Project

General

Profile

Actions

Cleanup #61408

closed

Improve reliability and organization around cephadm deployment

Added by John Mulligan 12 months ago. Updated 3 months ago.

Status:
Resolved
Priority:
Normal
Assignee:
-
Category:
cephadm (binary)
Target version:
-
% Done:

0%

Tags:
backport_processed
Backport:
reef
Reviewed:
Affected Versions:
Pull request ID:

Description

The cephadm binary provides a `deploy` command that is used to create services on nodes. This command uses a mix of various command line arguments that take single input values, command line arguments that take JSON-encoded strings, and json-encoded input on stdin.

This mish-mash of inputs makes fixes and extensions to cephadm more difficult. In addition, the current method of passing extra CLI options for entrypoint arguments and container engine arguments will fail if the argument should have spaces in it. The plan is to add a new `cephadm _orch deploy` command that takes the vast majority of configuration parameters as a single JSON encoded object on the stdin (or as a specific input file).

The `_orch` term is a new special "namespace" in cephadm that serves as a boundary between commands that can / should be run by users directly and commands that are meant primarily to be called by the orchestrator module and use by humans/scripts should be very rare. These commands are permitted to be tightly coupled with the needs of the orchestrator module.

This plan was discussed in the ceph orchestrators weekly meeting on May 16, 2023.


Related issues 1 (0 open1 closed)

Copied to Orchestrator - Backport #61806: reef: Improve reliability and organization around cephadm deploymentResolvedAdam KingActions
Actions

Also available in: Atom PDF