Call to logrotate-setup-validate can take a lot of time and fill the state
file with some message from logrotate and in the same time the promise can
kick in resulting with false-positive alarm.

By using the temporary file such race condition is avoided. Promise running
periodically will detect problem with logrotate setup.

While generating haproxy configuration (including it's CRT list) the specific
order of entries is used, so that wildcard domains end up last. Thanks to this
they work as a catch-all and allow specific domain to take precedence. Care
is taken to support *.example.example.com and *.example.com situation - so
tree like possibility of wildcards.

Anonymous in-place ACL are used per each domain, instead of per-shared
instance grouping in order to avoid situation like *.example.com and
example.com having single ACL, thus resulting with catch-all kicking in too
fast.

For the precision in the haproxy configuration and simplifcation of the regular
expressions the -m reg is used, so that host_only can be applied, which also
lowercases the hostname.

Notes:
 * test00cluster_request_instance_parameter_dict changed due to sorting slaves
   in test's requestSlaves
 * the test infrastructure has been improved to assure repetition of the
   situation
 * tests in TestSlaveHostHaproxyClash are asserting that correct domain AND
   that specific certificate have been used while serving given frontend
   configuration

See merge request nexedi/slapos!1438

this is a fixup of 54a08186

For slapos-node package compilation, we use DESTDIR pointing to temp
location and PREFIX pointing to future /opt/slapos.

Without this commit, bison look at wrong directory for data:

bison: cannot open file `/opt/slapos/parts/bison/share/bison/m4sugar/m4sugar.m4'

We configure haproxy with "verify optional", which makes haproxy request
a client certificate, but accept the case where client does not present
a certificate, but as described in [1], if client present a certificate
and this certificate can not be verified, handshake is aborted. This is
not what we want, we want to treat the case of a non verified
certificate same as the case of the absence of certificate.

This configures haproxy accordingly, using "crt-ignore-err all" to allow
handshake anyway.

Once this was fixed, there was a remaining problem with
client_cert_verified acl, haproxy acl are OR, but this rule was supposed
to be a AND (client present a certificate AND it is verified), this was
rewritten to use inline condition which are AND.

[1]: https://cbonte.github.io/haproxy-dconv/1.8/configuration.html#5.1-verify

Also adjust test_x_forwarded_for_stripped_when_no_certificate to assert
that there is no X-Forwarded-For header at all when no client
certificate.

See merge request nexedi/slapos!1433

  https://archive.fosdem.org/2023/schedule/event/innodb_change_buffer/
  https://jira.mariadb.org/browse/MDEV-27734
  https://mariadb.com/kb/en/innodb-change-buffering/

If a server reboots brutally, there will be a legacy httpd pid file stay
in the file system, and the OS may reuse the PID that remains in that
file, which will cause httpd service to fail to start even if there is
no such httpd service is running.

This commit implemented the following logic:
0. If there is no existing PID file, then run the httpd service directly
1. If the PID file exists and contains the PID used by another process,
we delete the PID file, then restart the httpd service
2. If the PID file exists and contains the PID used by another running
httpd service, we allow the httpd report the "already running" error
normally

This is the last version supporting python 2.7.

See merge request nexedi/slapos!1428