jira-confluence-full-instance-backup 0.5.0

Full-instance backup of Jira & Confluence Cloud (Standard plan) to pluggable cloud storage

Jira & Confluence Full-Instance Backup

Automated full-instance backup of Jira Cloud and Confluence Cloud for the Atlassian Standard plan. Run it by hand from an interactive menu, or unattended from Jenkins/cron. Backups are encrypted and shipped to the cloud of your choice — Google Cloud Storage, AWS S3 (and S3-compatible stores), Azure Blob, or a local/mounted directory — with notifications to any channel you use: Slack, Microsoft Teams, Discord, Google Chat, email, or a generic webhook.

---

Why?

On March 30, 2026, Atlassian deprecated the Backup Manager API for Jira Cloud. Direct API-token calls to the backup endpoint now return:

HTTP 403
{"error":"This feature is only accessible from the UI."}

The replacement v2 Backup & Restore API is Premium/Enterprise only, leaving Standard-plan customers with no automation path for full-instance backup — only the manual UI button. This tool restores that automation by replaying the browser UI session for Jira, while Confluence uses the OBM REST API (which still accepts API tokens). Both flow into one pipeline that archives, encrypts, uploads, and notifies.

For per-project Jira backup/restore, see the sibling project jira-project-backup-restore.

---

Features

Feature	Description
Full-instance backup	Jira (all projects, attachments, avatars, logos) + Confluence (all spaces, attachments) in one run
Run anywhere	Interactive menu for VMs/manual use and CLI flags for Jenkins/cron — same codebase; the Jenkinsfile runs on Linux and Windows agents
One-paste Jenkins setup	Export a Script Console script from your config that creates all credentials + the pipeline job — no manual credential forms
Pluggable storage	GCS · AWS S3 / S3-compatible (R2, B2, MinIO, Spaces) · Azure Blob · local — pick one flag; only that SDK is needed
Pluggable notifications	Slack · Teams · Discord · Google Chat · email (SMTP) · generic webhook — any combination, no extra deps
Optional encryption	7-Zip AES-256 with encrypted headers — on by default, switch off with --no-encrypt
Configurable compression	0 (store) … 9 (ultra)
Custom filenames	Name templates: {product} {site} {date} {time} {datetime} {timestamp}
Integrity & housekeeping	manifest.json with sha256 + --validate, --cleanup, --skip-existing, --dry-run
Cooldown-aware	Atlassian's 48h throttle (HTTP 412) is detected and skipped cleanly — no false failures
Connection test	Validates Jira cookies + Confluence token, warns when the Jira session is near expiry

---

Install

pip install jira-confluence-full-instance-backup            # core (requests)
# add the storage backend you use:
pip install "jira-confluence-full-instance-backup[s3]"      # or [gcs] / [azure]
# optional nicer interactive output (progress bars, color):
pip install "jira-confluence-full-instance-backup[ui]"
# everything:
pip install "jira-confluence-full-instance-backup[all]"

Or from source:

git clone https://github.com/davidmalko87/jira-confluence-full-instance-backup.git
cd jira-confluence-full-instance-backup
pip install -r requirements.txt          # + requirements-<provider>.txt as needed

Requirements: Python 3.10+, and 7z on PATH (apt install p7zip-full, or set SEVEN_ZIP_PATH).

---

Quick Start

1. Configure

cp .env.example .env      # fill in real values — .env is gitignored

Or use the guided menu (writes .env for you): jira-confluence-backup → Configure credentials. It explains every field — including how to obtain the Jira cookie blob — and validates what you paste.

Hidden input is intentional. When entering secrets (API token, cookie blob, passwords), nothing is echoed to the screen — the prompt shows [input hidden — paste, then Enter]. Paste and press Enter; the tool confirms back (e.g. "captured 5 cookie(s); all required present") so you know it registered, without ever displaying the value. Use Test connections to verify.

2. Run — interactive menu

jira-confluence-backup            # or: python main.py   /   python -m backup

=== Atlassian Full-Instance Backup ===
  Jira     https://<your-site>.atlassian.net
  Storage  s3:my-backups
  Notify   slack

  1) Backup Jira          7) Validate backup
  2) Backup Confluence    8) Cleanup backups
  3) Backup both          9) Test connections
  4) Full run            10) Configure credentials
  5) Archive ./out       11) Show configuration
  6) Upload ./archive    12) List local backups
                         13) Export Jenkins setup
                         14) Refresh Jira cookies
  0) Exit

3. Run — CLI (automation)

jira-confluence-backup --all                              # backup both -> archive -> upload -> notify
jira-confluence-backup --all --dry-run                    # preview only, no API calls / no cooldown burn
jira-confluence-backup --backup jira,confluence --archive --upload --notify
jira-confluence-backup --validate                         # check the archive against its manifest
jira-confluence-backup --cleanup --keep-days 28           # prune incomplete + old local backups
jira-confluence-backup --test-connection                  # exit 0 if both auth paths are OK

Output is plain ASCII ([INFO]/[OK]) by default — safe on any console, including legacy Windows. Install the ui extra for colored output and progress bars.

---

How It Works

Setup  ->  Jira  ->  Confluence  ->  Archive  ->  Upload          ->  Notify
(venv)    (cookies)  (API token)    (7z, opt.    (<provider>://      (your
                                     AES-256)     <dest>/Y/M/D/)      channels)

Stages are independent: a Jira cookie expiry does not stop the Confluence stage.

Auth model

Product	Endpoint	Auth	Why
Jira	/rest/backup/1/export/runbackup	Session cookies + UI headers	Atlassian gates this endpoint to UI sessions only — API tokens return 403
Confluence	/wiki/rest/obm/1.0/runbackup	Basic (email + API token)	OBM never received the UI-only lockdown

Do not replace the Jira side with an API token — it is gated to browser sessions and returns 403 "This feature is only accessible from the UI.". One dedicated Atlassian admin account supplies both: its API token (for Confluence) and its browser session cookies (for Jira).

---

Storage backends

Set STORAGE_PROVIDER + STORAGE_DEST (or --provider / --dest). Only the chosen SDK is imported; a missing one prints a pip install hint.

Provider	STORAGE_DEST	Optional SDK	Credentials
gcs	bucket	requirements-gcs.txt	GOOGLE_APPLICATION_CREDENTIALS (SA JSON)
s3	bucket	requirements-s3.txt	AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_DEFAULT_REGION
azure	container	requirements-azure.txt	AZURE_STORAGE_CONNECTION_STRING
local	directory	(none)	(none)

S3-compatible stores (Cloudflare R2, Backblaze B2, MinIO, DigitalOcean Spaces): use s3 plus S3_ENDPOINT_URL. Objects are written to <dest>/YYYY/MM/DD/. Retention is the bucket's job (set a lifecycle rule); use a write-only identity where possible.

---

Notification channels

NOTIFY_CHANNELS is a comma list — pick any. One report is built and rendered per channel; a failed channel is logged but never blocks the rest.

Channel	Needs	Notes
slack / discord / teams / google-chat	NOTIFY_WEBHOOK_URL	platform-specific incoming webhook
email	SMTP_*	stdlib SMTP; port 465 → SSL, else STARTTLS
webhook	NOTIFY_WEBHOOK_URL	raw JSON POST (PagerDuty / Opsgenie / your API)

---

Archiving: encryption, compression, names

• Encryption (default on): 7-Zip AES-256 with encrypted headers, password from ARCHIVE_PASSWORD. Turn it off with --no-encrypt or by leaving the password blank.
• Compression: ARCHIVE_COMPRESSION / --compression 0–9.
• Names: PRODUCT_NAME_TEMPLATE (the per-product .zip) and ARCHIVE_NAME_TEMPLATE (the .7z). Tokens: {product} {site} {date} {time} {datetime} {timestamp} {year} {month} {day}. Defaults: {product}-{date} and atlassian-backup-{date}.

Integrity & housekeeping

Each successful run writes a manifest.json (timestamp, products, per-file + archive sha256, complete: true, encrypted) next to the archive and uploads it too.

• --validate — re-checksum the local archive against the manifest.
• --cleanup [--keep-days N] — remove incomplete backups (no manifest) and, optionally, ones older than N days.
• --skip-existing — skip a product already backed up today.
• --dry-run — preview any flow without API calls, archiving, or uploads.

---

Jenkins

The Jenkinsfile is cross-platform — it runs on both Linux (sh) and Windows (powershell) agents. The agent needs Python 3.10+ and 7-Zip (apt install p7zip-full, or install 7-Zip on Windows). It runs weekly (cron('H 2 * * 4')) and installs only the selected provider's SDK.

📖 Full step-by-step guide (prerequisites, plugins, every credential, troubleshooting): docs/JENKINS_SETUP.md. The summary below is the quick version.

Fast setup — one paste (recommended)

1. Configure locally: python main.py → Configure credentials (writes .env).
2. Generate the setup script: python main.py → Export Jenkins setup (or python main.py --export-jenkins). This writes a gitignored jenkins-setup.groovy.
3. In Jenkins: Manage Jenkins → Script Console, paste the file's contents, Run. It creates every credential (with the exact IDs) and the atlassian-full-backup pipeline job pointing at this repo.
4. Open the job → Build Now, then delete jenkins-setup.groovy (it contains secrets).

The job uses Pipeline script from SCM, so Jenkins reads the Jenkinsfile straight from the repo at build time — there is no Jenkinsfile to generate or maintain inside Jenkins, and updates flow automatically from the repo.

Manual setup

Edit the Jenkinsfile environment block for your site / provider / channels:

environment {
    SITE_JIRA        = 'https://<YOUR_SITE>.atlassian.net'
    SITE_CONFLUENCE  = 'https://<YOUR_SITE>.atlassian.net/wiki'
    STORAGE_PROVIDER = 'gcs'            // gcs | s3 | azure | local
    STORAGE_DEST     = '<YOUR_BUCKET>'
    NOTIFY_CHANNELS  = 'slack'          // any comma list
}

Create a Pipeline job → Pipeline script from SCM → Git → this repo → */master → Script Path Jenkinsfile, and add the credentials below (IDs must match exactly — the pipeline binds only the ones your config needs):

Credential ID	Kind	Used when
jira-cookies	Secret text	always (Jira)
atlassian-email, atlassian-api-token	Secret text	always (Confluence)
archive-password	Secret text	always (may be blank = unencrypted)
gcp-backup-sa-key	Secret file	STORAGE_PROVIDER=gcs
aws-access-key-id, aws-secret-access-key	Secret text	STORAGE_PROVIDER=s3
azure-storage-connection-string	Secret text	STORAGE_PROVIDER=azure
notify-webhook-url	Secret text	chat/webhook channels
smtp-host/from/to/user/password	Secret text	email channel

No secrets live in the repo — they stay in the Jenkins Credentials store.

---

Cookie Refresh Procedure

The Jira tenant.session.token JWT expires roughly every 30 days. When it does, the Jira stage exits with code 2 (Cookie auth rejected — cookies likely expired). Refresh takes ~60 seconds:

1. Log into Atlassian as the backup admin account.
2. Open https://<your-site>.atlassian.net/secure/admin/CloudExport.jspa.
3. F12 → Network, reload the page, right-click the CloudExport.jspa request (or any /rest/backup/1/export/... request) → Copy → Copy as cURL.
4. Paste the whole cURL into the menu's Configure credentials (it extracts the cookies), or into the jira-cookies Jenkins credential / JIRA_COOKIES in .env.

The cookie blob must contain at least tenant.session.token and atlassian.xsrf.token. Other cookies such as JSESSIONID / AWSALB / AWSALBCORS are load-balancer/servlet cookies that some instances set and others don't — they're forwarded automatically when present, and not required. Using a real request's "Copy as cURL" (rather than the Application→Cookies list) ensures any that are needed come along.

Test connections warns you in advance when the token is within a few days of expiry.

---

Response Code Semantics

Code	Meaning	Behavior
200	Backup queued / status returned	Continue polling
403	Auth rejected (UI-only gate)	Exit 2 — refresh Jira cookies
412	48-hour cooldown active	Exit 0 + marker — stays green
400	Body schema rejected	Investigate body (Atlassian schema change)
406	Confluence cosmetic error	Ignore — backup actually started

---

Configuration Reference

All values come from environment variables, optionally loaded from .env (see .env.example). In Jenkins they are bound from the Credentials store at runtime — never from a file in the repo.

Env var	Purpose
SITE_JIRA / SITE_CONFLUENCE	Atlassian base URLs
JIRA_COOKIES	Browser session cookie blob for Jira
ATL_EMAIL / ATL_TOKEN	Confluence Basic auth
ARCHIVE_PASSWORD	7-Zip AES-256 passphrase (blank = unencrypted)
ARCHIVE_COMPRESSION	0–9
PRODUCT_NAME_TEMPLATE / ARCHIVE_NAME_TEMPLATE	Filename templates
STORAGE_PROVIDER / STORAGE_DEST	Backend + bucket/container/dir (aligned comma lists for multiple targets, e.g. gcs,s3 + bucketA,bucketB)
BACKUP_CRON	Jenkins schedule (default H 2 * * 4)
PYTHON_BIN	Jenkins agent Python path (blank = auto-detect; set if python isn't on the service PATH)
S3_ENDPOINT_URL	S3-compatible endpoint (s3 only)
GOOGLE_APPLICATION_CREDENTIALS / AWS_* / AZURE_STORAGE_CONNECTION_STRING	Provider credentials
NOTIFY_CHANNELS	Comma list of channels
NOTIFY_WEBHOOK_URL / SMTP_*	Notification delivery

---

Project Structure

jira-confluence-full-instance-backup/
├── main.py                   # Convenience shim (python main.py)
├── Jenkinsfile               # Declarative pipeline (provider/channel driven)
├── pyproject.toml            # Packaging + console script + ruff config
├── .env.example              # Local-testing template (real .env is gitignored)
├── requirements.txt          # Core (requests)
├── requirements-{gcs,s3,azure,ui}.txt   # Optional extras
├── docs/JENKINS_SETUP.md     # Full Jenkins setup guide
└── backup/
    ├── cli.py                # Dual-mode entrypoint (menu + CLI)
    ├── jira.py               # Cookie-authenticated Jira backup
    ├── confluence.py         # OBM Basic-auth Confluence backup
    ├── archive.py            # 7-Zip (optional AES-256, configurable level)
    ├── upload.py             # Multi-provider upload (gcs/s3/azure/local)
    ├── notify.py             # Multi-channel notifier
    ├── manifest.py           # manifest.json: completeness + sha256 integrity
    ├── config.py             # Env/.env config + Configure-menu persistence
    ├── jenkins_export.py     # Generate Script Console Groovy (creds + job)
    ├── naming.py             # Filename templating
    └── ui.py                 # Console UI (rich-optional, ASCII-safe)

---

Known Limitations

These are Atlassian platform constraints, not tool limitations:

• 48h Jira cooldown between full-instance backups (weekly cadence is fine).
• Cookie lifetime ~30 days; monthly manual refresh required.
• Confluence Filestore retention ~14 days (the tool downloads immediately, so this affects only the source file).
• No restore automation — restoring a full-instance backup is manual via Atlassian's UI; for per-project restore use jira-project-backup-restore.

---

Troubleshooting

Run Test connections (menu option 9) first — it pinpoints which of Jira / Confluence / storage is misconfigured. Common errors and fixes (403 vs the 204 "success", cookie refresh, the 48-hour cooldown, storage permissions, Jenkins) are in docs/TROUBLESHOOTING.md.

---

Contributing

See CONTRIBUTING.md. In short: no secrets in the repo, keep cloud SDKs optional, ASCII-only console output, and test auth/backup changes against a non-prod Atlassian instance. Do not switch Jira to API-token auth (see Auth model).

License

MIT — see LICENSE.

Related

• jira-project-backup-restore — per-project Jira Cloud backup/restore via REST API.