How did you determine the cpu activity?
Processes doing a lot of disk io operations will cause a lot of "D" states and will show in the load average but are generally using little or no actual cpu and won't show up on top.
And potentially creating (or attempting) to create links would be doing a lot of trivial disk iops that spend a lot of time waiting on the disk.
What kind of disk is the OS on? Even with a ssd the iop takes some small amount of real time (typically about 10-50us) but is a huge amount of time compared to the few cpu cycles it takes to create the iop.
On Thu, Jun 9, 2022 at 11:48 AM home user <mattisonw@xxxxxxxxxxx> wrote:
On 6/2/22 2:50 PM, Samuel Sieb wrote:
> On 6/2/22 13:03, home user wrote:
>> When doing my weekly patches (dnf upgrade), I get swamped with
>> "hardlink" error messages. There are a huge number of them, and they
>> fly past far too fast to see what dnf was doing when the messages
>> started. I found the messages in two log files. I put them on the
>> google drive. The file names with links to them on the google drive
>> are as follows:
>>
>> dnf.rpm.log.1
>> "https://drive.google.com/file/d/17VBjtSM_6YNSRcsvjmYRvuSRbnGkbMzp/view?usp=sharing"
>
> For some reason you have debug kernels installed. Try "dnf remove
> kernel-debug*" and see if that solves your problem.
I finally tested the fix this morning with the weekly run of "dnf
upgrade". I saw none of them "hardlink" errors. So Samuel's diagnosis
and cure is the solution.
This seems to have also fixed another problem. For the past few weeks,
the weekly "dnf upgrade" would take some 15-20 minutes doing some
"invisible" work, I think it was at the end of the cleanup phase. The
system monitor graph would show lots of CPU activity during that 15-20
minutes, but the system monitor process table and "top" would show very
few or no processes using more than 1% CPU. This workstation is a 3+
GHz, quad-core, 8 CPU machine. This did not occur during today's "dnf
upgrade". It does bother me that CPU-intensive processes seems to be
able to hide from "top" and the system monitor process table.
I do agree with Roger also. Some part of the patching process was not
being done properly.
I've marked this thread "SOLVED". Thank-you Samuel and Roger.
Bill.
_______________________________________________
users mailing list -- users@xxxxxxxxxxxxxxxxxxxxxxx
To unsubscribe send an email to users-leave@xxxxxxxxxxxxxxxxxxxxxxx
Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/users@xxxxxxxxxxxxxxxxxxxxxxx
Do not reply to spam on the list, report it: https://pagure.io/fedora-infrastructure
_______________________________________________ users mailing list -- users@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe send an email to users-leave@xxxxxxxxxxxxxxxxxxxxxxx Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@xxxxxxxxxxxxxxxxxxxxxxx Do not reply to spam on the list, report it: https://pagure.io/fedora-infrastructure