Bug 1860644 - unable to continue the installation after reset the size of LVM thin provisioning partition
Summary: unable to continue the installation after reset the size of LVM thin provisi...
Keywords:
Status: CLOSED DUPLICATE of bug 1868623
Alias: None
Product: Fedora
Classification: Fedora
Component: anaconda
Version: 33
Hardware: Unspecified
OS: Unspecified
unspecified
unspecified
Target Milestone: ---
Assignee: Anaconda Maintenance Team
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2020-07-26 06:53 UTC by lnie
Modified: 2020-09-04 13:41 UTC (History)
9 users (show)

Fixed In Version:
Clone Of:
Environment:
Last Closed: 2020-09-04 13:41:22 UTC
Type: Bug
Embargoed:


Attachments (Terms of Use)
screencast (69.27 KB, video/webm)
2020-07-26 06:53 UTC, lnie
no flags Details
storage.log (525.65 KB, text/plain)
2020-07-26 06:53 UTC, lnie
no flags Details
anaconda.log (32.96 KB, text/plain)
2020-07-26 06:55 UTC, lnie
no flags Details
screencast (651.58 KB, video/webm)
2020-08-18 03:29 UTC, lnie
no flags Details
screenshot of the pxe boot crash (86.82 KB, image/png)
2020-08-18 04:05 UTC, lnie
no flags Details
everything screenshot (160.09 KB, image/png)
2020-08-18 04:16 UTC, lnie
no flags Details

Description lnie 2020-07-26 06:53:09 UTC
Created attachment 1702420 [details]
screencast

Description of problem:
Boot the installer,set / partition'Device type to Lvm,then change it to Lvm thin provisioning, then change the size of the partition,and you will find the cursor keep rotating 

Version-Release number of selected component (if applicable):

always 
How reproducible:
always

Steps to Reproduce:
1.
2.
3.

Actual results:


Expected results:


Additional info:

Comment 1 lnie 2020-07-26 06:53:57 UTC
Created attachment 1702421 [details]
storage.log

Comment 2 lnie 2020-07-26 06:55:58 UTC
Created attachment 1702422 [details]
anaconda.log

Comment 3 Ben Cotton 2020-08-11 13:49:33 UTC
This bug appears to have been reported against 'rawhide' during the Fedora 33 development cycle.
Changing version to 33.

Comment 4 Vladimír Slávik 2020-08-17 17:55:32 UTC
Inie, thank you for the report! I can't reproduce using both latest master and 33.25, but that's on a everything pxe boot while you had a live cd. I'd like to try reproducing what you had and verifying it's fixed as well. Can you please tell me what version/build/date of live cd you used? 

FWIW, I really like the videos you make, they make it really easy to see "your perspective".

Comment 5 lnie 2020-08-18 03:24:16 UTC
I can't believe that I have forgotten to put the compose information,sorry about that.
Anyway,I reported the bug in 07-26,and I reproduced the bug with Fedora-Workstation-Live-x86_64-Rawhide-20200717.n.0.iso,
which is the best media I still keep on my local machine,and I'm gonna to attach a longer video for the reproducer.
Btw,the symptom  seems get worse with Fedora-Workstation-Live-x86_64-Rawhide-20200811.n.0.iso(and Fedora-Server-dvd-x86_64-Rawhide-20200811.n.0.iso),
and guess I need to report a new bug.

Comment 6 lnie 2020-08-18 03:29:36 UTC
Created attachment 1711671 [details]
screencast

Comment 7 lnie 2020-08-18 04:05:23 UTC
Created attachment 1711678 [details]
screenshot of the pxe boot crash

Comment 8 lnie 2020-08-18 04:15:34 UTC
> I can't reproduce using both latest master and 33.25, but that's on a everything pxe boot while you had a live cd
I've tested Fedora-Server(and everything)-x86_64-Rawhide-20200811.n.0,which has 33.25-1 on a beaker server,
which performs installation with pxe boot,and it affected by this bug

Comment 9 lnie 2020-08-18 04:16:15 UTC
Created attachment 1711681 [details]
everything screenshot

Comment 10 Vladimír Slávik 2020-08-20 18:32:14 UTC
Vojto, I think you return only positive sizes now? Or is this another instance of bug 1853071 and bug 1868623 as you predicted somewhere?

---

I managed to reproduce this with latest anaconda. For me the first resize did not break because it was to a smaller size. Resizing to a larger size did crash. I get two error dialogs, like the reporter's screenshots show. What's funny is that resizing down and then incrementally up did not crash, but resizing again down and then back up to the last reached amount did. That is repeatable too. As a precise example, 15.17 -> 13 -> 15 caused the crash in another attempt.

I tried to include blivet master in the updates image, which should include the recent fix. But currently that always ends for me in an error popup about broken module dependencies for ***:el :-/

---

The interesting part of the logs (first anaconda-tb-***) that matches the error in reporter's storage.log is probably this:
> WARNING org.fedoraproject.Anaconda.Modules.Storage:DEBUG:blivet:vg fedora_fedora has -4 MiB free
> WARNING org.fedoraproject.Anaconda.Modules.Storage:WARNING:dasbus.server.handler:The call org.fedoraproject.Anaconda.Modules.Storage.DeviceTree.Scheduler.GetContainerFreeSpace has failed with an exception:
> (...)
> WARNING org.fedoraproject.Anaconda.Modules.Storage:OverflowError: -4194304 not in range 0 to 18446744073709551615

The second traceback just adds another repetition of the exact same message sequence, it's probably the same error coming through a different module.

I think it's interesting that the error always mentions 4 MiB. I read that as the size being off by a single extent. Maybe there's some rounding error somewhere, such as using round instead of math.floor?

---

Also related: https://github.com/rhinstaller/anaconda/pull/2748

Comment 11 Vojtech Trefny 2020-08-21 07:10:47 UTC
(In reply to Vladimír Slávik from comment #10)
> Vojto, I think you return only positive sizes now? Or is this another
> instance of bug 1853071 and bug 1868623 as you predicted somewhere?
> 

I think this is a duplicate of 1868623. The -4 MiB free space appears immediately after recalculating thin metadata size, same as in 1868623.

DEBUG:blivet:Recommended metadata size: 8.5078125 MiB
DEBUG:blivet:Rounded metadata size to extents: 12 MiB
DEBUG:blivet:Adjusting size from 16364 MiB to 16360 MiB
...
DEBUG:blivet:fedora_localhost-live size is 20 GiB
DEBUG:blivet:vg fedora_localhost-live has -4 MiB free

We still consider a negative size to be a valid value. 1868623 was a bug in metadata size calculation, but that doesn't mean it is the only bug or that all negative sizes are bugs.

Comment 12 Vladimír Slávik 2020-09-04 13:41:22 UTC
I am not unable to reproduce using anaconda and blivet master. So I guess this is resolved by the blivet changes Vojta linked - thank you! Thus closing. Please feel free to reopen if needed.

*** This bug has been marked as a duplicate of bug 1868623 ***


Note You need to log in before you can comment on or make changes to this bug.