On 2023-03-18 14:53, Don Spam's Reckless Son wrote:
> Carlos E.R. wrote:
>> On 2023-03-18 08:35, Don Spam's Reckless Son wrote:
>>> Andrew wrote:
>>>> Tristan Miller wrote:
>>>>> Greetings.
>>>>>
>>>>> Occasionally when I boot my machine, the system pauses for a minute
>>>>> or two with the message, "A start job is running for Purge old
>>>>> kernels".
>>>>>
>>>>> If I understand correctly, purging old kernels simply means
>>>>> uninstalling them.  If this is the case, why is this something that
>>>>>  boot has to block for?  I mean, once the system is up an running,
>>>>> I can always use zypper or rpm to manually remove old kernels.  So
>>>>> it's obviously something that *can* be done without interfering
>>>>> with my use of the machine.  I get why the bootup script might want
>>>>> to clean up old kernels every once in a while, but why can't it
>>>>> just launch a process that does this unobtrusively in the
>>>>> background?
>>>>>
>>>>> Regards, Tristan
>>>>>
>>>>
>>>> Having just updated a kernel on a Leap 15.4 machine, the system still
>>>>  waits for the purge to complete.
>>>>
>>>> The purge of the old kernels runs in parallel to the setup of my
>>>> "wicked managed network interfaces".  On a system with SSD discs the
>>>> purge takes only slightly longer than the wicked setup.  On my older
>>>> system the purge takes well over a minute, I'd guess at 80 seconds.
>>>>
>>>> The older system has another problem anyway, my /boot partition is
>>>> over 500MB and has around 50% free with the current kernel and the -1
>>>> kernel. This is insufficient when it comes to installing a new kernel
>>>> and I'm going to have to start getting rid of the -1 kernel before
>>>> installing the new one.  The beast is dual-boot with Windows 10 and I
>>>> am not prepared to risk moving the main Windows partion which is just
>>>> behind /boot.
>>>
>>> An update has come out which fixes the problem.
>>> purge-kernels-service-0-150200.8.6.1.noarch
>>>
>>> It turns out the behaviour was deliberate, YaST -> Software
>>> Management -> [search for "purge"], and then look at "Change Log".
>>> The entry for 06 May 2021.
>>
>> Good find.
>>
>> Telcontar:~ # rpm -q --changelog purge-kernels-service | less
>>
>> * Thu May 06 2021 ...@suse.de
>> - Add ZYPP_LOCK_TIMEOUT=-1 to keep waiting for the lock (boo#1184399).
>>
>> Which is:
>>
>> https://bugzilla.opensuse.org/show_bug.cgi?id=1184399
>>
>> Basically, purge-kernels failed because the user was running zypper at
>> the same time. So they added loop or something waiting for package
>> management to be finished and then start the actual purge.
>>
>>
>>
>> But this is about a lock condition inside zypper, not about the
>> service holding the boot process. Different issue.
>>
>>
>>
>>
>> The update you mentions says:
>>
>>
>> openSUSE-SLE-15.4-2023-793 - Recommended update for purge-kernels-service
>>
>> This update for purge-kernels-service fixes the following issues:
>> - Change systemd service type to 'exec' (bsc#1198668)
>>
>> References:
>> 1198668 (bugzilla) : Tumbleweed: Purge old kernels service block
>> during boot
>>
>>
>> Comment 11 Michal Koutný 2022-05-13 12:57:12 UTC
>> ...
>> Thirdly, I can see the purge-kernels.service is of Type=oneshot, which
>> means it'll block subsequent jobs until the ExecStart= command
>> finishes. Not sure if there's actually necessity to wait after kernel
>> removal is done, the Type=simple or Type=exec would only wait for the
>> command start and the removal may run in parallel (even with jobs that
>> specify After=purge-kernels.service).
>>
>>
>>
>> Curious. We'll see.
>>
>>
>
> In the interests of full disclosure I have to point out that the date on
> the most recent entry in the Change Log (Change service type to exec
> (boo#1198668)) is 13 May 2022.
> Has it really taken 10 months to release that fix?

If you look at the bugzilla, they released it somewhere in mid 2022 for
Tumbleweed. Then somebody recently mentioned that the bug also affected
15.4, and *then* a patch for 15.4 was also released.

I told here to bring up the issue in the openSUSE mail list. I wrote it
myself (2022-04-20), but the people affected did not speak up and I was
subsequently busy. The next step would be to write a bugzilla.

If you don't write a bugzilla, things are never solved. Things written
here do not exist.

> I was watching out for things on my 10-year-old Laptop (no SSD of
> course) and saw no sign of the extended wait for a purge which I had
> seen after previous kernel updates.
> The system was running rather slowly after that boot completed, as I
> expected if a purge was running in the background.
> (this was running with the "-1" kernel because the newest one was
> freezing at a very early stage in the boot process, a different problem).

--
Cheers, Carlos.

Carlos E.R. wrote:
>
> If you look at the bugzilla, they released it somewhere in mid 2022 for
> Tumbleweed. Then somebody recently mentioned that the bug also affected
> 15.4, and *then* a patch for 15.4 was also released.
>
> I told here to bring up the issue in the openSUSE mail list. I wrote it
> myself (2022-04-20), but the people affected did not speak up and I was
> subsequently busy. The next step would be to write a bugzilla.
>
> If you don't write a bugzilla, things are never solved. Things written
> here do not exist.
>

I tried creating a bugzilla account a year or two ago. Tried. Can't
remember what the problem was, maybe I was not eligible, or maybe there
was a technical problem.

On 2023-03-18 23:31, Don Spam's Reckless Son wrote:
> Carlos E.R. wrote:
>>
>> If you look at the bugzilla, they released it somewhere in mid 2022
>> for Tumbleweed. Then somebody recently mentioned that the bug also
>> affected 15.4, and *then* a patch for 15.4 was also released.
>>
>> I told here to bring up the issue in the openSUSE mail list. I wrote
>> it myself (2022-04-20), but the people affected did not speak up and I
>> was subsequently busy. The next step would be to write a bugzilla.
>>
>> If you don't write a bugzilla, things are never solved. Things written
>> here do not exist.
>>
>
> I tried creating a bugzilla account a year or two ago.  Tried.  Can't
> remember what the problem was, maybe I was not eligible, or maybe there
> was a technical problem.

There is no "eligible" thing.

--
Cheers, Carlos.