[Feature] diagnostic library for Autoware design.

[hakuturu583]

Design

Now, there is no diagnostic library for Autoware.
It is very dangerous.
This issue is related to #341

diagnostic lib contains diag_manager class.
It provides various kinds of functions for diagnostics.

Now, log file format is like below.

[2018-09-13T03:25:25.340543] : in /ndt_matching: topic /ndt_matching/data subscribe rate was low (Warn)
[2018-09-13T03:25:25.341312] : in /ndt_matching: The input value hogehoge is out of range.
[2018-09-13T03:25:25.441295] : in /ndt_matching: The input value hogehoge is out of range.
[2018-09-13T03:25:25.541326] : in /ndt_matching: The input value hogehoge is out of range.
[2018-09-13T03:25:25.641427] : in /ndt_matching: The input value hogehoge is out of range.
[2018-09-13T03:25:25.741318] : in /ndt_matching: The input value hogehoge is out of range.
[2018-09-13T03:25:25.841311] : in /ndt_matching: The input value hogehoge is out of range.
[2018-09-13T03:25:25.941436] : in /ndt_matching: The input value hogehoge is out of range.
[2018-09-13T03:25:26.041322] : in /ndt_matching: The input value hogehoge is out of range.
[2018-09-13T03:25:26.141353] : in /ndt_matching: The input value hogehoge is out of range.
[2018-09-13T03:25:26.340464] : in /ndt_matching: topic /ndt_matching/data subscribe rate was low (Warn)
[2018-09-13T03:25:27.340491] : in /ndt_matching: topic /ndt_matching/data subscribe rate was low (Warn)
[2018-09-13T03:25:28.241331] : in /ndt_matching: exception was catched
[2018-09-13T03:25:28.241375] : in /ndt_matching: Divided by zero.
[2018-09-13T03:25:28.340549] : in /ndt_matching: topic /ndt_matching/data subscribe rate was low (Warn)
[2018-09-13T03:25:29.340556] : in /ndt_matching: topic /ndt_matching/data subscribe rate was low (Warn)
[2018-09-13T03:25:30.340551] : in /ndt_matching: topic /ndt_matching/data subscribe rate was low (Warn)
[2018-09-13T03:25:30.341377] : in /ndt_matching: The input value hogehoge is out of range.
[2018-09-13T03:25:30.441393] : in /ndt_matching: The input value hogehoge is out of range.
[2018-09-13T03:25:30.541348] : in /ndt_matching: The input value hogehoge is out of range.
[2018-09-13T03:25:30.641382] : in /ndt_matching: The input value hogehoge is out of range.

Each diag_manager publish diagnostic infomation as a diag_msgs/diag_error type message when it found someting wrong.
Watchdog node collect all diagnostic information and publish /watchdog_node/all/diag topic.

Error code defines in yaml format like below.

/watchdog_node:
  node_number: 0
  errors:
  - num: 0
    name: required file does not exist
    category: 7    
    description: required file does not exist
/pure_pursuit:
  node_number: 0
  errors:
  - num: 0
    name:  out of range
    category: 8    
    description: The input value hogehoge is out of range.
  - num: 1
    name:  node is ded
    category: 3    
    description: /pure_pursuit node is dead
/ndt_matching:
  node_number: 1
  errors:
  - num: 0
    name: out of range    
    category: 8    
    description: The input value hogehoge is out of range.
  - num: 1
    name: exception    
    category: 2    
    description: exception was catched
  - num: 2
    name: low subscribe rate
    category: 4
    description: topic /nft_matching/data subscribe rate was low (Warn)
    threshold: 10.0
    level: warn
  - num: 3
    name: low subscribe rate
    category: 4
    description: topic /nft_matching/data subscribe rate was low (Error)
    threshold: 5.0
    level: error
  - num: 4
    name: low reliability
    category: 12
    description: low reliability
  - num: 5
    name:  node is ded
    category: 3    
description: /ndt_matching node is dead

documentation is here.

[kfunaoka]

@hakuturu583 This issue focus on software error, and hardware error such as sensors is another topic. Is it right?

[hakuturu583]

Yes.
This library only handles software diagnostics.
Hardware diagnostics will add later.

[hakuturu583]

this Issue was updated by autowarefoundation/autoware#1943