Changelog
makedumpfile (1:1.5.9-6) sid; urgency=medium
* [ Juerg Haefliger <email address hidden> ]
Limit the number of dumps kept on the local machine
Introduce a new config variable KDUMP_NUM_DUMPS that specifies how
many dumps should be kept on the machine. This can be use to prevent
running out of disk space. It is ignored if remote dumping (ssh or nfs)
is enabled. If there are too many dump files, older dumps are purged but
only after the current dump operation succeeded.
* [ Linn Crosetto <email address hidden> ]
kdump-config: add support for UEFI Secure Boot
kdump-config needs to query the state of UEFI Secure Boot in order to correctly
use kexec to boot the crash kernel. Specifically, kexec needs the "-s" option
to use the kexec_file_load system call to validate the image before loading it.
(Closes: #823003)
* [Hari Bathini <email address hidden>]
Fix networked kdump failure to reach remote server. Avoids
"Network is unreachable" message when trying to do remote dumps on either
SSH or NFS. (LP: #1571590)
* Add support for architecture-specific crashkernel param
On Debian, the crashkernel boot parameter is left undefined. On Ubuntu
it was defined in the kexec-tools package, modified specifically.
This parameter is now defined by the kdump-tools package and uses an
architecture-specific file containing the definition. A generic symlink
called /etc/default/grub.d/kdump-tools.cfg points to the architecture
specific file.
The default values are :
- All but ppc64el : 128M
- ppc64el : 2G-4G:320M,4G-32G:512M,32G-64G:1024M,64G-128G:2048M,128G-:4096M@32M
(LP: #1567539)
On s390x architecture, the crashkernel parameter is added to the
/etc/zipl.conf file.
/etc/default/kdump-tools is also modified to use the cio_ignore parameter
when kexec loads the crash kernel (LP: #1570775)
-- Louis Bouchard <email address hidden> Mon, 30 May 2016 11:39:07 +0200
