Khéops 101

cd /home/jez/volumes/data/prj/bell/training/tiger-ansible/ext/kheops
echo $PWD
export KHEOPS_NAMESPACE=ex1_enc 
export KHEOPS_CONFIG=examples/kheops.yml
rm -rf "examples/$KHEOPS_NAMESPACE"
/home/jez/volumes/data/prj/bell/training/tiger-ansible/ext/kheops

Command line

Let’s check first that kheops works correclty, and start to read the manual.

kheops --help
usage: kheops [-h] [-v] [-c CONFIG]
              {schema,gen_doc,lookup,demo,subcommand2} ...

Kheops, hierarchical data lookup tool

options:
  -h, --help            show this help message and exit
  -v, --verbose         Increase verbosity (KHEOPS_VERBOSE) (default: 0)
  -c CONFIG, --config CONFIG
                        Kheops configuration file (KHEOPS_CONFIG) (default:
                        examples/kheops.yml)

subcommands:
  valid subcommands

  {schema,gen_doc,lookup,demo,subcommand2}

So we have a working kheops command, and we will focus on the lookup command. On it’s simplest form, a lookup consists in querying a key for a given scope. The output of the key will change depending the scope value. A key is in simple word.

Defining a hierarchy

To illustrate how Khéops works, let’s start with a simple example, we will try to lookup the profile key of the following two (fictive) servers:

  • web.domain.org: which act as a webserver role

  • mysql.domain.org: which act as mysql role

But first we need to create our hierarchy. It’s as simple as creating directories and put some json or yaml data into different files. Let’s create our hierarchy. We will first create the default profile:

From our use case, we will build a lookup tree. We want to be able to organise data depending the 3 criterias:

  • node: name of the node

  • role: assigned role to the node

  • environment: it can either be dev or prod

Let’s create our lookup hierarchy:

default: lookups: - path: default # Simplest form, just a path - path: “roles/{role}” # If list, it’s auto expanded like in bash - path: “env_{env}” # If list, it’s auto expanded like in bash - path: “nodes/{node}”

So for a given key, Khéops will iterate all over those paths to find the requested key , and then it will merge all results. Some paths are variabilized, and those variable comes from the scope. The scope come along the key, it’s can be any metadata. For complex metadata you may want to store those in a file and load your scope with the -f <yaml_scope_file> option:

kheops lookup -e <var1=val1> -e <var2=val2> <key>

A scope is completely optional while keys are required.


Basic hierarchy

# We create a fresh hierachy
mkdir -p examples/$KHEOPS_NAMESPACE

# We create a profile key, which is a dict
cat > examples/$KHEOPS_NAMESPACE/default.yml <<EOF
---
profile:
  env: "NO_ENV"
  product: "NO_PRODUCT"
  
EOF

# Let's inspect our hierarchy
tree examples/$KHEOPS_NAMESPACE
cat examples/$KHEOPS_NAMESPACE/default.yml
examples/ex1_enc
└── default.yml

0 directories, 1 file
---
profile:
  env: "NO_ENV"
  product: "NO_PRODUCT"

From this point, we defined our profile with two attribute, team and product. As it’s the default case, we set them both unconfigured.

You are now already able to query your hierarchy:

kheops lookup profile
env: NO_ENV
product: NO_PRODUCT

Good, no surprise. But, we mentionned we wanted to get the profile of two instances, this how would do that:

kheops lookup -e node=web.infra.net profile
env: NO_ENV
product: NO_PRODUCT

Roles

However, same result as before, which is expected as we did not finished to configure our hierarchy. Among our instances, we identified 2 roles: web and mysql. Let’s create those two roles:

mkdir -p examples/$KHEOPS_NAMESPACE/roles

# We create a new web role
cat > examples/$KHEOPS_NAMESPACE/roles/web.yml <<EOF
---
profile:
  product: "httpd_server"

  web_top_domain: ""
  web_app: "NO_APP"
  web_port: 80
  web_user_list:
    - sysadmins
  
EOF

# We create a new mysql role
cat > examples/$KHEOPS_NAMESPACE/roles/mysql.yml <<EOF
---
profile:
  product: "mysql_server"

  mysql_database: "NO_DATABASE"
  mysql_users:
    - "sysadmin@10.0.42%"
  mysql_port: 3306
  mysql_cluster: False
  
EOF

# Let's inspect our hierarchy
tree examples/$KHEOPS_NAMESPACE
examples/ex1_enc
├── default.yml
└── roles
    ├── mysql.yml
    └── web.yml

1 directory, 3 files
tail -n 999 examples/$KHEOPS_NAMESPACE/{*.yml,*/*.yml}
==> examples/ex1_enc/default.yml <==
---
profile:
  env: "NO_ENV"
  product: "NO_PRODUCT"
  

==> examples/ex1_enc/roles/mysql.yml <==
---
profile:
  product: "mysql_server"

  mysql_database: "NO_DATABASE"
  mysql_users:
    - "sysadmin@10.0.42%"
  mysql_port: 3306
  mysql_cluster: False
  

==> examples/ex1_enc/roles/web.yml <==
---
profile:
  product: "httpd_server"

  web_top_domain: ""
  web_app: "NO_APP"
  web_port: 80
  web_user_list:
    - sysadmins
kheops lookup -e node=web.infra.net -e role=web profile
kheops lookup -e node=mysql.infra.net -e role=mysql profile
env: NO_ENV
product: httpd_server
web_top_domain: ''
web_app: NO_APP
web_port: 80
web_user_list:
- sysadmins

env: NO_ENV
product: mysql_server
mysql_database: NO_DATABASE
mysql_users:
- sysadmin@10.0.42%
mysql_port: 3306
mysql_cluster: false

Per node override

It’s getting better, we can see that the profile key has been merged with the key values, across the different locations.

However, we will have those placeholders, and we want to have personalized value, depending if it’s aweb server, it need an unique domain and some unique parameters. So let’s create a nodes directory and place some data inside.

mkdir -p examples/$KHEOPS_NAMESPACE/nodes

# We create a new web role
cat > examples/$KHEOPS_NAMESPACE/nodes/web.infra.net.yml <<EOF
---
profile:
  web_app: 'myapp'
  web_user_list:
    - domain_org
    - domain_org_external
  
EOF

# We create a new mysql role
cat > examples/$KHEOPS_NAMESPACE/nodes/mysql.infra.net.yml <<EOF
---
profile:  
  mysql_database: "app_domain_org"
  mysql_users:
    - "app_domain_org@10.0.51%"
  
EOF

# Let's inspect our hierarchy
tree examples/$KHEOPS_NAMESPACE
examples/ex1_enc
├── default.yml
├── nodes
│   ├── mysql.infra.net.yml
│   └── web.infra.net.yml
└── roles
    ├── mysql.yml
    └── web.yml

2 directories, 5 files

And we try again:

kheops lookup -e node=web.infra.net -e role=web profile
kheops lookup -e node=mysql.infra.net -e role=mysql profile
env: NO_ENV
product: httpd_server
web_top_domain: ''
web_app: myapp
web_port: 80
web_user_list:
- sysadmins
- domain_org
- domain_org_external

env: NO_ENV
product: mysql_server
mysql_database: app_domain_org
mysql_users:
- sysadmin@10.0.42%
- app_domain_org@10.0.51%
mysql_port: 3306
mysql_cluster: false

Environment override

Let’s say you want to support environment, it’s the same:

# We create a new dev environment
cat > examples/$KHEOPS_NAMESPACE/env_dev.yml <<EOF
---
profile:
  env: dev
  
  # We change the top domain for dev environment, and reduce the cache
  web_top_domain: dev.infra.net
  web_cache: 1m
  
  # We want a debug users
  web_user_list:
    - debug_user
  mysql_users:
    - debug@10.0.%

  debug: true

EOF

# We create a new mysql role
cat > examples/$KHEOPS_NAMESPACE/env_prod.yml <<EOF
---
profile:
  env: prod
  
  # On production environment, we always want to use public faced domain and 12 hour cache.
  web_top_domain: infra.com
  web_cache: 12h
  
EOF

# Let's inspect our hierarchy
tree examples/$KHEOPS_NAMESPACE
examples/ex1_enc
├── default.yml
├── env_dev.yml
├── env_prod.yml
├── nodes
│   ├── mysql.infra.net.yml
│   └── web.infra.net.yml
└── roles
    ├── mysql.yml
    └── web.yml

2 directories, 7 files

So it’s become quite easy to compare the difference between environment, with a simple variable switch:

kheops lookup -e node=web.infra.net -e role=web -e env=prod profile
kheops lookup -e node=web.infra.net -e role=web -e env=dev profile
env: prod
product: httpd_server
web_top_domain: infra.com
web_app: myapp
web_port: 80
web_user_list:
- sysadmins
- domain_org
- domain_org_external
web_cache: 12h

env: dev
product: httpd_server
web_top_domain: dev.infra.net
web_app: myapp
web_port: 80
web_user_list:
- sysadmins
- debug_user
- domain_org
- domain_org_external
web_cache: 1m
mysql_users:
- debug@10.0.%
debug: true

Same for mysql:

kheops lookup -e node=mysql.infra.net -e role=mysql -e env=prod profile
kheops lookup -e node=mysql.infra.net -e role=mysql -e env=dev profile
env: prod
product: mysql_server
mysql_database: app_domain_org
mysql_users:
- sysadmin@10.0.42%
- app_domain_org@10.0.51%
mysql_port: 3306
mysql_cluster: false
web_top_domain: infra.com
web_cache: 12h

env: dev
product: mysql_server
mysql_database: app_domain_org
mysql_users:
- sysadmin@10.0.42%
- debug@10.0.%
- app_domain_org@10.0.51%
mysql_port: 3306
mysql_cluster: false
web_top_domain: dev.infra.net
web_cache: 1m
web_user_list:
- debug_user
debug: true

You have to keep in mind you can query the key with a different scope, and get different views:

kheops lookup profile

echo "==> Per environment view"
kheops lookup -e env=prod profile
kheops lookup -e env=dev profile

echo "==> Per role and  environment view"
kheops lookup -e role=mysql -e env=prod profile
kheops lookup -e role=web -e env=prod profile

echo "==> Per node view"
kheops lookup -e node=web.infra.net -e role=web -e env=dev profile
env: NO_ENV
product: NO_PRODUCT

==> Per environment view
env: prod
product: NO_PRODUCT
web_top_domain: infra.com
web_cache: 12h

env: dev
product: NO_PRODUCT
web_top_domain: dev.infra.net
web_cache: 1m
web_user_list:
- debug_user
mysql_users:
- debug@10.0.%
debug: true

==> Per role and  environment view
env: prod
product: mysql_server
mysql_database: NO_DATABASE
mysql_users:
- sysadmin@10.0.42%
mysql_port: 3306
mysql_cluster: false
web_top_domain: infra.com
web_cache: 12h

env: prod
product: httpd_server
web_top_domain: infra.com
web_app: NO_APP
web_port: 80
web_user_list:
- sysadmins
web_cache: 12h

==> Per node view
env: dev
product: httpd_server
web_top_domain: dev.infra.net
web_app: myapp
web_port: 80
web_user_list:
- sysadmins
- debug_user
- domain_org
- domain_org_external
web_cache: 1m
mysql_users:
- debug@10.0.%
debug: true

Even if somwaht clunky, this method can help to troubleshoot wrong data by dichotomy.


Tooling and applications






Troubleshooting

Sometimes, it can may be hard to navigate across file and hierachy, but GNU Utils are here to help. There is a selection of small tips:

set -x

: Find where a key has been defined
grep -r '^profile:' examples/$KHEOPS_NAMESPACE

: Find where a key has been defined and 5 first lines
grep -r -A 5 'web_user_list:' examples/$KHEOPS_NAMESPACE

: Search from anything related to database
grep -R -C 3  'database' examples/$KHEOPS_NAMESPACE

set +x
+ : Find where a key has been defined
+ grep --colour=auto -r '^profile:' examples/ex1_enc
examples/ex1_enc/env_prod.yml:profile:
examples/ex1_enc/roles/mysql.yml:profile:
examples/ex1_enc/roles/web.yml:profile:
examples/ex1_enc/nodes/mysql.infra.net.yml:profile:  
examples/ex1_enc/nodes/web.infra.net.yml:profile:
examples/ex1_enc/default.yml:profile:
examples/ex1_enc/env_dev.yml:profile:
+ : Find where a key has been defined and 5 first lines
+ grep --colour=auto -r -A 5 web_user_list: examples/ex1_enc
examples/ex1_enc/roles/web.yml:  web_user_list:
examples/ex1_enc/roles/web.yml-    - sysadmins
examples/ex1_enc/roles/web.yml-  
--
examples/ex1_enc/nodes/web.infra.net.yml:  web_user_list:
examples/ex1_enc/nodes/web.infra.net.yml-    - domain_org
examples/ex1_enc/nodes/web.infra.net.yml-    - domain_org_external
examples/ex1_enc/nodes/web.infra.net.yml-  
--
examples/ex1_enc/env_dev.yml:  web_user_list:
examples/ex1_enc/env_dev.yml-    - debug_user
examples/ex1_enc/env_dev.yml-  mysql_users:
examples/ex1_enc/env_dev.yml-    - debug@10.0.%
examples/ex1_enc/env_dev.yml-
examples/ex1_enc/env_dev.yml-  debug: true
+ : Search from anything related to database
+ grep --colour=auto -R -C 3 database examples/ex1_enc
examples/ex1_enc/roles/mysql.yml-profile:
examples/ex1_enc/roles/mysql.yml-  product: "mysql_server"
examples/ex1_enc/roles/mysql.yml-
examples/ex1_enc/roles/mysql.yml:  mysql_database: "NO_DATABASE"
examples/ex1_enc/roles/mysql.yml-  mysql_users:
examples/ex1_enc/roles/mysql.yml-    - "sysadmin@10.0.42%"
examples/ex1_enc/roles/mysql.yml-  mysql_port: 3306
--
examples/ex1_enc/nodes/mysql.infra.net.yml----
examples/ex1_enc/nodes/mysql.infra.net.yml-profile:  
examples/ex1_enc/nodes/mysql.infra.net.yml:  mysql_database: "app_domain_org"
examples/ex1_enc/nodes/mysql.infra.net.yml-  mysql_users:
examples/ex1_enc/nodes/mysql.infra.net.yml-    - "app_domain_org@10.0.51%"
examples/ex1_enc/nodes/mysql.infra.net.yml-  
+ set +x

The tail/head command is quite usefull to look at multiple files at the same time, it add a nice header for each file:

head -n 999 examples/ex1_enc/roles/*
==> examples/ex1_enc/roles/mysql.yml <==
---
profile:
  product: "mysql_server"

  mysql_database: "NO_DATABASE"
  mysql_users:
    - "sysadmin@10.0.42%"
  mysql_port: 3306
  mysql_cluster: False
  

==> examples/ex1_enc/roles/web.yml <==
---
profile:
  product: "httpd_server"

  web_top_domain: ""
  web_app: "NO_APP"
  web_port: 80
  web_user_list:
    - sysadmins

You can also have a view of all files with this command:

find . -type f| xargs head -n 999 | less

From there, you will be able to have a nice overview of your data.

You can even diff your change with this command. There is this simple trick to compare the data difference between 2 lookups:

diff -u \
<(kheops lookup -e node=web.infra.net -e role=web -e env=prod profile) \
<(kheops lookup -e node=web.infra.net -e role=web -e env=dev profile)
--- /dev/fd/63	2022-02-01 20:10:53.094525316 -0500
+++ /dev/fd/62	2022-02-01 20:10:53.094525316 -0500
@@ -1,11 +1,15 @@
-env: prod
+env: dev
 product: httpd_server
-web_top_domain: infra.com
+web_top_domain: dev.infra.net
 web_app: myapp
 web_port: 80
 web_user_list:
 - sysadmins
+- debug_user
 - domain_org
 - domain_org_external
-web_cache: 12h
+web_cache: 1m
+mysql_users:
+- debug@10.0.%
+debug: true

You can also ask Kheops to explain you how he built the result, you can use the -x flag:

kheops lookup -e role=web profile -x
    INFO: Explain lookups:
+------------------------+------------------------------+
| Config                 | Runtime                      |
+------------------------+------------------------------+
|                        |                              |
| Config:{               | Runtime:{                    |
|   "path": "default",   |   "scope": {                 |
|   "backend": "file",   |     "role": "web"            |
|   "continue": true     |   },                         |
| }                      |   "key": "profile",          |
|                        |   "conf": {                  |
|                        |     "index": 0               |
|                        |   },                         |
|                        |   "raw_path": "default"      |
|                        | }                            |
|                        |                              |
| Config:{               | Runtime:{                    |
|   "path": "roles/web", |   "scope": {                 |
|   "backend": "file",   |     "role": "web"            |
|   "continue": true     |   },                         |
| }                      |   "key": "profile",          |
|                        |   "conf": {                  |
|                        |     "index": 1               |
|                        |   },                         |
|                        |   "raw_path": "roles/{role}" |
|                        | }                            |
+------------------------+------------------------------+
    INFO: Explain candidates:
+----------------------------------------------------------------------------------+-------------------------------+------------------------------+
| Status                                                                           | Runtime                       | Data                         |
+----------------------------------------------------------------------------------+-------------------------------+------------------------------+
|                                                                                  |                               |                              |
| Status:{                                                                         | Runtime:{                     | Data:{                       |
|   "path": "/home/jez/volumes/data/prj/bell/training/tiger-ansible/ext/kheops ... |   "scope": {                  |   "env": "NO_ENV",           |
|   "status": "found",                                                             |     "role": "web"             |   "product": "NO_PRODUCT"    |
|   "rel_path": "examples/ex1_enc/default.yml"                                     |   },                          | }                            |
| }                                                                                |   "key": "profile",           |                              |
|                                                                                  |   "conf": {                   |                              |
|                                                                                  |     "index": 0                |                              |
|                                                                                  |   },                          |                              |
|                                                                                  |   "raw_path": "default",      |                              |
|                                                                                  |   "backend_index": 0          |                              |
|                                                                                  | }                             |                              |
|                                                                                  |                               |                              |
| Status:{                                                                         | Runtime:{                     | Data:{                       |
|   "path": "/home/jez/volumes/data/prj/bell/training/tiger-ansible/ext/kheops ... |   "scope": {                  |   "product": "httpd_server", |
|   "status": "found",                                                             |     "role": "web"             |   "web_top_domain": "",      |
|   "rel_path": "examples/ex1_enc/roles/web.yml"                                   |   },                          |   "web_app": "NO_APP",       |
| }                                                                                |   "key": "profile",           |   "web_port": 80,            |
|                                                                                  |   "conf": {                   |   "web_user_list": [         |
|                                                                                  |     "index": 1                |     "sysadmins"              |
|                                                                                  |   },                          |   ]                          |
|                                                                                  |   "raw_path": "roles/{role}", | }                            |
|                                                                                  |   "backend_index": 1          |                              |
|                                                                                  | }                             |                              |
+----------------------------------------------------------------------------------+-------------------------------+------------------------------+
env: NO_ENV
product: httpd_server
web_top_domain: ''
web_app: NO_APP
web_port: 80
web_user_list:
- sysadmins