On 3/26/19 2:27 PM, Axel Beckert wrote:

Hi JC,

On Tue, Mar 26, 2019 at 01:37:32PM -0700, Japheth Cleaver wrote:

...

I'm pushing for a release of 4.3.29 relatively soon. Cool, thanks!

...

I've been trying to go through the backlog to identify un-applied patches, but I know there are some that I'm missing. If you have build fixes or runtime changes that have not yet been put in in 4.3.29 already (see: https://sourceforge.net/p/xymon/code/HEAD/tree/branches/4.3.29/Changes), I'd appreciate if you could point them out. Here are the patches Debian applies to 4.3.28: https://salsa.debian.org/debian/xymon/tree/master/debian/patches

I'll try to categorize them quickly:

Missing fixes:

• 39_kfreebsd-makefile.patch: Support for the GNU/kFreeBSD architecture/platform (i.e. FreeBSD kernel with GNU instead of BSD userland).

• 42_bbcombotest-fix.patch: Fix bbcombotest: "Could not access hobbitd board, error 0". (Don't have more details, sorry. Christoph might perhaps remember more details.)

• 63_netstat-ant-vs-ipv6-address-truncating.patch: Port monitoring seems to cut off IPv6 addresses. This seems to be unrelated to "Fix RRD parsing for recent netstat (net-tools) on Linux".

• 66_apache2.4.patch: Some Apache 2.4 fixes. Since Apache 2.2 is End of Life already, IMHO you do not need to care for the old syntax anymore. Then again, some distributions with long-term support might still have Apache 2.2, so you might want to cross-check if that has some impact there.

• 84_fix_compilation_on_GNU_Hurd.patch: Fixes compilation on the GNU Hurd architecture/platform.

• 90_fix-spelling-errors.patch: Spelling error fixes.

Can't be used 1:1, but shows what is missing:

• 51_hardening-buildflags.patch: CFLAGS are hardcoded and don't respect any such environment variable.

Feature patches:

• 27_hobbit_files_ifexist.patch: Adds an "ifexist" feature to the files check.

Might be no more needed:

• 69_disk-no-duplicate-root.patch: Ignore duplicate submissions for the "/" partition. For a while Debian and some other distributions report the root partition twice because it got remounted during boot. I though don't see that behaviour on a current Debian Stable anymore.

State unclear:

• 33_526176-ldap.patch: There seems to be an LDAP API change necessary some time in the future (for 10 years now) and this seems to be the lazy workaround. See https://bugs.debian.org/526176

  I must admit, I'm not sure if that's still necessary. This has been added to Debian in April 2009, but https://sourceforge.net/p/xymon/code/HEAD/tree/branches/4.3.29/Changes says that Hernik worked on this in November 2008 and it was included in the 4.2.2 release in December 2008. And Debian shipped 4.2.2 in January 2009 and added this patch in April 2009 with 4.3.0 beta2. So those things don't seem to have overlapped each other. And these changelog entries don't sound as if this has been fixed properly

  ---

  r5990 | storner | 2008-11-28 07:43:02 +0100 (Fri, 28 Nov 2008) | 1 line Changed paths: M /branches/4.2.2/build/test-ldap.c

  Build properly with new OpenLDAP API by using deprecated functions.

  […]

  r5995 | storner | 2008-11-28 10:27:42 +0100 (Fri, 28 Nov 2008) | 1 line Changed paths: M /branches/4.2.2/bbnet/ldaptest.h

  Build properly with new OpenLDAP API by using deprecated functions (missed ldaptest.h in previous commit)

  But then again, I'd wonder why this "lazy" fix worked for now 10 years without causing issues with the deprecated LDAP functions finally being removed.

(Probably)already included patches:

• 00_htmlcontenttype.patch: Seems to be "Ensure Content-Type always set in HTML headers (Thanks, Christoph Berg)"

• 24_hobbitclient-tmpfs.patch: Might be "Ignore additional common tmpfs partitions on recent Linux", but you might want to cross-check.

• 87_fix_logfetch_FTBFS_with_glibc_2.26.patch: Applied in r8030.

These seem to be Debian specific patches are are marked as not needing to be forwarded to upstream, so you can safely ignore them:

• 03_doc-paths.patch: Debian-specific path changes.

• 09_hobbitclient-debian.patch: Adds a dpkg section to client reports.

• 12_hobbitvars.patch: Debian-specific path changes.

• 21_FHS-instead-FSSTND-in-example-in-man-page.patch: Debian-specific path changes.

• 30_prefer-packaged-temp-plugin-over-unpackaged-devmon.patch: Different default settings with regards to one plugin.

• 45_fix-configure-for-multiarch.patch: Debian-specific generated paths.

• 48_png-multiarch.patch: Debian-specific generated paths.

  Kind regards, Axel

---

Xymon mailing list Xymon at xymon.com http://lists.xymon.com/mailman/listinfo/xymon

while Apache 2.2 is EOL, there are a LOT of distros and installations that still have it running running back ports.

I'd say it costs nothing to leave in and has the potential of creating a lot of heartache by taking it out... Even if Debian does like it that way.

On 3/26/19 4:41 PM, Axel Beckert wrote:

Hi,

On Tue, Mar 26, 2019 at 04:23:33PM -0700, Bruce Ferrell wrote:

...

...

• 66_apache2.4.patch: Some Apache 2.4 fixes. Since Apache 2.2 is End of Life already, IMHO you do not need to care for the old syntax anymore. Then again, some distributions with long-term support might still have Apache 2.2, so you might want to cross-check if that has some impact there. while Apache 2.2 is EOL, there are a LOT of distros and installations that still have it running running back ports. So far I only can come up with RHEL/CentOS 6, Debian 7 Wheezy ELTS (EoL in about two months) and Ubuntu 12.04 ESM (i.e. paid-only LTS) who still provide Apache 2.2.

At least Wheezy and Ubuntu 12.04 ship Xymon out of the box and I would be surprised if anyone on these releases follows upstream source code instead of using the distro-provided packages (with older Xymon versions).

So probably the only still relevant distro shipping Apache 2.2 is RHEL/CentOS 6. JC probably knows better how wide-spread Xymon is there.

And variants of RHEL/Centos... The first that comes to mind is Scientific Linux out of Fermi Labs.

I do agree the Deb/Ubuntu types probably don't build from scratch.

...

I'd say it costs nothing to leave in IMHO the cost is primarily to figure out if a case switch is needed between Apache 2.2 and 2.4.

The patch _is_ needed if you want to run your Apache 2.4 without mod_access_compat — which probably will vanish in some future Apache release as its sole purpose is to provide Apache 2.2 authentication/authorization syntax in Apache 2.4, i.e. needs to be applied at some point in the future for sure.

My observation on my systems, the "case switch" is right in the Apache configuration file as Apache directives, testing for the correct apache module.

The fact the the Apache configuration "just works" no matter WHAT, was a very pleasant change from ye olden dayes, when I had to craft the config myself.... Thanks JC!

...

and has the potential of creating a lot of heartache by taking it out... Only if there are people out there who don't rely on distro-built packages and want the newest version of Xymon on their oldest boxes,

You mean the vast majority of systems admins?  As a systems admin, I don't *tend" to do forklift updates... the entire system to update an application.  That is a HUGE amount of pain and effort and I tend to develop resentments around it... To the point of finding things that aren't so problematic.

So dropping it means running systems suddenly can't update.  OK, can't is probably an overstatement... It's not as bad as demanding a whole different glibc but I'd say it's NOT particularly friendly to people who use the code or you want to use it.