Closed n-rodriguez closed 4 years ago
@myii ping
@n-rodriguez I'll check this soon, thanks for the reminder.
@n-rodriguez Can you check all of the Travis failures? From the first couple, there's a common error:
local:
Data failed to compile:
----------
Rendering SLS 'base:influxdb.package.install' failed: Jinja variable 'dict object' has no attribute 'toml_pkg'
@myii it seems that influxdb is not available on this OpenSuse version (leap15) : https://software.opensuse.org/package/influxdb
default-centos-6-2017-7-py2
doesn't have any python-pytoml
package
default-amazonlinux-2-2019-2-py2
doesn't have any python-pytoml
package
@n-rodriguez Like last time, do you want to enable all of the instances, to see which ones work?
@n-rodriguez Like last time, do you want to enable all of the instances, to see which ones work?
done : https://travis-ci.com/saltstack-formulas/influxdb-formula/builds/126296726
@n-rodriguez The opensuse
instance was working prior to this (specifically default-opensuse-leap-15-2018-3-py2
). Can we get that working again?
@myii ping
@n-rodriguez OK, to get an idea of what this PR is doing, I've taken advantage of the new debug
mode to compare runs:
In terms of TOFS specifically, there wasn't much to check. In fact, before and after aren't even the same states (when searching for source:
):
https://travis-ci.org/myii/influxdb-formula/jobs/590033880#L1099-L1102 (before):
influxdb_logrotate:
file.managed:
- name: /etc/logrotate.d/influxdb
- source: salt://influxdb/files/logrotate.conf.jinja
https://travis-ci.org/myii/influxdb-formula/jobs/590038193#L991-L997 (after):
influxdb/config/install:
file.managed:
- name: /etc/influxdb/influxdb.conf
- source:
- salt://influxdb/files/1d1421e8f779/influxdb.tmpl.jinja
- salt://influxdb/files/Debian/influxdb.tmpl.jinja
- salt://influxdb/files/default/influxdb.tmpl.jinja
Just checked right now using the following search:
So it looks like I'm going to have to do both runs again, with more states defined in kitchen.yml
. Let me get back with that feedback, when I get a chance to do that.
@n-rodriguez So influxdb.cli
has been removed? Then this PR involves a breaking change (which isn't an issue but) that needs to be added to the commit message.
@n-rodriguez OK, I only needed to modify the run done before TOFS, to get the necessary comparisons:
This breakdown is part of the familiarisation, so if I ask questions, please don't take it the wrong way.
https://travis-ci.org/myii/influxdb-formula/jobs/590048079#L1047-L1050:
influxdb_config:
file.managed:
- name: /etc/influxdb/config.toml
- source: salt://influxdb/files/influxdb.config.jinja
https://travis-ci.org/myii/influxdb-formula/jobs/590048079#L1065-L1068:
influxdb_default:
file.managed:
- name: /etc/default/influxdb
- source: salt://influxdb/files/influxdb.etc_default
https://travis-ci.org/myii/influxdb-formula/jobs/590048079#L1162-L1165:
influxdb_logrotate:
file.managed:
- name: /etc/logrotate.d/influxdb
- source: salt://influxdb/files/logrotate.conf.jinja
If anything else is being removed by this PR, it will also need to be added as a breaking change. Multiple breaking changes can be supplied in the commit body, like has been done here:
@myii @n-rodriguez Why did you closed this ??
Initial feedback, I haven't looked at this in any detail at all. But in terms of
yamllint
, this will cause an error with Jinja tags in the YAML files. Ideally, we need to resolve this by pre or post-processing inmap.jinja
(or similar). Worst case, we can ignore the YAML files containing the Jinja, but that then loses the benefit of havingyamllint
.