Give feedback.
1 reply
-
|
At the time, Norway's first rocket launch blew up spectacularly, so it felt like a humorous parallel. |
Beta Was this translation helpful? Give feedback.
-
Fresh OS latest version of OpenZFS. Performance looks awesome Storage Spaces would be like 10x slower.
Will test how stable it is when i copy some data.
|
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
This discussion was created from the release zfs-windows-2.3.1rc1.
Beta Was this translation helpful? Give feedback.
All reactions