...(rhbz#1611631)
Adapt spec file more akin to upstream version including:
. assuredly skip servicelog-related binaries even when build-time
prerequisites are present on suitable systems (9f24448d8)
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
...for release candidate: Pacemaker-2.0.0-rc6,
for full details, see included ChangeLog file or
https://github.com/ClusterLabs/pacemaker/releases/tag/Pacemaker-2.0.0-rc6
Adapt spec file more akin to upstream version including:
. new procps-ng and psmisc dependencies with -cli and -cts, for e.g.
"ps/sysctl/uptime" and "killall" invocations, respectively (a4ad8183a)
. move crm_node to -cli (a94a1ed58)
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
...for release candidate: Pacemaker-2.0.0-rc5,
for full details, see included ChangeLog file or
https://github.com/ClusterLabs/pacemaker/releases/tag/Pacemaker-2.0.0-rc5
Adapt spec file more akin to upstream version including:
. new coreutils dependency for "post" scriptlet of -cli,
for "mv" invocation (c2b16165d)
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
...for release candidate: Pacemaker-2.0.0-rc4,
for full details, see included ChangeLog file or
https://github.com/ClusterLabs/pacemaker/releases/tag/Pacemaker-2.0.0-rc4
. as a special note, previous release candidate, rc3, had rolling upgrades
broken, and if that is required, that particular release shall be
skipped in the upgrade path altogether
Adapt spec file more akin to upstream version including:
. as part of the update process, possibly move old log files as implicitly
used prior to 2.0 so there's a (limited) continuity with the new implicit
location, preventing clutter and confusion (ce2e74c99, 765eb3f24)
. move cts-exec-helper from -cli under main package (a2dc2a67e)
. -cts backed with new helpers and, tangentially, dummy systemd service
file transiently generated on-demand again (fa2d43445, d52b001b1)
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
...for release candidate: Pacemaker-2.0.0-rc3,
for full details, see included ChangeLog file or
https://github.com/ClusterLabs/pacemaker/releases/tag/Pacemaker-2.0.0-rc3
Adapt spec file more akin to upstream version including:
. new --without legacy_links conditional (c8a7e5225)
. reflect name change of the auxiliary daemons
(e4f4a0d64, db5536e40, e2fdc2bac + 9ecbfea1c, 038c465e2 + ed8ce4055a)
. new dummy systemd service for -cts (bf0a22812)
. honor system-wide crypto policies once for all, via package-build-time
configurable "pcmk_gnutls_priorities" defaulting to @SYSTEM as prescribed
in https://fedoraproject.org/wiki/Packaging:CryptoPolicies
(based on b3dfce1d3)
Adapt spec file akin to current packaging guidelines including:
. make -nagios-plugins-metadata package noarch
Drop patches that got subsumed in upstream
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
... for release candidate: Pacemaker-2.0.0-rc2,
for full details, see included ChangeLog file or
https://github.com/ClusterLabs/pacemaker/releases/tag/Pacemaker-2.0.0-rc2
Adapt spec file more akin to upstream version including:
. out-of-tree change from 1.1.18-2 build got subsumed (508ad52e7)
. %%{_sysconfdir}/pacemaker path got properly owned
(-cli package; f6e3ab98d)
. -libs package started to properly declare Requires(pre): shadow-utils
(293fcc1e8 + b3d49d210)
. some build conditionals and dependencies dropped for no longer
(snmp, esmtp; f24bdc6f2 and 1f7374884, respectively) or never
being relevant (~bison, byacc, flex; 61aef8af4)
. some dependencies were constrained with new or higher lower bounds:
corosync needs to be of version 2+ unconditionally (ccd58fe29),
ditto some others components (~GLib, 1ac2e7cbb), plus both 2 and 3
versions of Python are now (comprehensively for the auxiliary
functionality where used) supported upstream with the latter being
a better fit (453355f8f)
. package descriptions got to reflect the drop of legacy low-level
cluster infrastructures (55ab749bf)
Adapt spec file akin to current packaging guidelines including:
. drop some redundant/futile expressions (defattr, "-n %%{name}-libs"
instead of plain "libs", "timezone hack"), add some notes for future
. make -cts and -doc packages noarch (former enabled with 088a5e7d4)
. simplify "systemd_requires" macro invocation, and relax it to
"systemd_ordering" for -remote package where possible so as not
to drag systemd into a lightweight system setup (e.g. container)
needlessly
. adjust, in a compatible way, common ldconfig invocation with
post{,un} scriptlets
(https://fedoraproject.org/wiki/Changes/Removing_ldconfig_scriptlets)
. drop some more unuseful conditionals (upstart_job)
Apply some regression fixes on top as patches (PR #1457, #1459)
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
... for point release: Pacemaker-1.1.18,
for full details, see included ChangeLog file or
https://github.com/ClusterLabs/pacemaker/releases/tag/Pacemaker-1.1.18
Make -libs-devel package dependencies arch-qualified
(-cts hasn't been switched at this time, pending further cleanup)
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
...beyond Pacemaker-1.1.16 (commit a39ea6491), including:
. prevent FTBFS with new GCC 7 (a7476dd96)
Adapt spec file more akin to upstream version including:
. better pre-release vs. tags logic (4581d4366)
... for full details, see included ChangeLog file or
https://github.com/ClusterLabs/pacemaker/releases/tag/Pacemaker-1.1.16
- Adapt spec file more akin to upstream version including:
. clarify licensing, especially for -doc (f01f734)
. fix pacemaker-remote upgrade (779e0e3)
. require python >= 2.6 (31ef7f0)
. older libqb is sufficient (based on 30fe1ce)
. remove openssl-devel and libselinux-devel as BRs (2e05c17)
. make systemd BR pkgconfig-driven (6285924)
. defines instead of some globals + error suppression (625d427)
- Rectify -nagios-plugins-metadata declared license and install
also respective license text
...using the upstream patches overhauling toolchain hardening (Fedora
natively uses -fstack-protector-strong so this effectively relaxed
stack protection is the only effect as hardened flags are already
used by default:
https://fedoraproject.org/wiki/Changes/Harden_All_Packages)
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
... for full details, see included ChangeLog file or
https://github.com/ClusterLabs/pacemaker/releases/tag/Pacemaker-1.1.15
- Adapt spec file more akin to upstream version:
. move xml schema files + PCMK-MIB.txt (81ef956), logrotate configuration
file (ce576cf; drop it from -remote package as well), attrd_updater
(aff80ae), the normal resource agents (1fc7287), and common directories
under /var/lib/pacemaker (3492794) from main package under -cli
. simplify docdir build parameter passing and drop as of now
redundant chmod invocations (e91769e)
... Pacemaker-1.1.15-rc3, for full details, see included ChangeLog file or
https://github.com/ClusterLabs/pacemaker/releases/tag/Pacemaker-1.1.15-rc3
- Drop fence_pcmk (incl. man page) from the package (no use where no
CMAN)
- Drop license macro emulation for cases when not supported natively
(several recent Fedora releases do not need that)
...Pacemaker-1.1.15-rc1, for full details, see included ChangeLog file or
https://github.com/ClusterLabs/pacemaker/releases/tag/Pacemaker-1.1.15-rc1
- Adapt spec file more akin to upstream version (also to reflect recent
changes like ability to built explicitly without Publican-based docs)
Note that patch disabling original built-in notification mechanism is
no longer needed as the same is achieved natively in the sources,
and there's brand new, more convoluted (and allegedly compatible)
implementation of the same.
...beyond Pacemaker-1.1.14
(commit 5a6cdd1), but restore old-style notifications to the state at
Pacemaker-1.1.14 point release (disabled)
- Definitely get rid of Corsync v1 (Flatiron) hypothetical support
- Remove some of the spec file cruft, not required for years
(BuildRoot, AutoReqProv, "clean" scriptlet, etc.) and adapt the file
per https://github.com/ClusterLabs/pacemaker/pull/965
... introduced with Fedora 23 Mass Rebuild.
It will likely happen again as the incrementing script, rpmdev-bumpspec,
doesn't account for "integer will be yielded from the macro expansion"
(not that it can do it in any simple way, and even if it could,
propagation of the increment back to the origin is pure sci-fi now).
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
