orion at cora.nwra.com
Tue Feb 23 00:10:35 UTC 2016
On 02/18/2016 03:16 PM, Heyman, Jerrold wrote:
> I previously wrote the below, and wanted to add that I have since tested
> on a newer version of Mageia, version 4.
> It makes use of rpm version 4.11.1 and has the same behavior as 184.108.40.206
> Does this mean that is the new behavior? Or is Mageia 4 broken as Mageia
> 3 is?
> Earlier today (2016/02/18), Jerry Heyman Jerrold.Heyman at emc.com
> <mailto:Jerrold.Heyman at emc.com> wrote:
> A quick scan of the archives leaves me confused about %preun sciptlets
> returning error (non-zero).
> There is a thread that says using %pre to fail an install/uninstall is a
> bad idea.
> I have been requested fail the removal of an RPM if the binary that was
> installed is currently an active process.
> I’ve written a %preun scriptlet that returns non-zero if the binary is
> actively running.
> On CentOS 6.7 (rpm version 4.8.0) I get the expected behavior. When the
> scriptlet returns non-zero, the removal is terminated and no change to
> the rpmdb.
> On Mageia 3 (rpm version 220.127.116.11), the non-zero return code appears to
> be ignored and the rpm is removed (files and rpmdb updated).
> Was there a change somewhere between 4.8 -> 18.104.22.168 that permanently
> altered the behavior or is it just a bug in 22.214.171.124?
I'm not authoritative, but it's my experience that rpm has been moving
to ignoring the return codes of all of the scripts in order to keep the
overall transaction process deterministic.
Technical Manager 303-415-9701 x222
NWRA/CoRA Division FAX: 303-415-9702
3380 Mitchell Lane orion at cora.nwra.com
Boulder, CO 80301 http://www.cora.nwra.com
More information about the Rpm-list