On 20.07.25 09:45, Niels Thykier wrote:
@Bacula team: I have split this into two bugs for you.
- #1109499 remains RC and is about preinst defaulting to fail on a
non-interactive upgrade test.
The problem here is that this question and the default to "no" is a very
important safeguard against a critical upgrade problem WRT the DB
changes that are going to happen.
The upgrade path to v15 changes the biggest table in the database in a
way that it needs huge amount of storage (nearly a TB for my production
system for example) and if you fail to provide this then the upgrade
will fail and leave your database in an unusable and very hard to
recover state, leading to loss of data or the very least non running
backup and non-functioning restores.
We are kind-of between a rock and a hard place here, since there is no
way to programmatically check if the upgrade will succeed with remote
databases and non-standard database locations involved.
So IMO the only way to prevent a catastrophic failure is to ask the
person running the upgrade if it is safe and default to "no" to prevent
yes-yes-yes-spamming from ruining the day.
Grüße,
Sven.
