Thus we have more possibilities in filtering than just excluding
a full directory.

It can exclude directories from the hash computation, which may
be handy for more than one software release.
Its first application will be to exclude logrotated logs from the signature
file, and the zodb backup, as they integrate their own integrity mechanisme

To provide uselful functions that can be reused in other scripts
used in the context of webrunner's resiliency

Rewrite the [runner-exporter bash script](https://lab.nexedi.com/nexedi/slapos/blob/2a01586bfe17f46e2ef26ec6e45bef692435ed28/software/slaprunner/template/runner-export.sh.jinja2) in python.

Also parametarize the signature function to not use only sha256sum which can create issues on some systems. This should be a separated commit, but that shouldn't be a blocker.

Write unit tests for the main functions of this exporter script. As every trick in the original bash script were not obvious / understood / found by me, I may have missed some.

/reviewed-on nexedi/slapos.toolbox!44

Starting from Debian 9, ping binary can understand IPv6 addresses and
since we have more and more testnode on Debian9, test the other way
around. Maybe we should get rid of this test...

assertEquals is deprecated, see https://docs.python.org/2.7/library/unittest.html#deprecated-aliases

This is the promise that will be used for the backupserver software release. It is working as is (with some needed changes in backupserver software release of course) but I may add more features.

I'd like to have a first review of my code.

/reviewed-on nexedi/slapos.toolbox!43

This promise will replace the old check_port_listening promise (https://lab.nexedi.com/nexedi/slapos/blob/master/slapos/recipe/check_port_listening.py). See https://www.erp5.com/slapos-TechnicalNote.General.SlapOS.Monitoring.Specifications for how to instantiate this promise in buildout.

/reviewed-on nexedi/slapos.toolbox!42

Since slapos!a222ccfabc10103c6ecb9195de4e0c0250705d11, haproxy process name is variable,
so let's restart the full partition instead of guessing the name of the process.

/reviewed-on nexedi/slapos.toolbox!41

/reviewed-on nexedi/slapos.toolbox!40

/reviewed-on nexedi/slapos.toolbox!39

/reviewed-on nexedi/slapos.toolbox!38

/reviewed-on nexedi/slapos.toolbox!37

This solve indefinite loop in while condition on promise check_free_disk_space

check_free_disk_space is now in promise.plugin, with some improvement.
Database connection timeout is now 10 seconds
If database is locked, promise will try again 10 times before start to raise error with "database is locked" message.

When the URL /isSRReady is called if the software release link is not present in softwareLink folder, webrunner will launch build of software. The result is slapgrid trying to build sr every minutes, then resiliency fail.