Running apt-get -V full-upgrade which updated atop - atop:amd64
(2.10.0-3, 2.11.1-1)
    it finished with Error: Sub-process /usr/bin/dpkg returned an error
code (1)
    Said also something not being able to convert todays logfile.

    Ran it again later, same outcome. This time I looked, if it is
shutting down atop, while doing a conversion. It did not.

    Backed up all atop logs, shut down atop manually. Ran apt again.

    Now it complains after long pause, while running atopconvert, not
being able to convert yesterdays log.

    Tried manually converting yesterdays atop log. It finishes fine:
    atopconvert /var/log/atop/oldsh/atop_20250408
/var/log/atop/oldsh/atop_20250408_newformat
    Version of /var/log/atop/oldsh/atop_20250408: 2.10
    Version of /var/log/atop/oldsh/atop_20250408_newformat: 2.11
    Samples converted: 1441
    Tried other logs also, they all converted fine. Trying atopconvert on
todays log said it is already at 2.11

    Removed all atop logs, ran apt again and now it finally finishes
without complaints.


    rgds

So you're upgrading from unstable, 2.10.0.3 to unstable, 2.11.1-1.
Right?

It would be incredibly helpful to have the verbatim output of all calls.

Can you reproduce the issue with the backed up logs? In that case, can
you run postinst configure with bash -x so that we can see what the
script actually tries to do?

Greetings
Marc

Upgraded, done. After actions described it got installed. Debian
testing, yes, did reportbug output not include that?
script, it was "atop postinst: error converting
/var/log/atop/atop_20250409 to new atop raw file format"

It clearly had no problem writing files. It wrote many of them, like:
/var/log/atop/atop_20250409
/var/log/atop/atop_20250409_oldformat
/var/log/atop/atop_20250409_oldformat_oldformat
after few retries with apt full-upgrade.

Deleted them afterwards.

How do I reproduce it, when I have done installing/upgrading? As I have
repeatedly been told, debian does not support downgrades etc etc. I do
find it odd that debian testing makes it so cumbersome to simply
downgrade something, "this is how debian works" as I have been told.

Deleted these backed up logs also some time before bugreport, I had them
just for my atopsar runs which I have done for now and did not need them
anymore. Or so I thought.

Initial feeling was that perhaps atop running prevented postinst from
actually getting active logfile converted. Not sure. It failed with atop
shut down also. Only odd thing about my atop install is its log
interval, 60. Does postinst script have some timelimit for running? Just
guessing.

When I watched it work it was busy running atopconvert
mytodayslog_oldformat mytodayslog, thats all I recall seeing. Same
command was used to run my manual atopconvert attempts against all logs
one after another, just changed to keep original logs intact. All
converted ok, only todays one was saying it has already new format...

rgds,
hy

Control: tags -1 unreproducible
thanks

Unfortunately, reportbug doesn't report the version of the package being
upgraded from. A full transcript of the upgrading process would have
helped.

You should do that in the future. It makes your bugreports easier to act
on.

I will adapt the postinst to be a bit more robust and to not error out
if it doesn't convert. Better lose today's data than aborting the
upgrade.

Debian testing is an unreleased development version that is targeted at
developers who know their way around the distribution. It is generally
not recommended to run Debian testing if you don't know how to debug
package upgrades.

In the situation in question, you would have to pull the older version
of the package from snapshot.debian.org, install it (that is not
supported but works in most of the cases. If it doesn't you have to
clean up) and re-run the upgrade.

It would have been a good idea to have a backup.

I cannot reproduce the issue in my test environments. Without more
input and with the data deleted, there is probably nothing to do. Noone
else has reported a similar issue since the last testing migration, so
without a reproducer there is nothing I can do for you.

Greetings
Marc

You probably missed the "upgraded" part. Its upgraded, nothing more to
"do" for me. It was just to let you know, perhaps you can find way to
improve it for others when this goes to stable. If not, then not. My
local anomaly.

I do know about snapshots, yes. Done downgrading that way few times...
IMHO Debian could make these packages available for testing repo, so
such "tests" can be done without hunting stuff down from snapshots.

"I will adapt the postinst to be a bit more robust and to not error out

rgds,
hy

I didn't miss that. You're using an unreleased development version of
Debian, so we kind of expect you to be prepared to debug. It's kind of
sad that you aren't, so I'll *shrug* and continue with my work.

At least part of your body report pointed me to the fact that the script
needed love.

Greetings
Marc

Setting up atop (2.11.1-3) ...
atop postinst: error reading converted log file /var/log/atop/atop_20250825.


Debian testing. Also had 60 sec interval in use for atop. Also observed atop running during conversion attempt. Leftover file for todays log also remained...

e

Seeme liks gibberish to me, sorry.