Age | Commit message (Collapse) | Author |
|
subrepo:
subdir: "puppet/modules/openvpn"
merged: "26d4edc"
upstream:
origin: "https://leap.se/git/puppet_openvpn"
branch: "master"
commit: "26d4edc"
git-subrepo:
version: "0.3.0"
origin: "https://github.com/ingydotnet/git-subrepo"
commit: "1e79595"
Change-Id: I596766ccfb806b3ca2d1c755c4e24c5ad3d997f9
|
|
Change-Id: Ia7655153b556337f676e3d909559c4a7306bedd6
|
|
Configure DNSSEC validation for client verification, giving us a
stronger form of opportunistic TLS
Change-Id: Iab92d4f593c4a5a44e3b694295096b0d7f687a37
|
|
Without this set, dnssec will fail validation for internal domains,
which should not be validated
Change-Id: I8589332598fe97ad5218dd23825ac77af2d8def6
|
|
Change-Id: Ibdf39a721162b4a5663ef27c27b2db0261c6e8a5
|
|
In an attempt to resolve #8021, a template error was made, causing
duplicated entries to appear in the rsyslog template.
Change-Id: Ic41d6ef9aec9865cf64312c1eb96e408b39d441c
|
|
|
|
Change-Id: Icaab817870d005b7a854a3fb8c402705d0b2d77f
|
|
Change-Id: I0c6e27298c63bd37de1410985d054799818c22a4
|
|
When tor is not configured, then its possible to get this error on
deploy:
Error: tor is not a hash or array when accessing it with hidden_service
at /srv/leap/puppet/modules/site_static/manifests/init.pp:16 on node
rewdevstatic1.rewire.org
This commit only accesses the array when its enabled.
Change-Id: Ia75ac7a51179da980966adba0cc614b9cd642b0c
|
|
When tor hidden services were enabled for static sites, only a very
basic configuration was setup and it didn't take into account the
different location configurations that can be configured for a
static site.
This commit resolves that by making a site_static::hidden_service class
similar to the site_webapp::hidden_service class, and fixes up the
apache vhost template to properly create the location blocks for the
hidden service vhost.
Change-Id: Ice3586f4173bd2d1bd3defca29d21c7403d5a03a
|
|
We were creating the hidden service name without a newline, and then tor
would be restarted and change the hidden service hostname file to have a
newline, which would then require that the next deploy would change that
file to not have a newline again.
This fixes that problem by making the hostname have a newline so it
matches what tor wants.
Change-Id: I38f450684d557cf943ec94f2f8e19cda3aefdf66
|
|
Change-Id: I3d733b6645c804a5fb337ad4b8edc59a66ad50b5
|
|
Change-Id: Icaab817870d005b7a854a3fb8c402705d0b2d77f
|
|
|
|
|
|
Change-Id: Iab9597f5f0336f66df9b73fea9d79c789cbb8302
|
|
The Trace method is enabled because of the Apache module, but it is not the
default in Debian, and it should not be enabled, for more information see the
following:
https://www.kb.cert.org/vuls/id/867593
Change-Id: I06a06ae679dbf7049f26a017125b61e5e38f6268
|
|
The onlyif check was incorrectly specified in the original implementation in
commit id: 15b83d88dcedab496a19cef57f11c5c8e091dd4a this inverts it so it
is properly detected.
Change-Id: I531e206fff1ca61780adcd195e1f917011e50fb4
|
|
After including everything into a `node default` scope
in puppet/manifests/site.pp to make puppet-catalog-test happy
(see commit 62ea45d47), we get this error:
Error: member(): Requires array to work with at
/srv/leap/puppet/modules/site_obfsproxy/manifests/init.pp:14
Moving the `services` hiera avaluation out of the node scope back
to top level scope will solve this.
|
|
Change-Id: Ic12b243b195e40482a70dd70219212c3697899ba
|
|
Change-Id: I772c3b6e489e3c1848c45c6bcaa240324fc88928
|
|
|
|
Change-Id: I7675dbaba4d896a62dab9fcf4817092ea69f1298
|
|
|
|
|
|
It turns out that in some corner-cases, the script is not called:
(1) start the deploy, create files in /var/lib/puppet/stunnel4/config
(2) halt puppet before apply finishes
(3) re-run deploy
in this scenario, next time you run deploy, refresh_stunnel will never
get called to populate /etc/stunnel, because the files in
/var/lib/puppet/stunnel4/config haven't changed.
This problem can be really confusing when it happens.
To fix this, we just run refresh_stunnel every, it is pretty fast and
the script has more complete logic for what to do than puppet, which has
only an asymmetrical view on the situation.
Change-Id: I9e5fad1d081c2fe07f3ac8f07cfb87d86b88f7c9
|
|
|
|
|
|
Fix opendkim milter location (#8163).
The unix socket method for connecting to the milter was incorrectly
reverted, this puts it back to how it should be.
Change-Id: Ifde669c920a249c782f577a112f4d45e60a889a2
See merge request !4
|
|
if this is set in the config, the deamons do not
start anymore. From the debian changelog:
clamav (0.99.2+dfsg-0+deb8u1) stable; urgency=medium
* Import new Upstream.
* Drop AllowSupplementaryGroups option which is default now
(Closes: #822444).
|
|
|
|
The unix socket method for connecting to the milter was incorrectly
reverted, this puts it back to how it should be.
Change-Id: Ifde669c920a249c782f577a112f4d45e60a889a2
|
|
|
|
|
|
Disable puppet-agent daemon from running.
The agent wakes up every two minutes and tries to connect to the default
server, failing with a certificate warning. We don't use the agent, so
we can safely disable it (#8032)
Change-Id: I707f42b59205993325431aba283552b1b73a0ad1
See merge request !1
|
|
check_mk operations can take a long time (such as when doing a
re-inventory using "check_mk -II") when multiple hosts are down. This
decreases the connect timeout to 5 seconds.
Change-Id: I1eac5f14bad2afc2ffc4cbf8c950c24b052a0d6e
|
|
The agent wakes up every two minutes and tries to connect to the default
server, failing with a certificate warning. We don't use the agent, so
we can safely disable it (#8032)
Change-Id: I707f42b59205993325431aba283552b1b73a0ad1
|
|
Automatic background couchdb db compaction frees a huge
amount of diskspace.
- Resolves: #8118
|
|
|
|
|
|
Sometimes a floating point exception or segfault of
a process results in systemd restarting it, we want
to recognize this from the syslog
i.e.:
systemd[1]: pixelated-server.service: main process exited,
code=killed, status=8/FPE
systemd[1]: Unit pixelated-server.service entered failed state.
- Related: https://github.com/pixelated/pixelated-user-agent/issues/683
|
|
|
|
Otherwise, the nagios config will get regenerated and nagios gets
reloaded before all checks are registered by a check_mk inventory.
- Related: #6873
|
|
After upgrading the platform, there might be old check_mk checks
registered on the monitor hosts. We now run a check_mk inventory
on every run that also purged old non-existng checks.
- Resolves: #6873
|
|
|
|
Change-Id: I20a28ae77c98071aefc1933e0ea73e5f3b895acb
|
|
Shorewall in jessie doesn't come with a proper unit file, and
as a result, it doesn't properly start with systemd.
To solve this, we provide the systemd unit file that comes with stretch,
add a systemd submodule that provides the exec resources needed for when
systemd units or configuration files are changed
Change-Id: I861fa951835928b4741abfbf969adcee4b8f147b
|
|
|
|
- ignore puppet lint error about inheriting from different namespace
|