Management.vue gains a second tab listing MySQL delivery tasks from the backend (search, status filter, detail dialog, one-time credential reveal with copy/download). components.d.ts regenerated for the new Element Plus components.
Also add narrow-viewport (<=760px) style tweaks to AppHeader/AppMain/DefaultLayout; these are partial small-screen optimizations, not a full mobile adaptation.
Rebuild the MySQL standardized delivery form in Catalog.vue: sectioned layout (scope/resources/db config/advanced params), client-generated root password with regenerate/copy, storage slider, install/data path hints, inline field validation, delivery history and task polling.
Align CreateMySQLDeliveryPayload with the backend MySQLDeliveryInput contract (cpu_milli/memory_mi/storage_gi required, legacy fields optional, drop nonexistent param_template) and send mysql_root_password/mysql_admin_password as a pair so the credential shown to the user is the one actually deployed.
Add retryRollback / releaseRollback / retryCloudDMRegistration API
bindings and wire the corresponding actions and status presentation
into the delivery task view.
- POST /delivery/tasks/:id/rollback/retry (platform admin) re-launches
the rollback job after a previous cleanup failure
- POST /delivery/tasks/:id/rollback/release (platform admin) releases
bookkeeping after the operator verified the target host manually
- POST /delivery/tasks/:id/clouddm/retry retries only the CloudDM
registration step for an already healthy instance
- task logs now include rollback AWX job stdout and localized
text/classes for the new rollback states
Introduce a compensating workflow that launches the dedicated AWX
rollback template when a deployment cannot be started or fails midway.
- add task states: rollback_pending, rolling_back, rolled_back,
rollback_failed, rollback_acknowledged
- add RollbackJob model to track the compensating AWX run separately
from the deploy run, preserving both job IDs for audit
- hold resource reservations in 'rollback' status until cleanup
succeeds so a failed cleanup cannot be masked by a later delivery
- poll rollback jobs with a 2-minute launch timeout; an unknown launch
result surfaces as a recoverable failure instead of re-launching
- protect finished/register_failed/rolled-back tasks from rollback;
register_failed keeps the healthy instance and its resource usage
- add DELIVERY_ROLLBACK_TEMPLATE_ID config; without it, failures are
marked rollback_failed and require manual cleanup
- use unique pending-<task_id> placeholder for executor job IDs
innodb_redo_log_capacity passed as '' or 'auto' from the delivery
payload previously bypassed the memory-based tier calculation and was
rendered verbatim into the instance config. Fall back to the automatic
tier when the value is empty or 'auto'.
Add a compensating playbook that removes a single delivered MySQL
instance from a target host. The rollback scope is pinned by
target_hosts + instance_name + data_disk, with pre-task assertions
rejecting instance names or disk paths outside the delivery layout
to prevent accidental deletion.
- stop and disable mysql-delivery@<instance>.service, reset failed state
- remove instance data dir, install dir, config file and run dir
- Per-host instance limit (DELIVERY_HOST_INSTANCE_LIMIT, default 4)
- Optional target_host to pin a host from the candidate pool
- Regenerate swagger docs
- 8.0 from Ubuntu archive, 8.4 from MySQL official APT repo
- 5.6/5.7 not supported for now; enforce one MySQL series per host
- Sync Go version whitelist with the playbook package map
- Add topology/port/data_disk, DB options and 8 advanced parameters
with whitelist validation, rendered via extra_vars into the playbook
- Hybrid port allocation over pool 13306-13999
- Tighten bounds: memory 2048-65536 MiB, storage 20-2000 GiB
Implement AWXClient with Launch, GetJob, and Cancel methods for
orchestrating Ansible job templates via AWX REST API. Uses Bearer
token auth with 30s timeout. Includes unit tests with mock transport.
Relates-to: #97
qudder
wonder