Compare commits
6 Commits
codex/host
...
codex/wind
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
1df830d8cc | ||
|
|
c96e00d6e9 | ||
|
|
8391d8d5a6 | ||
|
|
5c4f22f7da | ||
|
|
68dcc6bbd4 | ||
|
|
56d1d396b8 |
67
docs/proposals/AWOOOI-AG99-H110-BACKUP-RUNTIME-V6.md
Normal file
67
docs/proposals/AWOOOI-AG99-H110-BACKUP-RUNTIME-V6.md
Normal file
@@ -0,0 +1,67 @@
|
||||
# AWOOOI Agent99 Host110 Backup Runtime V6
|
||||
|
||||
Status: proposed; owner approval required before any production apply.
|
||||
|
||||
This artifact supersedes V2 through V5. It is evidence for central review, not
|
||||
an active policy and not production execution authority.
|
||||
|
||||
## Scope
|
||||
|
||||
- Agent99 runtime bundle: 19 files. Deploy, bootstrap, sender, receiver,
|
||||
preflight, Signoz consumer, and contract checks must agree on this count.
|
||||
- Host110 backup payload: 18 files, comprising 17 backup scripts and
|
||||
`backup-health-textfile-exporter.py`.
|
||||
- Host110 remote stage: 21 files, comprising the 18 payload files, the fixed
|
||||
executor, the independent verifier, and the source manifest.
|
||||
- Fixed production target: `wooo@192.168.0.110`, dispatched from Windows99
|
||||
Agent99 with an exact Gitea revision and digest manifest.
|
||||
|
||||
## Exact Diff Serialization
|
||||
|
||||
The reviewer must substitute the V6 Gitea live-ref SHA for
|
||||
`<candidate_git_sha>` and hash these exact stdout bytes:
|
||||
|
||||
```sh
|
||||
git diff --binary --full-index \
|
||||
e7e3bcf8a5a8a19c1253d73f51d9dca248af0aa3..<candidate_git_sha> |
|
||||
shasum -a 256
|
||||
```
|
||||
|
||||
The proposal message must bind all of the following independently:
|
||||
|
||||
- Gitea live ref and exact 40-character candidate Git SHA.
|
||||
- This tracked artifact path and its SHA-256 bytes hash.
|
||||
- The exact diff serialization command above and resulting SHA-256.
|
||||
|
||||
## Apply Effects
|
||||
|
||||
- Agent99 runtime promotion writes the 19-file runtime bundle on Windows99 and
|
||||
writes its bounded deployment manifest and receipts.
|
||||
- Host110 Apply writes/replaces the 18 payload destinations through one
|
||||
exclusive, digest-bound filesystem transaction.
|
||||
- The transaction does not itself run a backup, sync Google Drive, prune or
|
||||
delete snapshots, restart a VM, or change a network, firewall, DB, or service.
|
||||
- The broader source diff also changes the Gitea backup route. A later scheduled
|
||||
or explicit Gitea offline-consistency backup may stop and restart the Gitea
|
||||
container within its bounded backup window. That deferred runtime effect is
|
||||
not part of the Host110 filesystem deployment transaction and requires its
|
||||
own verifier receipt.
|
||||
|
||||
## Validation
|
||||
|
||||
- Post-rebase focused suite: 51 passed, 4 skipped.
|
||||
- Central read-only review: 36 passed, 4 skipped; Host110 broker: 7 passed.
|
||||
- Ansible validation: green.
|
||||
- Changed PowerShell parser readback on Windows99: zero errors.
|
||||
- Prior broad regression baseline: 552 passed, 4 skipped.
|
||||
|
||||
## Rollback
|
||||
|
||||
- Source rollback reverts the exact V6 candidate diff to
|
||||
`e7e3bcf8a5a8a19c1253d73f51d9dca248af0aa3` without force-pushing.
|
||||
- Agent99 rollback restores the complete prior runtime bundle and prior runtime
|
||||
manifest, then independently verifies prior hashes.
|
||||
- Host110 rollback restores every prior payload's content, SHA-256, uid, gid,
|
||||
and mode under the exclusive lock.
|
||||
- Any mismatch terminates as `rollback_unverified`; it cannot be reported as a
|
||||
successful rollback.
|
||||
@@ -1,65 +0,0 @@
|
||||
# AWOOOI Agent99 Host110 Backup Runtime V7
|
||||
|
||||
Status: proposed; owner approval required before any production apply.
|
||||
|
||||
This artifact supersedes V2 through V6. All V5 proposal, artifact, and diff
|
||||
hashes are void. This file is review evidence only; it is not active policy or
|
||||
production execution authority.
|
||||
|
||||
## Scope
|
||||
|
||||
- Agent99 runtime bundle: 19 files. Every producer and consumer is guarded by
|
||||
one exact parity test.
|
||||
- Host110 backup payload: 18 files, comprising 17 backup scripts and
|
||||
`backup-health-textfile-exporter.py`.
|
||||
- Host110 remote stage: 21 files, comprising the 18 payload files, the fixed
|
||||
executor, the independent verifier, and the source manifest.
|
||||
- Fixed target: `wooo@192.168.0.110`, dispatched by Windows99 Agent99 from an
|
||||
exact Gitea live revision and digest manifest.
|
||||
- Gitea offline backup and container lifecycle changes are excluded from this
|
||||
proposal. `backup-gitea.sh` has no candidate diff. The only Gitea-related
|
||||
diff is a shared deployment lock in the read-only restore-drill entrypoint;
|
||||
it adds no Docker command and is not executed by the deployment transaction.
|
||||
|
||||
## Exact Diff Serialization
|
||||
|
||||
Substitute the V7 Gitea live-ref SHA for `<candidate_git_sha>` and hash the
|
||||
exact stdout bytes from this command:
|
||||
|
||||
```sh
|
||||
git diff --binary --full-index \
|
||||
e7e3bcf8a5a8a19c1253d73f51d9dca248af0aa3..<candidate_git_sha> |
|
||||
shasum -a 256
|
||||
```
|
||||
|
||||
The review message independently binds the Gitea live ref and Git SHA, this
|
||||
tracked artifact path and bytes hash, and the serialized diff hash.
|
||||
|
||||
## Apply Effects
|
||||
|
||||
- Windows99 Agent99 runtime promotion writes the 19 runtime files plus bounded
|
||||
manifest and receipt state.
|
||||
- Host110 Apply writes/replaces the 18 payload paths through one exclusive,
|
||||
digest-bound filesystem transaction.
|
||||
- Apply does not run a backup, invoke a Gitea container lifecycle, sync Google
|
||||
Drive, prune or delete snapshots, restart a VM or service, or change a DB,
|
||||
network, or firewall.
|
||||
|
||||
## Validation
|
||||
|
||||
- Post-split focused runtime, broker, backup, and parity suites must pass.
|
||||
- Ansible and shell/Python static validation must pass.
|
||||
- Changed PowerShell must parse on Windows99 with zero errors.
|
||||
- Central review must independently reproduce the live ref, artifact hash,
|
||||
diff hash, and 19/18/21 counts.
|
||||
|
||||
## Rollback
|
||||
|
||||
- Source rollback reverts the exact V7 diff to
|
||||
`e7e3bcf8a5a8a19c1253d73f51d9dca248af0aa3` without force-pushing.
|
||||
- Agent99 rollback restores the prior runtime files and manifest and verifies
|
||||
the prior hashes independently.
|
||||
- Host110 rollback restores every prior payload's content, SHA-256, uid, gid,
|
||||
and mode under the exclusive lock.
|
||||
- Any mismatch terminates as `rollback_unverified` and cannot be reported as a
|
||||
successful rollback.
|
||||
@@ -17,6 +17,9 @@
|
||||
- backup-host188-products.sh
|
||||
- verify-host188-products-backup.sh
|
||||
- verify-host188-products-archive.py
|
||||
- backup-gitea.sh
|
||||
- gitea-full-backup-restore-drill.sh
|
||||
- backup-configs.sh
|
||||
backup_runtime_stage2_files:
|
||||
- backup-awoooi.sh
|
||||
- backup-awoooi-frequent.sh
|
||||
@@ -35,7 +38,7 @@
|
||||
ansible.builtin.assert:
|
||||
that:
|
||||
- inventory_hostname == "host_110"
|
||||
- backup_runtime_files | length == 14
|
||||
- backup_runtime_files | length == 17
|
||||
- backup_runtime_files | unique | length == backup_runtime_files | length
|
||||
- backup_runtime_fault_injection_after_stage1 | type_debug == "bool"
|
||||
fail_msg: backup_runtime_deploy_boundary_failed
|
||||
|
||||
@@ -224,6 +224,7 @@
|
||||
- backup-all.sh
|
||||
- backup-status.sh
|
||||
- backup-gitea.sh
|
||||
- gitea-full-backup-restore-drill.sh
|
||||
- gitea-repo-bundle-backup.sh
|
||||
- gitea-bundle-backup-sync-188.sh
|
||||
- backup-harbor.sh
|
||||
|
||||
@@ -201,6 +201,13 @@ main() {
|
||||
write_cmd_output "110-systemd-unit-files" systemctl list-unit-files || failed=$((failed + 1))
|
||||
write_cmd_output "110-docker-containers" docker ps --format 'table {{.Names}}\t{{.Image}}\t{{.Status}}\t{{.Ports}}' || true
|
||||
|
||||
if [ -r /home/wooo/vibework-act-runner/config.yaml ]; then
|
||||
record_config_status "110-gitea-runner-config" true true "110"
|
||||
else
|
||||
record_config_status "110-gitea-runner-config" true false "110"
|
||||
failed=$((failed + 1))
|
||||
fi
|
||||
|
||||
if tar_local "110-host-configs" \
|
||||
/etc/nginx \
|
||||
/etc/systemd/system \
|
||||
@@ -224,6 +231,7 @@ main() {
|
||||
/home/wooo/scripts \
|
||||
/home/wooo/awoooi \
|
||||
/home/wooo/awoooi-ops \
|
||||
/home/wooo/vibework-act-runner/config.yaml \
|
||||
/backup/scripts; then
|
||||
record_config_status "110-host-configs" true true "110"
|
||||
else
|
||||
|
||||
@@ -7,6 +7,7 @@
|
||||
# =============================================================================
|
||||
|
||||
set -euo pipefail
|
||||
umask 077
|
||||
|
||||
source "$(dirname "$0")/common.sh"
|
||||
|
||||
@@ -17,7 +18,22 @@ DUMP_DIR="/tmp/gitea-backup-$$"
|
||||
TEXTFILE_DIR="${NODE_EXPORTER_TEXTFILE_DIR:-/home/wooo/node_exporter_textfiles}"
|
||||
TEXTFILE_PATH="${GITEA_FULL_BACKUP_RECEIPT_TEXTFILE:-${TEXTFILE_DIR}/gitea_full_backup_receipt.prom}"
|
||||
HOST_LABEL="${AIOPS_HOST_LABEL:-110}"
|
||||
GITEA_FULL_BACKUP_COMPONENTS="database repositories app_settings lfs_objects package_registry attachments hooks_custom_assets"
|
||||
GITEA_FULL_BACKUP_COMPONENTS="database repositories app_settings lfs_objects package_registry attachments actions_logs actions_artifacts avatars hooks_custom_assets"
|
||||
GITEA_RESTORE_DRILL="$(dirname "$0")/gitea-full-backup-restore-drill.sh"
|
||||
GITEA_BACKUP_LOCK="${GITEA_BACKUP_LOCK:-/tmp/gitea-backup.lock}"
|
||||
GITEA_STOP_TIMEOUT_SECONDS="${GITEA_STOP_TIMEOUT_SECONDS:-15}"
|
||||
GITEA_DUMP_TIMEOUT_SECONDS="${GITEA_DUMP_TIMEOUT_SECONDS:-360}"
|
||||
GITEA_DUMP_KILL_AFTER_SECONDS="${GITEA_DUMP_KILL_AFTER_SECONDS:-10}"
|
||||
GITEA_DOCKER_CONTROL_TIMEOUT_SECONDS="${GITEA_DOCKER_CONTROL_TIMEOUT_SECONDS:-10}"
|
||||
GITEA_DOCKER_CONTROL_KILL_AFTER_SECONDS="${GITEA_DOCKER_CONTROL_KILL_AFTER_SECONDS:-3}"
|
||||
GITEA_START_TIMEOUT_SECONDS="${GITEA_START_TIMEOUT_SECONDS:-60}"
|
||||
GITEA_MAX_OUTAGE_SECONDS="${GITEA_MAX_OUTAGE_SECONDS:-600}"
|
||||
GITEA_LOCAL_HEALTH_URL="${GITEA_LOCAL_HEALTH_URL:-http://127.0.0.1:3001/api/healthz}"
|
||||
TRANSIENT_CONTAINER=""
|
||||
PRIMARY_STOPPED=0
|
||||
PRIMARY_RESTART_VERIFIED=0
|
||||
OUTAGE_STARTED_AT=0
|
||||
OUTAGE_SECONDS=0
|
||||
|
||||
label_escape() {
|
||||
printf '%s' "$1" | sed -e 's/\\/\\\\/g' -e 's/"/\\"/g'
|
||||
@@ -28,6 +44,10 @@ write_gitea_full_backup_receipt() {
|
||||
local timestamp="$2"
|
||||
local duration="$3"
|
||||
local snapshot_id="${4:-unknown}"
|
||||
local consistent_offline_window="${5:-0}"
|
||||
local service_restart_verified="${6:-0}"
|
||||
local outage_seconds="${7:-0}"
|
||||
local structural_drill_performed="${8:-0}"
|
||||
local tmp
|
||||
local host
|
||||
local service_label
|
||||
@@ -66,6 +86,12 @@ write_gitea_full_backup_receipt() {
|
||||
echo "# TYPE awoooi_gitea_full_backup_secret_value_collected gauge"
|
||||
echo "# HELP awoooi_gitea_full_backup_production_restore_performed Whether this receipt restored anything into production."
|
||||
echo "# TYPE awoooi_gitea_full_backup_production_restore_performed gauge"
|
||||
echo "# HELP awoooi_gitea_full_backup_consistent_offline_window Whether the dump was created while the primary Gitea service was stopped."
|
||||
echo "# TYPE awoooi_gitea_full_backup_consistent_offline_window gauge"
|
||||
echo "# HELP awoooi_gitea_full_backup_service_restart_verified Whether Gitea was independently reachable after the bounded offline window."
|
||||
echo "# TYPE awoooi_gitea_full_backup_service_restart_verified gauge"
|
||||
echo "# HELP awoooi_gitea_full_backup_service_outage_seconds Duration of the bounded Gitea offline backup window."
|
||||
echo "# TYPE awoooi_gitea_full_backup_service_outage_seconds gauge"
|
||||
echo "awoooi_gitea_full_backup_last_success_timestamp{host=\"${host}\",service=\"${service_label}\"} $([ "${ok}" = "1" ] && echo "${timestamp}" || echo 0)"
|
||||
echo "awoooi_gitea_full_backup_last_run_failed{host=\"${host}\",service=\"${service_label}\",status=\"${status_label}\"} ${failed}"
|
||||
echo "awoooi_gitea_full_backup_duration_seconds{host=\"${host}\",service=\"${service_label}\"} ${duration}"
|
||||
@@ -74,16 +100,231 @@ write_gitea_full_backup_receipt() {
|
||||
for component in ${GITEA_FULL_BACKUP_COMPONENTS}; do
|
||||
echo "awoooi_gitea_full_backup_component_receipt{host=\"${host}\",service=\"${service_label}\",component=\"$(label_escape "${component}")\",receipt=\"gitea_dump_restic\"} ${ok}"
|
||||
done
|
||||
echo "awoooi_gitea_full_backup_restore_drill_performed{host=\"${host}\",service=\"${service_label}\"} 0"
|
||||
echo "awoooi_gitea_full_backup_restore_drill_performed{host=\"${host}\",service=\"${service_label}\"} ${structural_drill_performed}"
|
||||
echo "awoooi_gitea_full_backup_secret_value_collected{host=\"${host}\",service=\"${service_label}\"} 0"
|
||||
echo "awoooi_gitea_full_backup_production_restore_performed{host=\"${host}\",service=\"${service_label}\"} 0"
|
||||
echo "awoooi_gitea_full_backup_consistent_offline_window{host=\"${host}\",service=\"${service_label}\"} ${consistent_offline_window}"
|
||||
echo "awoooi_gitea_full_backup_service_restart_verified{host=\"${host}\",service=\"${service_label}\"} ${service_restart_verified}"
|
||||
echo "awoooi_gitea_full_backup_service_outage_seconds{host=\"${host}\",service=\"${service_label}\"} ${outage_seconds}"
|
||||
} >"${tmp}"
|
||||
mv "${tmp}" "${TEXTFILE_PATH}" || return 0
|
||||
chmod 0644 "${TEXTFILE_PATH}" || true
|
||||
}
|
||||
|
||||
cleanup() {
|
||||
local status=$?
|
||||
trap - EXIT HUP INT TERM
|
||||
if [ -n "${TRANSIENT_CONTAINER}" ]; then
|
||||
if ! remove_transient_container; then
|
||||
status=1
|
||||
fi
|
||||
fi
|
||||
if [ "${PRIMARY_STOPPED}" -eq 1 ]; then
|
||||
if ! start_and_verify_primary; then
|
||||
status=1
|
||||
fi
|
||||
fi
|
||||
rm -rf "${DUMP_DIR}"
|
||||
exit "${status}"
|
||||
}
|
||||
|
||||
validate_positive_integer() {
|
||||
case "$2" in
|
||||
''|*[!0-9]*|0) log_error "$1 must be a positive integer"; return 64 ;;
|
||||
esac
|
||||
}
|
||||
|
||||
require_commands() {
|
||||
local command_name configured_outage_bound
|
||||
for command_name in curl docker flock python3 restic timeout; do
|
||||
if ! command -v "${command_name}" >/dev/null 2>&1; then
|
||||
log_error "Required command missing: ${command_name}"
|
||||
return 69
|
||||
fi
|
||||
done
|
||||
[ -x "${GITEA_RESTORE_DRILL}" ] || {
|
||||
log_error "Gitea structural restore verifier is unavailable"
|
||||
return 69
|
||||
}
|
||||
[ -r "${RESTIC_PASSWORD_FILE}" ] || {
|
||||
log_error "Restic password reference is unavailable"
|
||||
return 69
|
||||
}
|
||||
validate_positive_integer GITEA_STOP_TIMEOUT_SECONDS "${GITEA_STOP_TIMEOUT_SECONDS}"
|
||||
validate_positive_integer GITEA_DUMP_TIMEOUT_SECONDS "${GITEA_DUMP_TIMEOUT_SECONDS}"
|
||||
validate_positive_integer GITEA_DUMP_KILL_AFTER_SECONDS "${GITEA_DUMP_KILL_AFTER_SECONDS}"
|
||||
validate_positive_integer GITEA_DOCKER_CONTROL_TIMEOUT_SECONDS "${GITEA_DOCKER_CONTROL_TIMEOUT_SECONDS}"
|
||||
validate_positive_integer GITEA_DOCKER_CONTROL_KILL_AFTER_SECONDS "${GITEA_DOCKER_CONTROL_KILL_AFTER_SECONDS}"
|
||||
validate_positive_integer GITEA_START_TIMEOUT_SECONDS "${GITEA_START_TIMEOUT_SECONDS}"
|
||||
validate_positive_integer GITEA_MAX_OUTAGE_SECONDS "${GITEA_MAX_OUTAGE_SECONDS}"
|
||||
|
||||
configured_outage_bound=$((
|
||||
GITEA_STOP_TIMEOUT_SECONDS
|
||||
+ GITEA_DUMP_TIMEOUT_SECONDS
|
||||
+ GITEA_DUMP_KILL_AFTER_SECONDS
|
||||
+ GITEA_START_TIMEOUT_SECONDS
|
||||
+ (4 * GITEA_DOCKER_CONTROL_TIMEOUT_SECONDS)
|
||||
+ (4 * GITEA_DOCKER_CONTROL_KILL_AFTER_SECONDS)
|
||||
))
|
||||
if [ "${configured_outage_bound}" -gt "${GITEA_MAX_OUTAGE_SECONDS}" ]; then
|
||||
log_error "Configured Gitea backup outage bound exceeds ${GITEA_MAX_OUTAGE_SECONDS}s"
|
||||
return 64
|
||||
fi
|
||||
}
|
||||
|
||||
bounded_docker_control() {
|
||||
timeout --signal=TERM \
|
||||
--kill-after="${GITEA_DOCKER_CONTROL_KILL_AFTER_SECONDS}s" \
|
||||
"${GITEA_DOCKER_CONTROL_TIMEOUT_SECONDS}s" \
|
||||
docker "$@"
|
||||
}
|
||||
|
||||
remove_transient_container() {
|
||||
local transient="${TRANSIENT_CONTAINER}"
|
||||
local remove_status=0
|
||||
|
||||
[ -n "${transient}" ] || return 0
|
||||
if ! bounded_docker_control rm -f "${transient}" >/dev/null 2>&1; then
|
||||
log_error "Gitea one-shot backup container removal timed out or failed"
|
||||
remove_status=1
|
||||
fi
|
||||
TRANSIENT_CONTAINER=""
|
||||
return "${remove_status}"
|
||||
}
|
||||
|
||||
wait_for_primary_health() {
|
||||
local deadline=$((SECONDS + GITEA_START_TIMEOUT_SECONDS))
|
||||
local health_body
|
||||
while [ "${SECONDS}" -lt "${deadline}" ]; do
|
||||
if [ "$(bounded_docker_control inspect -f '{{.State.Running}}' "${GITEA_CONTAINER}" 2>/dev/null || true)" = "true" ]; then
|
||||
health_body="$(curl -fsS --max-time 5 "${GITEA_LOCAL_HEALTH_URL}" 2>/dev/null || true)"
|
||||
if printf '%s' "${health_body}" | grep -Eq '"status"[[:space:]]*:[[:space:]]*"pass"'; then
|
||||
return 0
|
||||
fi
|
||||
fi
|
||||
sleep 2
|
||||
done
|
||||
return 1
|
||||
}
|
||||
|
||||
start_and_verify_primary() {
|
||||
local running
|
||||
|
||||
running="$(bounded_docker_control inspect -f '{{.State.Running}}' "${GITEA_CONTAINER}" 2>/dev/null || true)"
|
||||
if [ "${running}" != "true" ]; then
|
||||
if ! bounded_docker_control start "${GITEA_CONTAINER}" >/dev/null; then
|
||||
log_error "Gitea failed to restart after the backup window"
|
||||
return 1
|
||||
fi
|
||||
fi
|
||||
if ! wait_for_primary_health; then
|
||||
log_error "Gitea post-backup health verification timed out"
|
||||
return 1
|
||||
fi
|
||||
PRIMARY_STOPPED=0
|
||||
PRIMARY_RESTART_VERIFIED=1
|
||||
if [ "${OUTAGE_STARTED_AT}" -gt 0 ]; then
|
||||
OUTAGE_SECONDS=$(($(date +%s) - OUTAGE_STARTED_AT))
|
||||
fi
|
||||
if [ "${OUTAGE_SECONDS}" -gt "${GITEA_MAX_OUTAGE_SECONDS}" ]; then
|
||||
log_error "Gitea backup outage exceeded ${GITEA_MAX_OUTAGE_SECONDS}s"
|
||||
return 1
|
||||
fi
|
||||
log_success "Gitea primary restarted and independently reachable (outage=${OUTAGE_SECONDS}s)"
|
||||
}
|
||||
|
||||
offline_gitea_dump() {
|
||||
local image_id network_rows network_name network_count git_uid host_uid timestamp dump_status stop_status transient_cleanup_status
|
||||
|
||||
[ "$(docker inspect -f '{{.State.Running}}' "${GITEA_CONTAINER}" 2>/dev/null || true)" = "true" ] || {
|
||||
log_error "Gitea primary is not running before backup"
|
||||
return 1
|
||||
}
|
||||
wait_for_primary_health || {
|
||||
log_error "Gitea primary is not reachable before backup"
|
||||
return 1
|
||||
}
|
||||
|
||||
image_id="$(docker inspect -f '{{.Image}}' "${GITEA_CONTAINER}")"
|
||||
[[ "${image_id}" =~ ^sha256:[0-9a-f]{64}$ ]] || {
|
||||
log_error "Gitea image identity is invalid"
|
||||
return 1
|
||||
}
|
||||
network_rows="$(docker inspect -f '{{range $name, $value := .NetworkSettings.Networks}}{{$name}}{{"\n"}}{{end}}' "${GITEA_CONTAINER}" | sed '/^$/d')"
|
||||
network_count="$(printf '%s\n' "${network_rows}" | sed '/^$/d' | wc -l | tr -d ' ')"
|
||||
[ "${network_count}" -eq 1 ] || {
|
||||
log_error "Gitea backup requires exactly one fixed Docker network"
|
||||
return 1
|
||||
}
|
||||
network_name="$(printf '%s\n' "${network_rows}" | head -n 1)"
|
||||
[[ "${network_name}" =~ ^[A-Za-z0-9_.-]+$ ]] || {
|
||||
log_error "Gitea Docker network identity is invalid"
|
||||
return 1
|
||||
}
|
||||
git_uid="$(docker exec -u git "${GITEA_CONTAINER}" id -u)"
|
||||
host_uid="$(id -u)"
|
||||
[ "${git_uid}" = "${host_uid}" ] || {
|
||||
log_error "Gitea dump staging UID does not match the controller UID"
|
||||
return 1
|
||||
}
|
||||
|
||||
timestamp="$(date +%Y%m%d%H%M%S)"
|
||||
TRANSIENT_CONTAINER="agent99-gitea-backup-${timestamp}-$$"
|
||||
OUTAGE_STARTED_AT="$(date +%s)"
|
||||
PRIMARY_STOPPED=1
|
||||
set +e
|
||||
timeout --signal=TERM \
|
||||
--kill-after="${GITEA_DOCKER_CONTROL_KILL_AFTER_SECONDS}s" \
|
||||
"$((GITEA_STOP_TIMEOUT_SECONDS + GITEA_DOCKER_CONTROL_TIMEOUT_SECONDS))s" \
|
||||
docker stop --time "${GITEA_STOP_TIMEOUT_SECONDS}" "${GITEA_CONTAINER}" >/dev/null
|
||||
stop_status=$?
|
||||
set -e
|
||||
if [ "${stop_status}" -ne 0 ]; then
|
||||
log_error "Gitea primary could not enter the bounded offline backup window"
|
||||
start_and_verify_primary || true
|
||||
return 1
|
||||
fi
|
||||
|
||||
set +e
|
||||
timeout --signal=TERM \
|
||||
--kill-after="${GITEA_DUMP_KILL_AFTER_SECONDS}s" \
|
||||
"${GITEA_DUMP_TIMEOUT_SECONDS}s" docker run --pull=never \
|
||||
--name "${TRANSIENT_CONTAINER}" \
|
||||
--network "${network_name}" \
|
||||
--volumes-from "${GITEA_CONTAINER}" \
|
||||
--mount "type=bind,src=${DUMP_DIR},dst=/backup-out" \
|
||||
--user "${git_uid}" \
|
||||
--workdir /backup-out \
|
||||
--cpus "${BACKUP_DOCKER_CPUS}" \
|
||||
--memory "${BACKUP_DOCKER_MEMORY}" \
|
||||
--memory-swap "${BACKUP_DOCKER_MEMORY_SWAP}" \
|
||||
--security-opt no-new-privileges:true \
|
||||
--entrypoint /usr/local/bin/gitea \
|
||||
"${image_id}" dump \
|
||||
-c /data/gitea/conf/app.ini \
|
||||
-f /backup-out/gitea-dump.zip \
|
||||
-t /tmp
|
||||
dump_status=$?
|
||||
set -e
|
||||
|
||||
transient_cleanup_status=0
|
||||
remove_transient_container || transient_cleanup_status=$?
|
||||
|
||||
if ! start_and_verify_primary; then
|
||||
return 1
|
||||
fi
|
||||
if [ "${transient_cleanup_status}" -ne 0 ]; then
|
||||
return "${transient_cleanup_status}"
|
||||
fi
|
||||
if [ "${dump_status}" -ne 0 ]; then
|
||||
log_error "Offline Gitea dump failed (exit=${dump_status})"
|
||||
return "${dump_status}"
|
||||
fi
|
||||
[ -s "${DUMP_DIR}/gitea-dump.zip" ] || {
|
||||
log_error "Offline Gitea dump did not produce a non-empty archive"
|
||||
return 1
|
||||
}
|
||||
log_success "Gitea consistent offline dump completed"
|
||||
}
|
||||
|
||||
main() {
|
||||
@@ -91,31 +332,44 @@ main() {
|
||||
local tags
|
||||
local snapshot_id
|
||||
local duration
|
||||
local structural_drill_performed=0
|
||||
|
||||
start_time=$(date +%s)
|
||||
trap cleanup EXIT
|
||||
trap 'exit 130' INT
|
||||
trap 'exit 143' TERM HUP
|
||||
|
||||
log_info "========== 開始 Gitea 備份 =========="
|
||||
mkdir -p "${DUMP_DIR}"
|
||||
|
||||
log_info "執行 Gitea dump..."
|
||||
if docker exec -u git "${GITEA_CONTAINER}" gitea dump -c /data/gitea/conf/app.ini -f /tmp/gitea-dump.zip 2>&1; then
|
||||
docker cp "${GITEA_CONTAINER}:/tmp/gitea-dump.zip" "${DUMP_DIR}/gitea-dump.zip"
|
||||
docker exec -u git "${GITEA_CONTAINER}" rm -f /tmp/gitea-dump.zip
|
||||
log_success "Gitea dump 完成"
|
||||
else
|
||||
duration=$(($(date +%s) - start_time))
|
||||
write_gitea_full_backup_receipt "dump_failed" "0" "${duration}" "none" || true
|
||||
log_error "Gitea dump 失敗"
|
||||
notify_clawbot "failed" "${SERVICE}" "Gitea dump 失敗"
|
||||
exit 1
|
||||
require_commands
|
||||
exec 9>"${GITEA_BACKUP_LOCK}"
|
||||
if ! flock -n 9; then
|
||||
log_error "Gitea backup is already running"
|
||||
return 75
|
||||
fi
|
||||
install -d -m 700 "${DUMP_DIR}"
|
||||
|
||||
log_info "執行一致性 Gitea offline dump..."
|
||||
if ! offline_gitea_dump; then
|
||||
duration=$(($(date +%s) - start_time))
|
||||
write_gitea_full_backup_receipt "offline_dump_failed" "0" "${duration}" "none" 0 "${PRIMARY_RESTART_VERIFIED}" "${OUTAGE_SECONDS}" 0 || true
|
||||
notify_clawbot "failed" "${SERVICE}" "Gitea 一致性備份失敗;服務重啟驗證=${PRIMARY_RESTART_VERIFIED}" "${duration}"
|
||||
return 1
|
||||
fi
|
||||
|
||||
if ! "${GITEA_RESTORE_DRILL}" --dump-zip "${DUMP_DIR}/gitea-dump.zip"; then
|
||||
duration=$(($(date +%s) - start_time))
|
||||
write_gitea_full_backup_receipt "structural_drill_failed" "0" "${duration}" "none" 1 "${PRIMARY_RESTART_VERIFIED}" "${OUTAGE_SECONDS}" 0 || true
|
||||
log_error "Gitea dump structural verifier failed"
|
||||
notify_clawbot "failed" "${SERVICE}" "Gitea 備份結構驗證失敗" "${duration}"
|
||||
return 1
|
||||
fi
|
||||
structural_drill_performed=1
|
||||
|
||||
if [ ! -d "${LOCAL_REPO}/data" ]; then
|
||||
log_info "初始化本地 Restic 倉庫..."
|
||||
if ! restic -r "${LOCAL_REPO}" init --password-file "${RESTIC_PASSWORD_FILE}"; then
|
||||
duration=$(($(date +%s) - start_time))
|
||||
write_gitea_full_backup_receipt "restic_init_failed" "0" "${duration}" "none" || true
|
||||
write_gitea_full_backup_receipt "restic_init_failed" "0" "${duration}" "none" 1 "${PRIMARY_RESTART_VERIFIED}" "${OUTAGE_SECONDS}" "${structural_drill_performed}" || true
|
||||
log_error "Restic 初始化失敗"
|
||||
notify_clawbot "failed" "${SERVICE}" "Gitea Restic 初始化失敗"
|
||||
exit 1
|
||||
@@ -127,7 +381,7 @@ main() {
|
||||
--password-file "${RESTIC_PASSWORD_FILE}" \
|
||||
${tags}; then
|
||||
duration=$(($(date +%s) - start_time))
|
||||
write_gitea_full_backup_receipt "restic_backup_failed" "0" "${duration}" "none" || true
|
||||
write_gitea_full_backup_receipt "restic_backup_failed" "0" "${duration}" "none" 1 "${PRIMARY_RESTART_VERIFIED}" "${OUTAGE_SECONDS}" "${structural_drill_performed}" || true
|
||||
log_error "Restic 備份失敗"
|
||||
notify_clawbot "failed" "${SERVICE}" "Gitea Restic 備份失敗"
|
||||
exit 1
|
||||
@@ -142,7 +396,7 @@ main() {
|
||||
log_info "Offsite copy is handled by sync-offsite-backups.sh; no direct rclone sync here."
|
||||
|
||||
duration=$(($(date +%s) - start_time))
|
||||
write_gitea_full_backup_receipt "success" "$(date +%s)" "${duration}" "${snapshot_id:-unknown}" || true
|
||||
write_gitea_full_backup_receipt "success" "$(date +%s)" "${duration}" "${snapshot_id:-unknown}" 1 "${PRIMARY_RESTART_VERIFIED}" "${OUTAGE_SECONDS}" "${structural_drill_performed}" || true
|
||||
log_success "========== Gitea 備份完成 (${duration}s) =========="
|
||||
notify_clawbot "success" "${SERVICE}" "Gitea 備份完成" "${duration}"
|
||||
}
|
||||
|
||||
@@ -19,9 +19,9 @@ DUMP_ZIP="${AIOPS_GITEA_FULL_BACKUP_DRILL_DUMP_ZIP:-}"
|
||||
HOST_LABEL="${AIOPS_HOST_LABEL:-110}"
|
||||
TEXTFILE_PATH="${AIOPS_GITEA_FULL_BACKUP_RESTORE_DRILL_TEXTFILE:-/home/wooo/node_exporter_textfiles/gitea_full_backup_restore_drill.prom}"
|
||||
WRITE_TEXTFILE=0
|
||||
TIMEOUT_SECONDS="${TIMEOUT_SECONDS:-60}"
|
||||
TIMEOUT_SECONDS="${TIMEOUT_SECONDS:-300}"
|
||||
DRILL_SCOPE="structural_metadata_only"
|
||||
REQUIRED_COMPONENTS="database repositories app_settings lfs_objects package_registry attachments hooks_custom_assets"
|
||||
REQUIRED_COMPONENTS="database repositories app_settings lfs_objects package_registry attachments actions_logs actions_artifacts avatars hooks_custom_assets"
|
||||
|
||||
usage() {
|
||||
cat <<'USAGE'
|
||||
@@ -98,35 +98,50 @@ required = sys.argv[2:]
|
||||
|
||||
try:
|
||||
with zipfile.ZipFile(dump_zip) as archive:
|
||||
names = archive.namelist()
|
||||
bad_member = archive.testzip()
|
||||
infos = archive.infolist()
|
||||
if bad_member is not None:
|
||||
print("ERROR=crc_failed")
|
||||
sys.exit(1)
|
||||
except zipfile.BadZipFile:
|
||||
print("ERROR=bad_zip")
|
||||
sys.exit(1)
|
||||
except OSError:
|
||||
except (OSError, RuntimeError):
|
||||
print("ERROR=zip_open_failed")
|
||||
sys.exit(1)
|
||||
|
||||
lowered = [name.lower() for name in names]
|
||||
members = [
|
||||
(info.filename.lower().lstrip("./"), info.file_size)
|
||||
for info in infos
|
||||
if not info.is_dir()
|
||||
]
|
||||
|
||||
patterns = {
|
||||
"database": ("gitea-db", "database", ".sql"),
|
||||
"repositories": ("repo", "repos", "repositories", ".git"),
|
||||
"app_settings": ("app.ini", "app.example.ini", "conf/"),
|
||||
"lfs_objects": ("lfs",),
|
||||
"package_registry": ("package", "packages"),
|
||||
"attachments": ("attachment", "attachments"),
|
||||
"hooks_custom_assets": ("hook", "custom", "public/"),
|
||||
|
||||
def has_nonempty(*needles: str) -> bool:
|
||||
return any(size > 0 and any(needle in name for needle in needles) for name, size in members)
|
||||
|
||||
|
||||
data_bundle = has_nonempty("gitea-data.zip")
|
||||
custom_bundle = has_nonempty("gitea-custom.zip")
|
||||
|
||||
coverage = {
|
||||
"database": has_nonempty("gitea-db", "database", ".sql"),
|
||||
"repositories": has_nonempty("gitea-repo.zip", "gitea-repos.zip", "repos/", "repositories/", ".git/"),
|
||||
"app_settings": has_nonempty("app.ini", "app.example.ini"),
|
||||
"lfs_objects": data_bundle or has_nonempty("data/lfs/", "/lfs/"),
|
||||
"package_registry": data_bundle or has_nonempty("data/packages/", "/packages/"),
|
||||
"attachments": data_bundle or has_nonempty("data/attachments/", "/attachments/"),
|
||||
"actions_logs": data_bundle or has_nonempty("data/actions_log/", "/actions_log/"),
|
||||
"actions_artifacts": data_bundle or has_nonempty("data/actions_artifacts/", "/actions_artifacts/"),
|
||||
"avatars": data_bundle or has_nonempty("data/avatars/", "data/repo-avatars/", "/avatars/", "/repo-avatars/"),
|
||||
"hooks_custom_assets": custom_bundle or has_nonempty("custom/", "/hooks/", "public/"),
|
||||
}
|
||||
|
||||
missing = []
|
||||
for component in required:
|
||||
needles = patterns.get(component, (component,))
|
||||
if not any(any(needle in name for needle in needles) for name in lowered):
|
||||
missing.append(component)
|
||||
missing = [component for component in required if not coverage.get(component, False)]
|
||||
|
||||
print(f"ENTRY_COUNT={len(names)}")
|
||||
print(f"ENTRY_COUNT={len(infos)}")
|
||||
print(f"MISSING_COMPONENTS={','.join(missing)}")
|
||||
print("OK=1" if not missing and names else "OK=0")
|
||||
print("OK=1" if not missing and infos else "OK=0")
|
||||
PY
|
||||
)"
|
||||
drill_status=$?
|
||||
|
||||
@@ -18,6 +18,9 @@ EXPECTED_FILES = {
|
||||
"backup-host188-products.sh",
|
||||
"verify-host188-products-backup.sh",
|
||||
"verify-host188-products-archive.py",
|
||||
"backup-gitea.sh",
|
||||
"gitea-full-backup-restore-drill.sh",
|
||||
"backup-configs.sh",
|
||||
"check-backup-integrity.sh",
|
||||
"sync-offsite-backups.sh",
|
||||
"backup-offsite-readiness-gate.sh",
|
||||
@@ -26,7 +29,7 @@ EXPECTED_FILES = {
|
||||
}
|
||||
|
||||
|
||||
def test_backup_runtime_deploy_is_fixed_to_host_110_and_fourteen_files() -> None:
|
||||
def test_backup_runtime_deploy_is_fixed_to_host_110_and_seventeen_files() -> None:
|
||||
plays = yaml.safe_load(PLAYBOOK.read_text(encoding="utf-8"))
|
||||
assert len(plays) == 1
|
||||
play = plays[0]
|
||||
|
||||
@@ -5,6 +5,7 @@ from pathlib import Path
|
||||
|
||||
ROOT = Path(__file__).resolve().parents[3]
|
||||
BACKUP_GITEA = ROOT / "scripts" / "backup" / "backup-gitea.sh"
|
||||
BACKUP_CONFIGS = ROOT / "scripts" / "backup" / "backup-configs.sh"
|
||||
|
||||
|
||||
def test_backup_gitea_writes_full_server_component_receipts() -> None:
|
||||
@@ -15,7 +16,12 @@ def test_backup_gitea_writes_full_server_component_receipts() -> None:
|
||||
assert "awoooi_gitea_full_backup_restore_drill_performed" in text
|
||||
assert "awoooi_gitea_full_backup_secret_value_collected" in text
|
||||
assert "awoooi_gitea_full_backup_production_restore_performed" in text
|
||||
assert "gitea dump -c /data/gitea/conf/app.ini" in text
|
||||
assert "offline_gitea_dump" in text
|
||||
assert "--volumes-from" in text
|
||||
assert "--pull=never" in text
|
||||
assert "docker stop --time" in text
|
||||
assert "start_and_verify_primary" in text
|
||||
assert "gitea-full-backup-restore-drill.sh" in text
|
||||
assert "GITEA_FULL_BACKUP_COMPONENTS" in text
|
||||
|
||||
for component in [
|
||||
@@ -25,6 +31,9 @@ def test_backup_gitea_writes_full_server_component_receipts() -> None:
|
||||
"lfs_objects",
|
||||
"package_registry",
|
||||
"attachments",
|
||||
"actions_logs",
|
||||
"actions_artifacts",
|
||||
"avatars",
|
||||
"hooks_custom_assets",
|
||||
]:
|
||||
assert component in text
|
||||
@@ -37,7 +46,7 @@ def test_backup_gitea_receipt_contract_stays_no_secret_no_restore() -> None:
|
||||
assert "awoooi_gitea_full_backup_production_restore_performed" in text
|
||||
assert "awoooi_gitea_full_backup_restore_drill_performed" in text
|
||||
assert (
|
||||
'restore_drill_performed{host=\\"${host}\\",service=\\"${service_label}\\"} 0'
|
||||
'restore_drill_performed{host=\\"${host}\\",service=\\"${service_label}\\"} ${structural_drill_performed}'
|
||||
in text
|
||||
)
|
||||
assert (
|
||||
@@ -48,3 +57,34 @@ def test_backup_gitea_receipt_contract_stays_no_secret_no_restore() -> None:
|
||||
'production_restore_performed{host=\\"${host}\\",service=\\"${service_label}\\"} 0'
|
||||
in text
|
||||
)
|
||||
|
||||
|
||||
def test_backup_gitea_is_bounded_and_fail_safe_around_the_offline_window() -> None:
|
||||
text = BACKUP_GITEA.read_text(encoding="utf-8")
|
||||
|
||||
assert 'GITEA_BACKUP_LOCK="${GITEA_BACKUP_LOCK:-/tmp/gitea-backup.lock}"' in text
|
||||
assert "flock -n 9" in text
|
||||
assert "trap cleanup EXIT" in text
|
||||
assert 'if [ "${PRIMARY_STOPPED}" -eq 1 ]' in text
|
||||
assert "wait_for_primary_health" in text
|
||||
assert "http://127.0.0.1:3001/api/healthz" in text
|
||||
assert '--kill-after="${GITEA_DUMP_KILL_AFTER_SECONDS}s"' in text
|
||||
assert "bounded_docker_control rm -f" in text
|
||||
assert 'GITEA_MAX_OUTAGE_SECONDS="${GITEA_MAX_OUTAGE_SECONDS:-600}"' in text
|
||||
assert "configured_outage_bound" in text
|
||||
assert "--security-opt no-new-privileges:true" in text
|
||||
assert '--cpus "${BACKUP_DOCKER_CPUS}"' in text
|
||||
assert '--memory "${BACKUP_DOCKER_MEMORY}"' in text
|
||||
assert '--memory-swap "${BACKUP_DOCKER_MEMORY_SWAP}"' in text
|
||||
assert "awoooi_gitea_full_backup_consistent_offline_window" in text
|
||||
assert "awoooi_gitea_full_backup_service_restart_verified" in text
|
||||
assert "awoooi_gitea_full_backup_service_outage_seconds" in text
|
||||
|
||||
|
||||
def test_gitea_runner_configuration_is_in_the_encrypted_config_backup() -> None:
|
||||
text = BACKUP_CONFIGS.read_text(encoding="utf-8")
|
||||
|
||||
assert "/home/wooo/vibework-act-runner/config.yaml" in text
|
||||
assert 'record_config_status "110-gitea-runner-config" true true "110"' in text
|
||||
assert 'record_config_status "110-gitea-runner-config" true false "110"' in text
|
||||
assert "/home/wooo/vibework-act-runner/data" not in text
|
||||
|
||||
@@ -0,0 +1,300 @@
|
||||
from __future__ import annotations
|
||||
|
||||
import os
|
||||
import subprocess
|
||||
import textwrap
|
||||
from pathlib import Path
|
||||
|
||||
|
||||
ROOT = Path(__file__).resolve().parents[3]
|
||||
SCRIPT = ROOT / "scripts" / "backup" / "backup-gitea.sh"
|
||||
|
||||
|
||||
def _write_executable(path: Path, source: str) -> None:
|
||||
path.write_text(textwrap.dedent(source), encoding="utf-8")
|
||||
path.chmod(0o755)
|
||||
|
||||
|
||||
def _fake_runtime(tmp_path: Path) -> tuple[dict[str, str], Path, Path]:
|
||||
fake_bin = tmp_path / "bin"
|
||||
fake_bin.mkdir()
|
||||
state_dir = tmp_path / "state"
|
||||
state_dir.mkdir()
|
||||
(state_dir / "running").write_text("true\n", encoding="utf-8")
|
||||
command_log = state_dir / "commands.log"
|
||||
|
||||
_write_executable(
|
||||
fake_bin / "docker",
|
||||
r"""
|
||||
#!/usr/bin/env bash
|
||||
set -euo pipefail
|
||||
state="${FAKE_DOCKER_STATE:?}"
|
||||
printf '%s\n' "$*" >> "$state/commands.log"
|
||||
command_name="${1:-}"
|
||||
shift || true
|
||||
case "$command_name" in
|
||||
inspect)
|
||||
args="$*"
|
||||
case "$args" in
|
||||
*State.Running*) cat "$state/running" ;;
|
||||
*NetworkSettings.Networks*) printf 'gitea_gitea\n' ;;
|
||||
*'{{.Image}}'*) printf 'sha256:%064d\n' 0 ;;
|
||||
*) exit 64 ;;
|
||||
esac
|
||||
;;
|
||||
exec)
|
||||
args="$*"
|
||||
if [[ "$args" == *' id -u'* ]]; then
|
||||
id -u
|
||||
elif [[ "$args" == *'gitea --version'* ]]; then
|
||||
printf 'gitea version 1.25.5 built with test\n'
|
||||
else
|
||||
exit 64
|
||||
fi
|
||||
;;
|
||||
stop)
|
||||
printf 'false\n' > "$state/running"
|
||||
if [ "${FAKE_DOCKER_STOP_FAIL:-0}" = "1" ]; then
|
||||
exit 42
|
||||
fi
|
||||
;;
|
||||
start)
|
||||
printf 'true\n' > "$state/running"
|
||||
;;
|
||||
run)
|
||||
output=""
|
||||
for arg in "$@"; do
|
||||
case "$arg" in
|
||||
type=bind,src=*,dst=/backup-out)
|
||||
output="${arg#type=bind,src=}"
|
||||
output="${output%,dst=/backup-out}"
|
||||
;;
|
||||
esac
|
||||
done
|
||||
[ -n "$output" ] || exit 65
|
||||
if [ "${FAKE_DOCKER_RUN_FAIL:-0}" = "1" ]; then
|
||||
exit 42
|
||||
fi
|
||||
python3 - "$output/gitea-dump.zip" <<'PY'
|
||||
import sys
|
||||
import zipfile
|
||||
|
||||
entries = {
|
||||
"gitea-db.sql": "fixture\n",
|
||||
"repos/wooo/awoooi.git/HEAD": "ref: refs/heads/main\n",
|
||||
"custom/conf/app.ini": "[server]\n",
|
||||
"data/lfs/objects/aa/bb/object": "lfs\n",
|
||||
"data/packages/package.bin": "package\n",
|
||||
"data/attachments/attachment.bin": "attachment\n",
|
||||
"data/actions_artifacts/aa/bb/artifact.zip": "artifact\n",
|
||||
"data/actions_log/aa/bb/job.log": "log\n",
|
||||
"data/avatars/aa/bb/avatar.png": "avatar\n",
|
||||
"custom/hooks/pre-receive": "#!/bin/sh\n",
|
||||
}
|
||||
with zipfile.ZipFile(sys.argv[1], "w") as archive:
|
||||
for name, value in entries.items():
|
||||
archive.writestr(name, value)
|
||||
PY
|
||||
;;
|
||||
rm) ;;
|
||||
*) exit 64 ;;
|
||||
esac
|
||||
""",
|
||||
)
|
||||
_write_executable(
|
||||
fake_bin / "curl",
|
||||
r"""
|
||||
#!/usr/bin/env bash
|
||||
url="${!#}"
|
||||
if [ "$url" = "http://127.0.0.1:3001/api/healthz" ]; then
|
||||
printf '{"status":"pass"}'
|
||||
else
|
||||
printf '<html>another service</html>'
|
||||
fi
|
||||
""",
|
||||
)
|
||||
_write_executable(
|
||||
fake_bin / "timeout",
|
||||
"""
|
||||
#!/usr/bin/env bash
|
||||
set -euo pipefail
|
||||
while [[ "${1:-}" == --* ]]; do shift; done
|
||||
shift
|
||||
if [ "${FAKE_DOCKER_RUN_TIMEOUT:-0}" = "1" ] && [[ " $* " == *' docker run '* ]]; then
|
||||
exit 124
|
||||
fi
|
||||
exec "$@"
|
||||
""",
|
||||
)
|
||||
_write_executable(
|
||||
fake_bin / "flock",
|
||||
"""
|
||||
#!/usr/bin/env bash
|
||||
exit 0
|
||||
""",
|
||||
)
|
||||
_write_executable(
|
||||
fake_bin / "restic",
|
||||
r"""
|
||||
#!/usr/bin/env bash
|
||||
set -euo pipefail
|
||||
repo=""
|
||||
args=("$@")
|
||||
for ((i=0; i<${#args[@]}; i++)); do
|
||||
if [ "${args[$i]}" = "-r" ]; then repo="${args[$((i+1))]}"; fi
|
||||
done
|
||||
if [[ " $* " == *' init '* ]]; then
|
||||
mkdir -p "$repo/data" "$repo/snapshots"
|
||||
exit 0
|
||||
fi
|
||||
if [[ " $* " == *' backup '* ]]; then
|
||||
mkdir -p "$repo/data" "$repo/snapshots"
|
||||
exit 0
|
||||
fi
|
||||
if [[ " $* " == *' snapshots '* ]]; then
|
||||
printf '[{"short_id":"abcd1234","time":"2026-07-18T00:00:00Z"}]\n'
|
||||
exit 0
|
||||
fi
|
||||
if [[ " $* " == *' forget '* ]]; then exit 0; fi
|
||||
exit 64
|
||||
""",
|
||||
)
|
||||
|
||||
backup_root = tmp_path / "backup"
|
||||
(backup_root / "scripts").mkdir(parents=True)
|
||||
(backup_root / "scripts" / ".restic-password").write_text("fixture\n", encoding="utf-8")
|
||||
textfile = tmp_path / "gitea.prom"
|
||||
environment = os.environ | {
|
||||
"PATH": f"{fake_bin}:{os.environ['PATH']}",
|
||||
"BACKUP_BASE": str(backup_root),
|
||||
"BACKUP_LOG_DIR": str(backup_root / "logs"),
|
||||
"RESTIC_PASSWORD_FILE": str(backup_root / "scripts" / ".restic-password"),
|
||||
"GITEA_FULL_BACKUP_RECEIPT_TEXTFILE": str(textfile),
|
||||
"NODE_EXPORTER_TEXTFILE_DIR": str(tmp_path),
|
||||
"GITEA_BACKUP_LOCK": str(tmp_path / "gitea.lock"),
|
||||
"GITEA_START_TIMEOUT_SECONDS": "4",
|
||||
"GITEA_DUMP_TIMEOUT_SECONDS": "30",
|
||||
"GITEA_DUMP_KILL_AFTER_SECONDS": "2",
|
||||
"GITEA_DOCKER_CONTROL_TIMEOUT_SECONDS": "2",
|
||||
"GITEA_DOCKER_CONTROL_KILL_AFTER_SECONDS": "1",
|
||||
"GITEA_STOP_TIMEOUT_SECONDS": "2",
|
||||
"FAKE_DOCKER_STATE": str(state_dir),
|
||||
"BACKUP_RETENTION_MODE": "latest",
|
||||
}
|
||||
return environment, command_log, textfile
|
||||
|
||||
|
||||
def test_offline_backup_restarts_and_verifies_primary_on_success(tmp_path: Path) -> None:
|
||||
environment, command_log, textfile = _fake_runtime(tmp_path)
|
||||
|
||||
result = subprocess.run(
|
||||
["bash", str(SCRIPT)],
|
||||
env=environment,
|
||||
text=True,
|
||||
capture_output=True,
|
||||
check=False,
|
||||
)
|
||||
|
||||
assert result.returncode == 0, result.stderr
|
||||
commands = command_log.read_text(encoding="utf-8")
|
||||
assert commands.index("stop --time 2 gitea") < commands.index("run --pull=never")
|
||||
assert commands.index("run --pull=never") < commands.index("rm -f agent99-gitea-backup-")
|
||||
assert commands.index("rm -f agent99-gitea-backup-") < commands.index("start gitea")
|
||||
rendered = textfile.read_text(encoding="utf-8")
|
||||
assert "awoooi_gitea_full_backup_consistent_offline_window" in rendered
|
||||
assert "awoooi_gitea_full_backup_service_restart_verified" in rendered
|
||||
assert 'service="gitea"} 1' in rendered
|
||||
|
||||
|
||||
def test_offline_backup_restarts_primary_when_dump_fails(tmp_path: Path) -> None:
|
||||
environment, command_log, textfile = _fake_runtime(tmp_path)
|
||||
environment["FAKE_DOCKER_RUN_FAIL"] = "1"
|
||||
|
||||
result = subprocess.run(
|
||||
["bash", str(SCRIPT)],
|
||||
env=environment,
|
||||
text=True,
|
||||
capture_output=True,
|
||||
check=False,
|
||||
)
|
||||
|
||||
assert result.returncode != 0
|
||||
assert (Path(environment["FAKE_DOCKER_STATE"]) / "running").read_text(encoding="utf-8").strip() == "true"
|
||||
commands = command_log.read_text(encoding="utf-8")
|
||||
assert "stop --time 2 gitea" in commands
|
||||
assert commands.index("rm -f agent99-gitea-backup-") < commands.index("start gitea")
|
||||
rendered = textfile.read_text(encoding="utf-8")
|
||||
assert 'awoooi_gitea_full_backup_service_restart_verified{host="110",service="gitea"} 1' in rendered
|
||||
assert 'status="offline_dump_failed"} 1' in rendered
|
||||
|
||||
|
||||
def test_offline_backup_restarts_primary_when_dump_hits_hard_timeout(tmp_path: Path) -> None:
|
||||
environment, command_log, textfile = _fake_runtime(tmp_path)
|
||||
environment["FAKE_DOCKER_RUN_TIMEOUT"] = "1"
|
||||
|
||||
result = subprocess.run(
|
||||
["bash", str(SCRIPT)],
|
||||
env=environment,
|
||||
text=True,
|
||||
capture_output=True,
|
||||
check=False,
|
||||
)
|
||||
|
||||
assert result.returncode != 0
|
||||
assert (Path(environment["FAKE_DOCKER_STATE"]) / "running").read_text(encoding="utf-8").strip() == "true"
|
||||
commands = command_log.read_text(encoding="utf-8")
|
||||
assert commands.index("rm -f agent99-gitea-backup-") < commands.index("start gitea")
|
||||
assert 'status="offline_dump_failed"} 1' in textfile.read_text(encoding="utf-8")
|
||||
|
||||
|
||||
def test_offline_backup_recovers_when_stop_has_side_effect_then_fails(tmp_path: Path) -> None:
|
||||
environment, command_log, textfile = _fake_runtime(tmp_path)
|
||||
environment["FAKE_DOCKER_STOP_FAIL"] = "1"
|
||||
|
||||
result = subprocess.run(
|
||||
["bash", str(SCRIPT)],
|
||||
env=environment,
|
||||
text=True,
|
||||
capture_output=True,
|
||||
check=False,
|
||||
)
|
||||
|
||||
assert result.returncode != 0
|
||||
assert (Path(environment["FAKE_DOCKER_STATE"]) / "running").read_text(encoding="utf-8").strip() == "true"
|
||||
commands = command_log.read_text(encoding="utf-8")
|
||||
assert commands.index("stop --time 2 gitea") < commands.index("start gitea")
|
||||
assert 'service_restart_verified{host="110",service="gitea"} 1' in textfile.read_text(encoding="utf-8")
|
||||
|
||||
|
||||
def test_wrong_service_health_response_cannot_open_offline_window(tmp_path: Path) -> None:
|
||||
environment, command_log, _ = _fake_runtime(tmp_path)
|
||||
environment["GITEA_LOCAL_HEALTH_URL"] = "http://127.0.0.1:3000/"
|
||||
|
||||
result = subprocess.run(
|
||||
["bash", str(SCRIPT)],
|
||||
env=environment,
|
||||
text=True,
|
||||
capture_output=True,
|
||||
check=False,
|
||||
)
|
||||
|
||||
assert result.returncode != 0
|
||||
commands = command_log.read_text(encoding="utf-8")
|
||||
assert "stop --time" not in commands
|
||||
|
||||
|
||||
def test_oversized_outage_budget_fails_before_primary_stop(tmp_path: Path) -> None:
|
||||
environment, command_log, _ = _fake_runtime(tmp_path)
|
||||
environment["GITEA_MAX_OUTAGE_SECONDS"] = "10"
|
||||
|
||||
result = subprocess.run(
|
||||
["bash", str(SCRIPT)],
|
||||
env=environment,
|
||||
text=True,
|
||||
capture_output=True,
|
||||
check=False,
|
||||
)
|
||||
|
||||
assert result.returncode != 0
|
||||
assert "Configured Gitea backup outage bound exceeds 10s" in result.stdout
|
||||
assert not command_log.exists() or "stop --time" not in command_log.read_text(encoding="utf-8")
|
||||
@@ -9,15 +9,25 @@ ROOT = Path(__file__).resolve().parents[1]
|
||||
SCRIPT = ROOT / "gitea-full-backup-restore-drill.sh"
|
||||
|
||||
|
||||
def _write_dump(path: Path, *, include_lfs: bool = True) -> None:
|
||||
def _write_dump(
|
||||
path: Path,
|
||||
*,
|
||||
include_lfs: bool = True,
|
||||
include_actions_artifact: bool = True,
|
||||
empty_critical_payloads: bool = False,
|
||||
) -> None:
|
||||
entries = {
|
||||
"gitea-db.sql": "-- redacted fixture\n",
|
||||
"repos/wooo/awoooi.git/HEAD": "ref: refs/heads/main\n",
|
||||
"gitea-db.sql": "" if empty_critical_payloads else "-- redacted fixture\n",
|
||||
"repos/wooo/awoooi.git/HEAD": "" if empty_critical_payloads else "ref: refs/heads/main\n",
|
||||
"custom/conf/app.ini": "[server]\nAPP_NAME=fixture\n",
|
||||
"data/packages/package.bin": "package fixture\n",
|
||||
"data/attachments/attachment.bin": "attachment fixture\n",
|
||||
"data/actions_log/aa/bb/job.log": "actions log fixture\n",
|
||||
"data/avatars/aa/bb/avatar.png": "avatar fixture\n",
|
||||
"custom/hooks/pre-receive": "#!/bin/sh\n",
|
||||
}
|
||||
if include_actions_artifact:
|
||||
entries["data/actions_artifacts/aa/bb/artifact.zip"] = "actions artifact fixture\n"
|
||||
if include_lfs:
|
||||
entries["data/lfs/objects/aa/bb/object"] = "lfs fixture\n"
|
||||
with zipfile.ZipFile(path, "w") as archive:
|
||||
@@ -97,3 +107,64 @@ def test_gitea_full_backup_restore_drill_fails_closed_when_component_missing(
|
||||
rendered = textfile.read_text(encoding="utf-8")
|
||||
assert 'component="lfs_objects",drill_scope="structural_metadata_only"} 0' in rendered
|
||||
assert 'production_restore_performed{host="110",service="gitea",' in rendered
|
||||
|
||||
|
||||
def test_gitea_full_backup_restore_drill_rejects_empty_database_and_repository(
|
||||
tmp_path: Path,
|
||||
) -> None:
|
||||
dump = tmp_path / "gitea-dump.zip"
|
||||
_write_dump(dump, empty_critical_payloads=True)
|
||||
|
||||
result = subprocess.run(
|
||||
["bash", str(SCRIPT), "--dump-zip", str(dump)],
|
||||
text=True,
|
||||
capture_output=True,
|
||||
)
|
||||
|
||||
assert result.returncode == 1
|
||||
assert "ERROR=component_coverage_gap" in result.stdout
|
||||
|
||||
|
||||
def test_gitea_full_backup_restore_drill_requires_actions_log_and_artifact(
|
||||
tmp_path: Path,
|
||||
) -> None:
|
||||
dump = tmp_path / "gitea-dump.zip"
|
||||
textfile = tmp_path / "gitea_full_backup_restore_drill.prom"
|
||||
_write_dump(dump, include_actions_artifact=False)
|
||||
|
||||
result = subprocess.run(
|
||||
[
|
||||
"bash",
|
||||
str(SCRIPT),
|
||||
"--dump-zip",
|
||||
str(dump),
|
||||
"--write-textfile",
|
||||
"--textfile",
|
||||
str(textfile),
|
||||
],
|
||||
text=True,
|
||||
capture_output=True,
|
||||
)
|
||||
|
||||
assert result.returncode == 1
|
||||
rendered = textfile.read_text(encoding="utf-8")
|
||||
assert 'component="actions_logs",drill_scope="structural_metadata_only"} 1' not in rendered
|
||||
assert 'component="actions_artifacts",drill_scope="structural_metadata_only"} 0' in rendered
|
||||
|
||||
|
||||
def test_gitea_full_backup_restore_drill_rejects_crc_corruption(tmp_path: Path) -> None:
|
||||
dump = tmp_path / "gitea-dump.zip"
|
||||
_write_dump(dump)
|
||||
payload = dump.read_bytes()
|
||||
fixture = b"-- redacted fixture\n"
|
||||
assert fixture in payload
|
||||
dump.write_bytes(payload.replace(fixture, b"X" + fixture[1:], 1))
|
||||
|
||||
result = subprocess.run(
|
||||
["bash", str(SCRIPT), "--dump-zip", str(dump)],
|
||||
text=True,
|
||||
capture_output=True,
|
||||
)
|
||||
|
||||
assert result.returncode == 1
|
||||
assert "ERROR=crc_failed" in result.stdout
|
||||
|
||||
Reference in New Issue
Block a user