msgpack as it is a new dependency since the introduction of cachecontrol

 There are cases when the db isn't upgrade but the webrunner is,
 that the db migration may occur during the import, cause error for a certain period of time (until db inside runner0 is updated).

Due to missing monitor frontend in tests this change is not covered by tests.

/reviewed-on nexedi/slapos!592

As caddy_configuration_state provides information with exit code use it
directly in the if instead of running it with $(), which stops execution when
configuration is not changed.

Improve checking for too old result - do not remove the file, instead check
for too old result and then recalculate.

/reviewed-on nexedi/slapos!589

As Kedifa speaks HTTP check that it can reply with HTTP, not that just
it is possible to connect to.

/reviewed-on nexedi/slapos!591

/reviewed-on nexedi/slapos!590

bedba966

We need to sleep to give chance for Caddy to pick up the signal.

This SR has been developed for a demo. The purpose is to have a running pureftpd server which will call a script "/opt/pureftpd/upload_script" each time a file is uploaded. Pureftpd is used because it is already used in the current working environment of our partner and they don't want to change for now.

The script /opt/pureftpd/upload_script" is left out of the SR on purpose because it will be written by people outside Nexedi at first and contains sensitive information. Also, it is on purpose in /opt directory of the machine because this SR will be deployed on only one machine for now and the script will be shared by all pureftpd instances.

/reviewed-on nexedi/slapos!503

In some cases domain can come from "outside" of the profile, and be filled
with "garbage", so if custom_domain is set, do not overwrite it.

/reviewed-on nexedi/slapos!588

…ration state calculation""

The states has been correctly propagated and used. Original author had no
idea what he was doing, possibly because of some incorrect assumptions while
developing this fix.

/reviewed-on !587

Tests passing, merging right away.

/reviewed-on nexedi/slapos!586

in some webrunner, there are 2 keys in equeue.db:

```
>>> print(db.keys())
['/srv/slapgrid/slappart18/bin//runner-importer', '/srv/slapgrid/slappart18/bin/runner-importer']
```

We suppose that the 2nd key was used after the first key so we take the
timestamp of the last key.

/reviewed-on nexedi/slapos!578

otherwise it will stall VM execution.

This is observed inside qemu VM started by 

https://lab.nexedi.com/nexedi/slapos/blob/master/software/erp5testnode/testsuite/deploy-test/software.cfg

In qemu log one can see:
===
QEMU waiting for connection on: disconnected:unix:/srv/slapgrid/slappart6/var/qmp_socket,server

===

Affected version:

root@z600:/srv/slapgrid/slappart6# /opt/slapgrid/7903668a197f3cf8f94375d0739a21e8/parts/kvm/bin/qemu-system-x86_64 --version
QEMU emulator version 2.8.0
Copyright (c) 2003-2016 Fabrice Bellard and the QEMU Project developer

/reviewed-on nexedi/slapos!585

This parameter was documented but not implemented in the software release,
is not a feature provided by caucase, and is not a slapos-compliant
feature. It is now a place-holder in parameter schema.

The gitclone reciped is used for babeld to get submodules added in
the latest babeld version.
gcc is forced to use `-std=gnu99` because default std has changed
in version 5.1 and we get errors when compiling babeld with older gcc.

The last state value has been reverted by mistake, put back correct one.

This also means that caddy source is fetched directly from upstream, as all
required fixes has been incorporated into the upstream.

Drop direct usage of gowork for now, in order to have caddy built using go
module, support for gowork with go modules might come later.

Follow new way of certificate managament in Caddy 1 as noted
https://github.com/mholt/caddy/issues/2588#issuecomment-505367152

It's released, let's use the newest version.

Only the webrunner should have a monitor port different from the default,
to avoid most conflicts between the webrunner and any instance built inside.

-> webrunner: 8386
-> other: 8196

The previous value, '/', caused problems: some generated urls had two
consecutive slashes, which can be misinterpreted by the waitress server.

Commit b96cdf9c didn't propagate
correctly the monitor password from the root partition to subpartitions.

This will be required for WSGI. This depends on ERP5 > 2018-04-26
(commit 6d74ba22a962d08624ef35342708f333a21ceff5)

The scripts are already provided in caucase-eggs part, there is no reason to
duplicate them here, so drop this entry.

/reviewed-on nexedi/slapos!579

Frontend operator shall have easy access to information about rejected
slaves, possibly the best in the JSON file.

Also the keys for the human readable information are slave's titles, not
references.

The information is published via hand crafted HTTPS endpoint.

Note: The SSL certificate is generated manually. Existing caucase is special
      for KeDiFa, this is another step to move all generated certificates (or
      otherwise self-signed) to internal, full automatic caucase.