feat(infra): add HPMN backup workflow#740
Merged
vivekgsharma merged 7 commits intov1.0-devfrom Apr 15, 2026
Merged
Conversation
|
Note Reviews pausedIt looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the Use the following commands to manage reviews:
Use the checkboxes below for quick actions:
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (2)
🚧 Files skipped from review as they are similar to previous changes (2)
📝 WalkthroughWalkthroughAdds manual HP masternode backup and restore: GitHub Actions workflows, Ansible playbooks/roles (install/run/restore/finalize), backup & restore shell templates, role defaults/tasks, docs, a small deploy apt tweak, and a dashmate tasks conditional fix. Changes
Sequence Diagram(s)sequenceDiagram
actor User
participant GHA as "GitHub Actions\n(hpmn-backup / hpmn-restore)"
participant Ctrl as "Ansible\nControl Machine"
participant Host as "HP Masternode\nHost"
participant S3 as "AWS S3\nBucket"
User->>GHA: Trigger workflow_dispatch (network, label/uri, options)
GHA->>Ctrl: Prepare controller (checkout, install deps, clone configs, validate inputs)
Ctrl->>Host: (optional) Run install playbook -> deploy tooling & scripts
Ctrl->>Host: Run run playbook -> include role run task
Host->>Host: Execute script (discover, stage, manifest, tar)
Host->>S3: Upload tarball (AES256 or aws:kms)
S3-->>Host: Upload complete
Host-->>Ctrl: Return script stdout (BACKUP/RESTORE metadata)
Ctrl-->>GHA: Playbook complete
GHA-->>User: Workflow finished
Estimated code review effort🎯 4 (Complex) | ⏱️ ~45 minutes Possibly related issues
Poem
🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 3
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In @.github/workflows/hpmn-backup.yml:
- Line 42: The workflow currently disables SSH host verification via
ANSIBLE_HOST_KEY_CHECKING and uses StrictHostKeyChecking no which accepts any
host key; instead seed the runner's SSH known_hosts with pinned host key
fingerprints for the targets and github.com before any SSH/Ansible steps.
Replace the ANSIBLE_HOST_KEY_CHECKING/StrictHostKeyChecking no usage by adding a
step that adds verified host keys (via ssh-keyscan or a fixed fingerprint entry)
into ~/.ssh/known_hosts and ensure the Ansiblе/ssh steps reference that
known_hosts (and remove ANSIBLE_HOST_KEY_CHECKING/StrictHostKeyChecking
overrides). Target symbols to change: ANSIBLE_HOST_KEY_CHECKING,
StrictHostKeyChecking (ssh options), and the workflow steps that call
ssh-keyscan/known_hosts population around where github.com and remote hosts are
used.
In `@ansible/roles/hpmn_backup/tasks/run.yml`:
- Line 7: The resolved SSE mode assignment hpmn_backup_sse_mode_resolved
currently applies default(hpmn_backup_s3_sse_mode,
lookup('env','HPMN_BACKUP_S3_SSE_MODE')) in the wrong order so the role default
always wins; change it to use lookup('env','HPMN_BACKUP_S3_SSE_MODE') first and
then default to the role var hpmn_backup_s3_sse_mode (i.e.,
hpmn_backup_sse_mode_resolved: "{{ lookup('env', 'HPMN_BACKUP_S3_SSE_MODE') |
default(hpmn_backup_s3_sse_mode, true) }}") so the environment variable
overrides the role default and preserves consistent SSE/KMS behavior.
In `@ansible/roles/hpmn_backup/templates/backup-hpmn.sh.j2`:
- Around line 70-72: The backup script currently uses cp -a on live paths (see
use of cp -a "$path" "$payload_dir" and record_path), which can produce
inconsistent, non-restorable backups for active Tendermint/Tenderdash and Dash
Core volumes; update the template to quiesce or snapshot those services before
copying: add logic that either (a) stops or pauses the relevant
services/containers (referencing their service names used in your
playbooks/defaults) and waits for clean shutdown, or (b) creates a
filesystem-level consistent snapshot (LVM, ZFS, or cloud disk snapshot) of the
underlying volume and copy from the snapshot path, then resume/start services;
ensure the script records which method was used and only runs cp/rsync from the
quiesced or snapshot path before calling record_path so uploaded archives are
recovery-grade.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: ccd9a15d-27c3-4ee2-b195-a288c106c87b
📒 Files selected for processing (8)
.github/workflows/hpmn-backup.ymlansible/hpmn_backup_install.ymlansible/hpmn_backup_run.ymlansible/roles/hpmn_backup/defaults/main.ymlansible/roles/hpmn_backup/tasks/main.ymlansible/roles/hpmn_backup/tasks/run.ymlansible/roles/hpmn_backup/templates/backup-hpmn.sh.j2docs/hpmn-backup.md
| AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }} | ||
| AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }} | ||
| AWS_REGION: ${{ secrets.AWS_REGION }} | ||
| ANSIBLE_HOST_KEY_CHECKING: "false" |
There was a problem hiding this comment.
Don't disable SSH host verification for this workflow.
Line 42 and Lines 77-87 make the runner trust any host key for both Ansible and github.com. A spoofed repo endpoint or target host would then receive the backup commands and the AWS credentials that ansible/roles/hpmn_backup/tasks/run.yml forwards into the remote environment. Please seed known_hosts with pinned fingerprints instead of using StrictHostKeyChecking no / ANSIBLE_HOST_KEY_CHECKING=false.
Also applies to: 77-87
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In @.github/workflows/hpmn-backup.yml at line 42, The workflow currently
disables SSH host verification via ANSIBLE_HOST_KEY_CHECKING and uses
StrictHostKeyChecking no which accepts any host key; instead seed the runner's
SSH known_hosts with pinned host key fingerprints for the targets and github.com
before any SSH/Ansible steps. Replace the
ANSIBLE_HOST_KEY_CHECKING/StrictHostKeyChecking no usage by adding a step that
adds verified host keys (via ssh-keyscan or a fixed fingerprint entry) into
~/.ssh/known_hosts and ensure the Ansiblе/ssh steps reference that known_hosts
(and remove ANSIBLE_HOST_KEY_CHECKING/StrictHostKeyChecking overrides). Target
symbols to change: ANSIBLE_HOST_KEY_CHECKING, StrictHostKeyChecking (ssh
options), and the workflow steps that call ssh-keyscan/known_hosts population
around where github.com and remote hosts are used.
Comment on lines
+70
to
+72
| if [[ -e "$path" ]]; then | ||
| cp -a --parents "$path" "$payload_dir" | ||
| record_path "included" "$category" "$path" |
There was a problem hiding this comment.
cp -a on live node state won't give you a recovery-grade backup.
This helper is later fed active Tenderdash and Dash Core volume paths from ansible/roles/hpmn_backup/defaults/main.yml, so it can copy files from different points in time while the containers are still writing. The upload can succeed and still produce an archive that is not restorable. Please quiesce those services or snapshot/export the underlying volume before copying database/state directories.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@ansible/roles/hpmn_backup/templates/backup-hpmn.sh.j2` around lines 70 - 72,
The backup script currently uses cp -a on live paths (see use of cp -a "$path"
"$payload_dir" and record_path), which can produce inconsistent, non-restorable
backups for active Tendermint/Tenderdash and Dash Core volumes; update the
template to quiesce or snapshot those services before copying: add logic that
either (a) stops or pauses the relevant services/containers (referencing their
service names used in your playbooks/defaults) and waits for clean shutdown, or
(b) creates a filesystem-level consistent snapshot (LVM, ZFS, or cloud disk
snapshot) of the underlying volume and copy from the snapshot path, then
resume/start services; ensure the script records which method was used and only
runs cp/rsync from the quiesced or snapshot path before calling record_path so
uploaded archives are recovery-grade.
There was a problem hiding this comment.
Actionable comments posted: 5
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
ansible/roles/dashmate/tasks/main.yml (1)
129-136:⚠️ Potential issue | 🟡 MinorUse
{{ dashmate_user }}in the docker-group task.Line 131 still targets the literal
dashmateaccount, so any override ofdashmate_usercreates one user but adds a different user to thedockergroup. That leaves the actual service account unable to manage Docker.♻️ Proposed fix
- name: Add dashmate user to docker group ansible.builtin.user: - name: 'dashmate' + name: '{{ dashmate_user }}' groups: docker append: true🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@ansible/roles/dashmate/tasks/main.yml` around lines 129 - 136, The task "Add dashmate user to docker group" currently hardcodes name: 'dashmate' — change it to use the dashmate_user variable ({{ dashmate_user }}) so the user added to the docker group matches any overridden account; update the ansible.builtin.user entry for name accordingly while keeping groups: docker, append: true and the existing when conditions (not (skip_dashmate_image_update | default(false)) and dashmate_user_exists).
♻️ Duplicate comments (1)
ansible/roles/hpmn_backup/templates/backup-hpmn.sh.j2 (1)
72-74:⚠️ Potential issue | 🟠 MajorQuiesce or snapshot stateful paths before copying.
At Line 73,
cp -areads live Tenderdash/Core volume data. That can produce a successfully uploaded but non-restorable archive due to mixed-time state. Please copy from a quiesced or snapshot source and record the consistency method inmanifest.txt.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@ansible/roles/hpmn_backup/templates/backup-hpmn.sh.j2` around lines 72 - 74, The backup currently runs cp -a on live stateful paths (the if [[ -e "$path" ]]; then cp -a --parents "$path" "$payload_dir" record_path "included" "$category" "$path") which can produce inconsistent snapshots; before copying each stateful path referenced by $path, ensure the service/volume is quiesced or a filesystem/volume snapshot is created and copy from that snapshot location instead of the live path, and update the manifest consistency entry by passing/recording the consistency method (e.g., "quiesced" or "snapshot:<snapshot-id>") via the record_path invocation or by appending to manifest.txt so manifest.txt records how each $path was frozen for restoreability.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In @.github/workflows/hpmn-restore.yml:
- Around line 103-140: Add a preflight step before any ansible-playbook runs
that verifies the workflow input TARGET_HOST resolves to exactly one host from
the hp_masternodes group in the inventory networks/${NETWORK_NAME}.inventory;
implement this by using Ansible’s inventory listing to filter hp_masternodes by
the LIMIT (TARGET_HOST), count the resulting hosts, and fail the job unless the
count is exactly 1 (so the subsequent steps with --limit "${TARGET_HOST}" cannot
no-op on zero hosts). Ensure the step references TARGET_HOST, hp_masternodes,
and networks/${NETWORK_NAME}.inventory so reviewers can find it easily and make
the workflow exit non-zero with a clear error if the count is not 1.
In `@ansible/roles/hpmn_restore/tasks/run.yml`:
- Around line 8-30: Add a preflight check that fails fast when the restore
script is not present or not executable: use ansible.builtin.stat on the
hpmn_restore_script_path and assert that stat_result.exists and
(stat_result.mode is executable) before the "Run HP masternode restore script"
task (or replace the current "Validate restore inputs" with this extra
assertion). Reference the variable hpmn_restore_script_path and register the
stat (e.g., stat_result) so you can use an ansible.builtin.assert to abort with
a clear message when the script file is missing or not executable instead of
letting the command task hit ENOENT.
In `@ansible/roles/hpmn_restore/templates/restore-hpmn.sh.j2`:
- Around line 47-58: The manifest loop currently trusts each path and can be
tricked to escape payload_dir or write arbitrary destinations; update the loop
that reads status, category, path (and uses payload_dir and source_path) to
canonicalize and validate both source and destination before any mkdir/rsync:
compute a normalized absolute source (e.g., realpath -m
"${payload_dir}/${path}") and normalized destination (e.g., realpath -m
"${restore_root}/${path}" or ensure path is relative and join with restore
root), then ensure the source path starts with payload_dir canonical prefix and
the destination starts with the intended restore root prefix; if an included
entry's source does not exist or validation fails, log an error and exit
non-zero instead of silently skipping, otherwise proceed with mkdir and rsync
only for validated paths.
In `@docs/hpmn-backup.md`:
- Around line 145-149: The markdown link currently points to an absolute local
path (`/home/.../docs/hpmn-recovery.md`) which will break on GitHub; update the
link in the Restore section (the
`[`docs/hpmn-recovery.md`](/home/.../docs/hpmn-recovery.md)` instance) to a
repository-relative path such as `docs/hpmn-recovery.md` or `./hpmn-recovery.md`
so it resolves correctly in the repo and on GitHub.
- Around line 123-125: The docs contain inconsistent capitalization of "GitHub"
(e.g., the header "GitHub Actions Workflow" and other occurrences of "Github");
update all instances of "Github" to the correct "GitHub" spelling throughout the
document (including the header and the occurrence noted around the same
section), ensuring consistent casing everywhere.
---
Outside diff comments:
In `@ansible/roles/dashmate/tasks/main.yml`:
- Around line 129-136: The task "Add dashmate user to docker group" currently
hardcodes name: 'dashmate' — change it to use the dashmate_user variable ({{
dashmate_user }}) so the user added to the docker group matches any overridden
account; update the ansible.builtin.user entry for name accordingly while
keeping groups: docker, append: true and the existing when conditions (not
(skip_dashmate_image_update | default(false)) and dashmate_user_exists).
---
Duplicate comments:
In `@ansible/roles/hpmn_backup/templates/backup-hpmn.sh.j2`:
- Around line 72-74: The backup currently runs cp -a on live stateful paths (the
if [[ -e "$path" ]]; then cp -a --parents "$path" "$payload_dir" record_path
"included" "$category" "$path") which can produce inconsistent snapshots; before
copying each stateful path referenced by $path, ensure the service/volume is
quiesced or a filesystem/volume snapshot is created and copy from that snapshot
location instead of the live path, and update the manifest consistency entry by
passing/recording the consistency method (e.g., "quiesced" or
"snapshot:<snapshot-id>") via the record_path invocation or by appending to
manifest.txt so manifest.txt records how each $path was frozen for
restoreability.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: ca38acc5-e386-4c1a-8437-200686666e6f
📒 Files selected for processing (18)
.github/workflows/hpmn-backup.yml.github/workflows/hpmn-restore.ymlansible/deploy.ymlansible/hpmn_backup_install.ymlansible/hpmn_backup_run.ymlansible/hpmn_restore_finalize.ymlansible/hpmn_restore_install.ymlansible/hpmn_restore_run.ymlansible/roles/dashmate/tasks/main.ymlansible/roles/hpmn_backup/defaults/main.ymlansible/roles/hpmn_backup/tasks/run.ymlansible/roles/hpmn_backup/templates/backup-hpmn.sh.j2ansible/roles/hpmn_restore/defaults/main.ymlansible/roles/hpmn_restore/tasks/main.ymlansible/roles/hpmn_restore/tasks/run.ymlansible/roles/hpmn_restore/templates/restore-hpmn.sh.j2docs/hpmn-backup.mddocs/hpmn-recovery.md
✅ Files skipped from review due to trivial changes (8)
- ansible/deploy.yml
- ansible/hpmn_restore_install.yml
- ansible/hpmn_backup_install.yml
- docs/hpmn-recovery.md
- ansible/roles/hpmn_restore/defaults/main.yml
- .github/workflows/hpmn-backup.yml
- ansible/roles/hpmn_backup/defaults/main.yml
- ansible/hpmn_backup_run.yml
🚧 Files skipped from review as they are similar to previous changes (1)
- ansible/roles/hpmn_backup/tasks/run.yml
Comment on lines
+103
to
+140
| - name: Validate restore config | ||
| run: | | ||
| test -f "networks/${NETWORK_NAME}.inventory" | ||
| test -f "networks/${NETWORK_NAME}.yml" | ||
| test -n "$HPMN_RESTORE_S3_URI" | ||
| test -n "$AWS_REGION" | ||
|
|
||
| - name: Install restore tooling on target host | ||
| if: ${{ inputs.install_restore_tooling }} | ||
| run: | | ||
| ansible-playbook \ | ||
| -i "networks/${NETWORK_NAME}.inventory" \ | ||
| ansible/hpmn_restore_install.yml \ | ||
| -e "@networks/${NETWORK_NAME}.yml" \ | ||
| -e "dash_network_name=${NETWORK_NAME}" \ | ||
| --limit "${TARGET_HOST}" | ||
|
|
||
| - name: Restore target host from S3 backup | ||
| run: | | ||
| ansible-playbook \ | ||
| -i "networks/${NETWORK_NAME}.inventory" \ | ||
| ansible/hpmn_restore_run.yml \ | ||
| -e "@networks/${NETWORK_NAME}.yml" \ | ||
| -e "dash_network_name=${NETWORK_NAME}" \ | ||
| -e "hpmn_restore_s3_uri=${HPMN_RESTORE_S3_URI}" \ | ||
| -e "hpmn_restore_start_services=${{ inputs.start_services }}" \ | ||
| --limit "${TARGET_HOST}" | ||
|
|
||
| - name: Finalize restored target host | ||
| if: ${{ inputs.finalize_restore }} | ||
| run: | | ||
| ansible-playbook \ | ||
| -i "networks/${NETWORK_NAME}.inventory" \ | ||
| ansible/hpmn_restore_finalize.yml \ | ||
| -e "@networks/${NETWORK_NAME}.yml" \ | ||
| -e "dash_network_name=${NETWORK_NAME}" \ | ||
| -e "dash_network=${NETWORK_NAME}" \ | ||
| --limit "${TARGET_HOST}" |
There was a problem hiding this comment.
Validate that target_host resolves to exactly one HP masternode before running Ansible.
The playbook-level assert only protects the “more than one host” case. If --limit "${TARGET_HOST}" matches zero hosts, Ansible can no-op and this workflow still looks successful. Add a preflight inventory check and fail unless the limit resolves to exactly one host in hp_masternodes.
🛠️ Proposed fix
- name: Validate restore config
run: |
test -f "networks/${NETWORK_NAME}.inventory"
test -f "networks/${NETWORK_NAME}.yml"
test -n "$HPMN_RESTORE_S3_URI"
test -n "$AWS_REGION"
+ matched_hosts="$(
+ ansible hp_masternodes \
+ -i "networks/${NETWORK_NAME}.inventory" \
+ --limit "${TARGET_HOST}" \
+ --list-hosts | tail -n +2 | sed '/^[[:space:]]*$/d' | wc -l
+ )"
+ test "$matched_hosts" -eq 1📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| - name: Validate restore config | |
| run: | | |
| test -f "networks/${NETWORK_NAME}.inventory" | |
| test -f "networks/${NETWORK_NAME}.yml" | |
| test -n "$HPMN_RESTORE_S3_URI" | |
| test -n "$AWS_REGION" | |
| - name: Install restore tooling on target host | |
| if: ${{ inputs.install_restore_tooling }} | |
| run: | | |
| ansible-playbook \ | |
| -i "networks/${NETWORK_NAME}.inventory" \ | |
| ansible/hpmn_restore_install.yml \ | |
| -e "@networks/${NETWORK_NAME}.yml" \ | |
| -e "dash_network_name=${NETWORK_NAME}" \ | |
| --limit "${TARGET_HOST}" | |
| - name: Restore target host from S3 backup | |
| run: | | |
| ansible-playbook \ | |
| -i "networks/${NETWORK_NAME}.inventory" \ | |
| ansible/hpmn_restore_run.yml \ | |
| -e "@networks/${NETWORK_NAME}.yml" \ | |
| -e "dash_network_name=${NETWORK_NAME}" \ | |
| -e "hpmn_restore_s3_uri=${HPMN_RESTORE_S3_URI}" \ | |
| -e "hpmn_restore_start_services=${{ inputs.start_services }}" \ | |
| --limit "${TARGET_HOST}" | |
| - name: Finalize restored target host | |
| if: ${{ inputs.finalize_restore }} | |
| run: | | |
| ansible-playbook \ | |
| -i "networks/${NETWORK_NAME}.inventory" \ | |
| ansible/hpmn_restore_finalize.yml \ | |
| -e "@networks/${NETWORK_NAME}.yml" \ | |
| -e "dash_network_name=${NETWORK_NAME}" \ | |
| -e "dash_network=${NETWORK_NAME}" \ | |
| --limit "${TARGET_HOST}" | |
| - name: Validate restore config | |
| run: | | |
| test -f "networks/${NETWORK_NAME}.inventory" | |
| test -f "networks/${NETWORK_NAME}.yml" | |
| test -n "$HPMN_RESTORE_S3_URI" | |
| test -n "$AWS_REGION" | |
| matched_hosts="$( | |
| ansible hp_masternodes \ | |
| -i "networks/${NETWORK_NAME}.inventory" \ | |
| --limit "${TARGET_HOST}" \ | |
| --list-hosts | tail -n +2 | sed '/^[[:space:]]*$/d' | wc -l | |
| )" | |
| test "$matched_hosts" -eq 1 | |
| - name: Install restore tooling on target host | |
| if: ${{ inputs.install_restore_tooling }} | |
| run: | | |
| ansible-playbook \ | |
| -i "networks/${NETWORK_NAME}.inventory" \ | |
| ansible/hpmn_restore_install.yml \ | |
| -e "@networks/${NETWORK_NAME}.yml" \ | |
| -e "dash_network_name=${NETWORK_NAME}" \ | |
| --limit "${TARGET_HOST}" | |
| - name: Restore target host from S3 backup | |
| run: | | |
| ansible-playbook \ | |
| -i "networks/${NETWORK_NAME}.inventory" \ | |
| ansible/hpmn_restore_run.yml \ | |
| -e "@networks/${NETWORK_NAME}.yml" \ | |
| -e "dash_network_name=${NETWORK_NAME}" \ | |
| -e "hpmn_restore_s3_uri=${HPMN_RESTORE_S3_URI}" \ | |
| -e "hpmn_restore_start_services=${{ inputs.start_services }}" \ | |
| --limit "${TARGET_HOST}" | |
| - name: Finalize restored target host | |
| if: ${{ inputs.finalize_restore }} | |
| run: | | |
| ansible-playbook \ | |
| -i "networks/${NETWORK_NAME}.inventory" \ | |
| ansible/hpmn_restore_finalize.yml \ | |
| -e "@networks/${NETWORK_NAME}.yml" \ | |
| -e "dash_network_name=${NETWORK_NAME}" \ | |
| -e "dash_network=${NETWORK_NAME}" \ | |
| --limit "${TARGET_HOST}" |
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In @.github/workflows/hpmn-restore.yml around lines 103 - 140, Add a preflight
step before any ansible-playbook runs that verifies the workflow input
TARGET_HOST resolves to exactly one host from the hp_masternodes group in the
inventory networks/${NETWORK_NAME}.inventory; implement this by using Ansible’s
inventory listing to filter hp_masternodes by the LIMIT (TARGET_HOST), count the
resulting hosts, and fail the job unless the count is exactly 1 (so the
subsequent steps with --limit "${TARGET_HOST}" cannot no-op on zero hosts).
Ensure the step references TARGET_HOST, hp_masternodes, and
networks/${NETWORK_NAME}.inventory so reviewers can find it easily and make the
workflow exit non-zero with a clear error if the count is not 1.
Comment on lines
+8
to
+30
| - name: Validate restore inputs | ||
| ansible.builtin.assert: | ||
| that: | ||
| - hpmn_restore_s3_uri_resolved | length > 0 | ||
| - hpmn_restore_aws_region_resolved | length > 0 | ||
| fail_msg: >- | ||
| HPMN restore requires HPMN_RESTORE_S3_URI or hpmn_restore_s3_uri, plus AWS_REGION. | ||
|
|
||
| - name: Run HP masternode restore script | ||
| ansible.builtin.command: | ||
| argv: | ||
| - "{{ hpmn_restore_script_path }}" | ||
| environment: | ||
| AWS_ACCESS_KEY_ID: "{{ lookup('env', 'AWS_ACCESS_KEY_ID') }}" | ||
| AWS_SECRET_ACCESS_KEY: "{{ lookup('env', 'AWS_SECRET_ACCESS_KEY') }}" | ||
| AWS_SESSION_TOKEN: "{{ lookup('env', 'AWS_SESSION_TOKEN') }}" | ||
| AWS_REGION: "{{ hpmn_restore_aws_region_resolved }}" | ||
| AWS_DEFAULT_REGION: "{{ hpmn_restore_aws_region_resolved }}" | ||
| HPMN_RESTORE_S3_URI: "{{ hpmn_restore_s3_uri_resolved }}" | ||
| HPMN_RESTORE_NETWORK: "{{ hpmn_restore_network_name }}" | ||
| HPMN_RESTORE_DASHMATE_HOME: "{{ hpmn_restore_dashmate_home }}" | ||
| HPMN_RESTORE_START_SERVICES: "{{ 'true' if hpmn_restore_start_services_resolved else 'false' }}" | ||
| register: hpmn_restore_run_result |
There was a problem hiding this comment.
Fail fast if the restore script was never installed.
.github/workflows/hpmn-restore.yml defaults install_restore_tooling to false, but this role invokes {{ hpmn_restore_script_path }} unconditionally. On a fresh replacement node that becomes a late ENOENT instead of a clear preflight error.
🛠️ Proposed fix
- name: Validate restore inputs
ansible.builtin.assert:
that:
- hpmn_restore_s3_uri_resolved | length > 0
- hpmn_restore_aws_region_resolved | length > 0
fail_msg: >-
HPMN restore requires HPMN_RESTORE_S3_URI or hpmn_restore_s3_uri, plus AWS_REGION.
+- name: Check restore script is installed
+ ansible.builtin.stat:
+ path: "{{ hpmn_restore_script_path }}"
+ register: hpmn_restore_script_stat
+
+- name: Validate restore script presence
+ ansible.builtin.assert:
+ that:
+ - hpmn_restore_script_stat.stat.exists
+ - hpmn_restore_script_stat.stat.isreg
+ fail_msg: >-
+ Restore script {{ hpmn_restore_script_path }} is missing. Run hpmn_restore_install first
+ or set install_restore_tooling=true.
+
- name: Run HP masternode restore script
ansible.builtin.command:
argv:
- "{{ hpmn_restore_script_path }}"📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| - name: Validate restore inputs | |
| ansible.builtin.assert: | |
| that: | |
| - hpmn_restore_s3_uri_resolved | length > 0 | |
| - hpmn_restore_aws_region_resolved | length > 0 | |
| fail_msg: >- | |
| HPMN restore requires HPMN_RESTORE_S3_URI or hpmn_restore_s3_uri, plus AWS_REGION. | |
| - name: Run HP masternode restore script | |
| ansible.builtin.command: | |
| argv: | |
| - "{{ hpmn_restore_script_path }}" | |
| environment: | |
| AWS_ACCESS_KEY_ID: "{{ lookup('env', 'AWS_ACCESS_KEY_ID') }}" | |
| AWS_SECRET_ACCESS_KEY: "{{ lookup('env', 'AWS_SECRET_ACCESS_KEY') }}" | |
| AWS_SESSION_TOKEN: "{{ lookup('env', 'AWS_SESSION_TOKEN') }}" | |
| AWS_REGION: "{{ hpmn_restore_aws_region_resolved }}" | |
| AWS_DEFAULT_REGION: "{{ hpmn_restore_aws_region_resolved }}" | |
| HPMN_RESTORE_S3_URI: "{{ hpmn_restore_s3_uri_resolved }}" | |
| HPMN_RESTORE_NETWORK: "{{ hpmn_restore_network_name }}" | |
| HPMN_RESTORE_DASHMATE_HOME: "{{ hpmn_restore_dashmate_home }}" | |
| HPMN_RESTORE_START_SERVICES: "{{ 'true' if hpmn_restore_start_services_resolved else 'false' }}" | |
| register: hpmn_restore_run_result | |
| - name: Validate restore inputs | |
| ansible.builtin.assert: | |
| that: | |
| - hpmn_restore_s3_uri_resolved | length > 0 | |
| - hpmn_restore_aws_region_resolved | length > 0 | |
| fail_msg: >- | |
| HPMN restore requires HPMN_RESTORE_S3_URI or hpmn_restore_s3_uri, plus AWS_REGION. | |
| - name: Check restore script is installed | |
| ansible.builtin.stat: | |
| path: "{{ hpmn_restore_script_path }}" | |
| register: hpmn_restore_script_stat | |
| - name: Validate restore script presence | |
| ansible.builtin.assert: | |
| that: | |
| - hpmn_restore_script_stat.stat.exists | |
| - hpmn_restore_script_stat.stat.isreg | |
| fail_msg: >- | |
| Restore script {{ hpmn_restore_script_path }} is missing. Run hpmn_restore_install first | |
| or set install_restore_tooling=true. | |
| - name: Run HP masternode restore script | |
| ansible.builtin.command: | |
| argv: | |
| - "{{ hpmn_restore_script_path }}" | |
| environment: | |
| AWS_ACCESS_KEY_ID: "{{ lookup('env', 'AWS_ACCESS_KEY_ID') }}" | |
| AWS_SECRET_ACCESS_KEY: "{{ lookup('env', 'AWS_SECRET_ACCESS_KEY') }}" | |
| AWS_SESSION_TOKEN: "{{ lookup('env', 'AWS_SESSION_TOKEN') }}" | |
| AWS_REGION: "{{ hpmn_restore_aws_region_resolved }}" | |
| AWS_DEFAULT_REGION: "{{ hpmn_restore_aws_region_resolved }}" | |
| HPMN_RESTORE_S3_URI: "{{ hpmn_restore_s3_uri_resolved }}" | |
| HPMN_RESTORE_NETWORK: "{{ hpmn_restore_network_name }}" | |
| HPMN_RESTORE_DASHMATE_HOME: "{{ hpmn_restore_dashmate_home }}" | |
| HPMN_RESTORE_START_SERVICES: "{{ 'true' if hpmn_restore_start_services_resolved else 'false' }}" | |
| register: hpmn_restore_run_result |
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@ansible/roles/hpmn_restore/tasks/run.yml` around lines 8 - 30, Add a
preflight check that fails fast when the restore script is not present or not
executable: use ansible.builtin.stat on the hpmn_restore_script_path and assert
that stat_result.exists and (stat_result.mode is executable) before the "Run HP
masternode restore script" task (or replace the current "Validate restore
inputs" with this extra assertion). Reference the variable
hpmn_restore_script_path and register the stat (e.g., stat_result) so you can
use an ansible.builtin.assert to abort with a clear message when the script file
is missing or not executable instead of letting the command task hit ENOENT.
Comment on lines
+47
to
+58
| while IFS='|' read -r status category path; do | ||
| [[ "$status" == "included" ]] || continue | ||
| source_path="${payload_dir}${path}" | ||
|
|
||
| if [[ -d "$source_path" ]]; then | ||
| mkdir -p "$path" | ||
| rsync -a --delete "${source_path}/" "${path}/" | ||
| elif [[ -f "$source_path" ]]; then | ||
| mkdir -p "$(dirname "$path")" | ||
| rsync -a "$source_path" "$path" | ||
| fi | ||
| done < "$manifest_path" |
There was a problem hiding this comment.
Reject unexpected manifest paths before restoring as root.
Lines 47-58 trust every path in manifest.txt as both a source and destination path. Because the archive URI is operator-supplied, a crafted archive can use .. segments or out-of-scope absolute paths to escape payload_dir and overwrite arbitrary files as root. This loop also silently skips included entries that are missing from payload, which can turn archive corruption into a partial “successful” restore.
🔒 Proposed fix
while IFS='|' read -r status category path; do
[[ "$status" == "included" ]] || continue
+ [[ "$path" == /* ]] || {
+ echo "Manifest path must be absolute: $path" >&2
+ exit 1
+ }
+ [[ "$path" != *"/../"* && "$path" != "../"* && "$path" != *"/.." ]] || {
+ echo "Manifest path contains parent traversal: $path" >&2
+ exit 1
+ }
+ case "$path" in
+ /home/dashmate/.dashmate/*|\
+ /var/lib/docker/volumes/dashmate_"${network_name}"_drive_tenderdash/_data*|\
+ /var/lib/docker/volumes/dashmate_"${network_name}"_core_data/_data/.dashcore/testnet3/llmq*)
+ ;;
+ *)
+ echo "Unexpected restore path in manifest: $path" >&2
+ exit 1
+ ;;
+ esac
source_path="${payload_dir}${path}"
if [[ -d "$source_path" ]]; then
mkdir -p "$path"
rsync -a --delete "${source_path}/" "${path}/"
elif [[ -f "$source_path" ]]; then
mkdir -p "$(dirname "$path")"
rsync -a "$source_path" "$path"
+ else
+ echo "Manifest entry marked included is missing from payload: $path" >&2
+ exit 1
fi
done < "$manifest_path"📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| while IFS='|' read -r status category path; do | |
| [[ "$status" == "included" ]] || continue | |
| source_path="${payload_dir}${path}" | |
| if [[ -d "$source_path" ]]; then | |
| mkdir -p "$path" | |
| rsync -a --delete "${source_path}/" "${path}/" | |
| elif [[ -f "$source_path" ]]; then | |
| mkdir -p "$(dirname "$path")" | |
| rsync -a "$source_path" "$path" | |
| fi | |
| done < "$manifest_path" | |
| while IFS='|' read -r status category path; do | |
| [[ "$status" == "included" ]] || continue | |
| [[ "$path" == /* ]] || { | |
| echo "Manifest path must be absolute: $path" >&2 | |
| exit 1 | |
| } | |
| [[ "$path" != *"/../"* && "$path" != "../"* && "$path" != *"/.." ]] || { | |
| echo "Manifest path contains parent traversal: $path" >&2 | |
| exit 1 | |
| } | |
| case "$path" in | |
| /home/dashmate/.dashmate/*|\ | |
| /var/lib/docker/volumes/dashmate_"${network_name}"_drive_tenderdash/_data*|\ | |
| /var/lib/docker/volumes/dashmate_"${network_name}"_core_data/_data/.dashcore/testnet3/llmq*) | |
| ;; | |
| *) | |
| echo "Unexpected restore path in manifest: $path" >&2 | |
| exit 1 | |
| ;; | |
| esac | |
| source_path="${payload_dir}${path}" | |
| if [[ -d "$source_path" ]]; then | |
| mkdir -p "$path" | |
| rsync -a --delete "${source_path}/" "${path}/" | |
| elif [[ -f "$source_path" ]]; then | |
| mkdir -p "$(dirname "$path")" | |
| rsync -a "$source_path" "$path" | |
| else | |
| echo "Manifest entry marked included is missing from payload: $path" >&2 | |
| exit 1 | |
| fi | |
| done < "$manifest_path" |
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@ansible/roles/hpmn_restore/templates/restore-hpmn.sh.j2` around lines 47 -
58, The manifest loop currently trusts each path and can be tricked to escape
payload_dir or write arbitrary destinations; update the loop that reads status,
category, path (and uses payload_dir and source_path) to canonicalize and
validate both source and destination before any mkdir/rsync: compute a
normalized absolute source (e.g., realpath -m "${payload_dir}/${path}") and
normalized destination (e.g., realpath -m "${restore_root}/${path}" or ensure
path is relative and join with restore root), then ensure the source path starts
with payload_dir canonical prefix and the destination starts with the intended
restore root prefix; if an included entry's source does not exist or validation
fails, log an error and exit non-zero instead of silently skipping, otherwise
proceed with mkdir and rsync only for validated paths.
Comment on lines
+123
to
+125
| ## GitHub Actions Workflow | ||
|
|
||
| Use `.github/workflows/hpmn-backup.yml` for a manual `workflow_dispatch` run. |
There was a problem hiding this comment.
Capitalize GitHub consistently.
These references should use GitHub, not Github.
Also applies to: 192-192
🧰 Tools
🪛 LanguageTool
[uncategorized] ~125-~125: The official name of this software platform is spelled with a capital “H”.
Context: ...es ``` ## GitHub Actions Workflow Use .github/workflows/hpmn-backup.yml for a manual...
(GITHUB)
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@docs/hpmn-backup.md` around lines 123 - 125, The docs contain inconsistent
capitalization of "GitHub" (e.g., the header "GitHub Actions Workflow" and other
occurrences of "Github"); update all instances of "Github" to the correct
"GitHub" spelling throughout the document (including the header and the
occurrence noted around the same section), ensuring consistent casing
everywhere.
Comment on lines
+145
to
+149
| ## Restore | ||
|
|
||
| Restore is intentionally single-host only. Use a replacement or disposable HP masternode, not the whole group. | ||
|
|
||
| For the full tested step-by-step recovery procedure, including public IP cutover, replacement host provisioning, restore, finalize, verification, and troubleshooting, use [`docs/hpmn-recovery.md`](/home/vivek/code/dash-network-deploy/docs/hpmn-recovery.md). |
There was a problem hiding this comment.
Use a repository-relative link here.
Line 149 points at an absolute path from a local workstation, so the link will be broken in the repo and on GitHub. Link to docs/hpmn-recovery.md or ./hpmn-recovery.md instead.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@docs/hpmn-backup.md` around lines 145 - 149, The markdown link currently
points to an absolute local path (`/home/.../docs/hpmn-recovery.md`) which will
break on GitHub; update the link in the Restore section (the
`[`docs/hpmn-recovery.md`](/home/.../docs/hpmn-recovery.md)` instance) to a
repository-relative path such as `docs/hpmn-recovery.md` or `./hpmn-recovery.md`
so it resolves correctly in the repo and on GitHub.
There was a problem hiding this comment.
♻️ Duplicate comments (1)
docs/hpmn-backup.md (1)
147-151:⚠️ Potential issue | 🟡 MinorFix broken link to recovery documentation.
Line 151 contains an absolute local filesystem path (
/home/vivek/code/dash-network-deploy/docs/hpmn-recovery.md) that won't work in the repository or on GitHub. Use a relative path instead.-For the full tested step-by-step recovery procedure, including public IP cutover, replacement host provisioning, restore, finalize, verification, and troubleshooting, use [`docs/hpmn-recovery.md`](/home/vivek/code/dash-network-deploy/docs/hpmn-recovery.md). +For the full tested step-by-step recovery procedure, including public IP cutover, replacement host provisioning, restore, finalize, verification, and troubleshooting, see [`docs/hpmn-recovery.md`](./hpmn-recovery.md).🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@docs/hpmn-backup.md` around lines 147 - 151, The link in the Restore section uses an absolute local path; update the Markdown link target to the repository-relative path for the recovery doc (replace `/home/vivek/code/dash-network-deploy/docs/hpmn-recovery.md` with `docs/hpmn-recovery.md` or `./docs/hpmn-recovery.md`) so the [`docs/hpmn-recovery.md`] link resolves correctly on GitHub; locate the link text "docs/hpmn-recovery.md" in the Restore paragraph and change only the URL portion to the relative path.
🧹 Nitpick comments (1)
ansible/roles/hpmn_backup/templates/backup-hpmn.sh.j2 (1)
137-159: Empty SSE mode case appears unreachable.The empty string case (lines 138-139) won't be hit in practice since
ansible/roles/hpmn_backup/tasks/run.ymlvalidates thathpmn_backup_sse_mode_resolved in ['AES256', 'aws:kms']. The default fromdefaults/main.ymlis alsoAES256. This defense-in-depth is fine, but you could simplify by removing the empty case or explicitly failing on it.♻️ Optional: Fail explicitly on empty SSE mode
case "$sse_mode" in - "") - ;; AES256) aws_args+=(--sse AES256) if [[ -n "$kms_key_id" ]]; then echo "HPMN_BACKUP_S3_KMS_KEY_ID must be empty when SSE mode is AES256" >&2 exit 1 fi ;; aws:kms) aws_args+=(--sse aws:kms) if [[ -z "$kms_key_id" ]]; then echo "HPMN_BACKUP_S3_KMS_KEY_ID is required when SSE mode is aws:kms" >&2 exit 1 fi aws_args+=(--sse-kms-key-id "$kms_key_id") ;; *) echo "Unsupported HPMN_BACKUP_S3_SSE_MODE: $sse_mode" >&2 exit 1 ;; esac🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@ansible/roles/hpmn_backup/templates/backup-hpmn.sh.j2` around lines 137 - 159, The empty string case in the case "$sse_mode" block is unreachable and should be removed or made explicit; update the case handling around the sse_mode switch (the "" branch that currently does nothing) to either delete that branch entirely or replace it with an explicit failure that echoes a clear error (e.g., "HPMN_BACKUP_S3_SSE_MODE must be set") and exits non‑zero, keeping the AES256 and aws:kms branches and the default fallback intact; adjust any references to aws_args if needed so behavior remains consistent.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Duplicate comments:
In `@docs/hpmn-backup.md`:
- Around line 147-151: The link in the Restore section uses an absolute local
path; update the Markdown link target to the repository-relative path for the
recovery doc (replace
`/home/vivek/code/dash-network-deploy/docs/hpmn-recovery.md` with
`docs/hpmn-recovery.md` or `./docs/hpmn-recovery.md`) so the
[`docs/hpmn-recovery.md`] link resolves correctly on GitHub; locate the link
text "docs/hpmn-recovery.md" in the Restore paragraph and change only the URL
portion to the relative path.
---
Nitpick comments:
In `@ansible/roles/hpmn_backup/templates/backup-hpmn.sh.j2`:
- Around line 137-159: The empty string case in the case "$sse_mode" block is
unreachable and should be removed or made explicit; update the case handling
around the sse_mode switch (the "" branch that currently does nothing) to either
delete that branch entirely or replace it with an explicit failure that echoes a
clear error (e.g., "HPMN_BACKUP_S3_SSE_MODE must be set") and exits non‑zero,
keeping the AES256 and aws:kms branches and the default fallback intact; adjust
any references to aws_args if needed so behavior remains consistent.
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: 48cc7e0a-9f1f-41ac-a1ce-78ceef58d6a6
📒 Files selected for processing (5)
ansible/roles/hpmn_backup/defaults/main.ymlansible/roles/hpmn_backup/tasks/main.ymlansible/roles/hpmn_backup/templates/backup-hpmn.sh.j2docs/hpmn-backup.mddocs/hpmn-recovery.md
✅ Files skipped from review due to trivial changes (1)
- docs/hpmn-recovery.md
🚧 Files skipped from review as they are similar to previous changes (1)
- ansible/roles/hpmn_backup/defaults/main.yml
There was a problem hiding this comment.
♻️ Duplicate comments (1)
docs/hpmn-backup.md (1)
154-154:⚠️ Potential issue | 🟡 MinorUse a repository-relative link instead of a local absolute path.
The current link points to a workstation-specific path, so it will be broken in the repo and on GitHub. Please switch it to a repo-relative target (for example
./hpmn-recovery.md).Suggested fix
-For the full tested step-by-step recovery procedure, including public IP cutover, replacement host provisioning, restore, finalize, verification, and troubleshooting, use [`docs/hpmn-recovery.md`](/home/vivek/code/dash-network-deploy/docs/hpmn-recovery.md). +For the full tested step-by-step recovery procedure, including public IP cutover, replacement host provisioning, restore, finalize, verification, and troubleshooting, use [`docs/hpmn-recovery.md`](./hpmn-recovery.md).🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@docs/hpmn-backup.md` at line 154, The markdown link in docs/hpmn-backup.md uses a workstation absolute path (/home/vivek/.../hpmn-recovery.md); change that target to a repository-relative link (for example ./hpmn-recovery.md) so the anchor `[docs/hpmn-recovery.md](/home/vivek/code/dash-network-deploy/docs/hpmn-recovery.md)` becomes a repo-relative reference and will resolve correctly in the repository and on GitHub.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Duplicate comments:
In `@docs/hpmn-backup.md`:
- Line 154: The markdown link in docs/hpmn-backup.md uses a workstation absolute
path (/home/vivek/.../hpmn-recovery.md); change that target to a
repository-relative link (for example ./hpmn-recovery.md) so the anchor
`[docs/hpmn-recovery.md](/home/vivek/code/dash-network-deploy/docs/hpmn-recovery.md)`
becomes a repo-relative reference and will resolve correctly in the repository
and on GitHub.
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: 425867fa-ac67-464c-a679-dd0b9526486e
📒 Files selected for processing (5)
.gitignoreansible/roles/hpmn_restore/defaults/main.ymlansible/roles/hpmn_restore/templates/restore-hpmn.sh.j2docs/hpmn-backup.mddocs/hpmn-recovery.md
✅ Files skipped from review due to trivial changes (2)
- .gitignore
- docs/hpmn-recovery.md
🚧 Files skipped from review as they are similar to previous changes (2)
- ansible/roles/hpmn_restore/templates/restore-hpmn.sh.j2
- ansible/roles/hpmn_restore/defaults/main.yml
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Issue being fixed or feature implemented
Implements the first HP masternode backup workflow for
testnetas follow-up work for infra issue#25.This change is scoped to
hp_masternodesonly and is intended to provide a repeatable backup path for the runtime state that matters most for HPMN recovery, without relying on full-instance snapshots ormodifying the existing
deploy.ymlflow.Related context:
#25#24What was done?
Added a dedicated HPMN backup implementation under Ansible, plus documentation and a GitHub Actions workflow scaffold.
Changes include:
ansible/roles/hpmn_backupansible/hpmn_backup_install.ymlfor one-time backup tooling installation on HPMNsansible/hpmn_backup_run.ymlfor on-demand backup execution.github/workflows/hpmn-backup.ymlforworkflow_dispatch-based triggeringdocs/hpmn-backup.mdThe backup flow is isolated from normal provisioning and does not require using
./bin/deploy -p.The backup currently collects the following runtime state on each HP masternode:
/home/dashmate/.dashmate/config.json/home/dashmate/.dashmate/<network>/platform/drive/tenderdash/home/dashmate/.dashmate/<network>/platform/gateway/ssl/var/lib/docker/volumes/dashmate_<network>_drive_tenderdash/_data/var/lib/docker/volumes/dashmate_<network>_core_data/_data/.dashcore/testnet3/llmqThe backup archive is created locally on the HPMN and uploaded directly to S3 with stamped object paths using:
This implementation is designed to be non-disruptive:
dashmate resetHow Has This Been Tested?
Functional testing was performed against live
testnetinfrastructure onhp-masternode-1.Read-only verification on
hp-masternode-1confirmed the presence of:llmqIt also confirmed that
priv_validator_key.json/priv_validator_state.jsonwere not present on the host or visible in the running container during inspection.Live validation performed:
ansible/hpmn_backup_install.ymlagainsthp-masternode-1hp-masternode-1Validated S3 bucket:
s3://dash-testnet-hpmns-backupsValidated uploaded test object:
s3://dash-testnet-hpmns-backups/hpmn-backups/testnet/hp-masternode-1/20260414T121720Z_second-test.tar.gzObserved result:
AES256server-side encryption_datallmqBreaking Changes
None.
Checklist:
Summary by CodeRabbit
New Features
Documentation
Bug Fixes
Chores