Files
distcloud/distributedcloud/dcmanager/orchestrator/states/lock_host.py
Cristian Mondo a6a6b84258 Subcloud Name Reconfiguration
This change adds the capability to rename the subcloud after
bootstrap or during subcloud rehome operation.

Added a field in the database to separate the region name
from the subcloud name.
The region name determines the subcloud reference in the
Openstack core, through which it is possible to access
the endpoints of a given subcloud. Since the region name
cannot be changed, this commit adds the ability to maintain
a unique region name based on the UUID format, and allows
subcloud renaming when necessary without any endpoint
impact.
The region is randomly generated to configure the subcloud
when it is created and only applies to future subclouds.
For those systems that have existing subclouds, the region
will be the same as on day 0, that is, region will keep the
same name as the subcloud, but subclouds can be renamed.

This topic involves changes to dcmanager, dcmanager-client
and GUI. To ensure the region name reference needed by the
cert-monitor, a mechanism to determine if the request is
coming from the cert-monitor has been created.

Usage for subcloud rename:
dcmanager subcloud update <subcloud-name> --name <new-name>

Usage for subcloud rehoming:
dcmanager subcloud add --name <subcloud-name> --migrate ...

Note: Upgrade test from StarlingX 8 -> 9 for this commit
is deferred until upgrade functionality in master is
restored. Any issue found during upgrade test will be
addressed in a separate commit

Test Plan:
PASS: Run dcmanager subcloud passing subcommands:
      - add/delete/migrate/list/show/show --detail
      - errors/manage/unmanage/reinstall/reconfig
      - update/deploy
PASS: Run dcmanager subcloud add supplying --name
      parameter and validate the operation is not allowed
PASS: Run dcmanager supplying subcommands:
      - kube/patch/prestage strategies
PASS: Run dcmanager to apply patch and remove it
PASS: Run dcmanager subcloud-backup:
      - create/delete/restore/show/upload
PASS: Run subcloud-group:
      - add/delete/list/list-subclouds/show/update
PASS: Run dcmanager subcloud strategy for:
      - patch/kubernetes/firmware
PASS: Run dcmanager subcloud update command passing --name
      parameter supplying the following values:
      - current subcloud name (not changed)
      - different existing subcloud name
PASS: Run dcmanager to migrate a subcloud passing --name
      parameter supplying a new subcloud name
PASS: Run dcmanager to migrate a subcloud without --name
      parameter
PASS: Run dcmanager to migrate a subcloud passing --name
      parameter supplying a new subcloud name and
      different subcloud name in bootstrap file
PASS: Test dcmanager API response using cURL command line
      to validate new region name field
PASS: Run full DC sanity and regression

Story: 2010788
Task: 48217

Signed-off-by: Cristian Mondo <cristian.mondo@windriver.com>
Change-Id: Id04f42504b8e325d9ec3880c240fe4a06e3a20b7
2023-09-07 10:30:06 -03:00

78 lines
3.1 KiB
Python

#
# Copyright (c) 2020-2023 Wind River Systems, Inc.
#
# SPDX-License-Identifier: Apache-2.0
#
import time
from dcmanager.common import consts
from dcmanager.common.exceptions import StrategyStoppedException
from dcmanager.orchestrator.states.base import BaseState
# Max time: 10 minutes = 60 queries x 10 seconds
DEFAULT_MAX_QUERIES = 60
DEFAULT_SLEEP_DURATION = 10
class LockHostState(BaseState):
"""Orchestration state for locking a host"""
def __init__(self, next_state, region_name, hostname):
super(LockHostState, self).__init__(
next_state=next_state, region_name=region_name)
self.target_hostname = hostname
# max time to wait (in seconds) is: sleep_duration * max_queries
self.sleep_duration = DEFAULT_SLEEP_DURATION
self.max_queries = DEFAULT_MAX_QUERIES
def perform_state_action(self, strategy_step):
"""Locks a host on the subcloud
Returns the next state in the state machine on success.
Any exceptions raised by this method set the strategy to FAILED.
"""
# Create a sysinv client on the subcloud
sysinv_client = self.get_sysinv_client(strategy_step.subcloud.region_name)
host = sysinv_client.get_host(self.target_hostname)
# if the host is already in the desired state, no need for action
if host.administrative == consts.ADMIN_LOCKED:
msg = "Host: %s already: %s." % (self.target_hostname,
host.administrative)
self.info_log(strategy_step, msg)
return self.next_state
# Invoke the action
# ihost_action is 'lock' and task is set to 'Locking'
response = sysinv_client.lock_host(host.id)
if (response.ihost_action != 'lock' or response.task != 'Locking'):
raise Exception("Unable to lock host %s" % self.target_hostname)
# this action is asynchronous, query until it completes or times out
counter = 0
while True:
# If event handler stop has been triggered, fail the state
if self.stopped():
raise StrategyStoppedException()
# query the administrative state to see if it is the new state.
host = self.get_sysinv_client(
strategy_step.subcloud.region_name).get_host(self.target_hostname)
if host.administrative == consts.ADMIN_LOCKED:
msg = "Host: %s is now: %s" % (self.target_hostname,
host.administrative)
self.info_log(strategy_step, msg)
break
counter += 1
if counter >= self.max_queries:
raise Exception("Timeout waiting for lock to complete. "
"Please check sysinv.log on the subcloud "
"for details.")
time.sleep(self.sleep_duration)
# If we are here, the loop broke out cleanly and the action succeeded
# When we return from this method without throwing an exception, the
# state machine can proceed to the next state
return self.next_state