[elbe-devel] Elbe 2.2 requests more memory for PostgreSQL clusters than 1.2 did, which breaks existing devices running Debian 7 (wheezy)

[Manuel Traut]

Hi,

> Building a <suite>wheezy</suite> image with elbe 2.2 results in a different PostgreSQL cluster configuration than with elbe 1.2. In particular the shared_buffers setting in /etc/postgresql/9.1/main/postgresql.conf differs - with elbe 1.2 it used to be:
> 
> shared_buffers = 24MB
> 
> while with elbe 2.2 it is:
> 
> shared_buffers = 32MB
> 
> The increased amount of shared buffer memory requested by PostgreSQL cannot be allocated on my hardware (existing product, which we still need to support) and the RDBMS fails to start.
> 
> Long story short: same XML, same build machine, same target device, different elbe version -> product broken.
> 
> This has probably something to do with the fact, that elbe 2.2 creates a Debian 8 (jessie) initvm while the 1.2 does Debian 7 (wheezy). Note, that the PostgreSQL conffiles in /etc/postgresql/ get generated upon package installation, so obviously some maintainer script works differently when run in chroot on Debian 8 and 7 - maybe the postinst script gets something from the host's environment that chroot cannot hide?
> 
> Any idea how to fix that?

not at the moment, i created an issue on github for this:
https://github.com/Linutronix/elbe/issues/148