Blog ·
Jira for DevOps: link the code, and schedule the chores
Jira for DevOps comes down to two jobs. The first is connecting Jira to the tools that ship code, so a work item shows the branch, the commits, the build and the deployment that delivered it. The second is the one most teams skip: putting the recurring operational chores on the board. Dependency upgrades, secret rotation, restore drills, cost reviews and runbook checks come back on a clock, and in most teams they live in one engineer's memory. This guide sets up both halves, and spends extra time on the second, because that is where the misses happen.
Statements about Jira's built-in features are as of October 2026, from Atlassian's own documentation.
What a DevOps team needs from Jira #
A DevOps team asks Jira for three things, and each one has a different source of truth:
| Need | Where the truth lives | What Jira should do |
|---|---|---|
| Planned delivery work | Your roadmap and backlog | Hold it on a board, with owners and dates |
| Traceability from ticket to production | Git, CI and your deploy tool | Show the branch, commits, builds and deployments on the work item |
| Recurring operational chores | The calendar | Create the ticket on time, every cycle |
| Incidents and alerts | Your monitoring and paging tools | Receive a ticket for follow-up work after the fire is out |
The third row is the one with no owner by default. Delivery work gets tickets because somebody planned it. Incidents get tickets because something broke. A quarterly secret rotation gets a ticket only when somebody remembers it is due.
Connect Jira to your code and pipelines #
Traceability is the reason most engineering teams accept Jira in the first place. It works on one convention: put the work item key everywhere the work goes.
Atlassian's guide to referencing work items in development work, as of October 2026, says to check out "a new branch in your repo, using the key in the branch name", and continues:
When committing changes to your branch, use the key in your commit message to link those commits to the development panel in your Jira work item.
In practice, for a work item called OPS-142:
- Branch:
OPS-142-rotate-ci-deploy-key. - Commits: start each message with
OPS-142. - Pull request: put
OPS-142in the title. - Builds and deployments: connect your CI tool to Jira through its integration, so build and deploy events show up on the same work item.
The same Atlassian page warns: "Make sure you've formatted the Jira work item key correctly using capital letters." A lowercase key is the most common reason a commit fails to appear, and a pre-commit hook or a branch naming rule in your Git host catches it before it reaches the main branch.
Once the links flow, the development panel answers the questions people used to ask in chat: is this merged, did it build, is it in production?
One board for delivery, kept honest #
Most DevOps teams run kanban rather than sprints, because platform work arrives unevenly and interrupts are normal. A short workflow covers nearly everything:
| Status | What it means |
|---|---|
| To do | Agreed, not started |
| In progress | Somebody is working on it now |
| In review | A pull request is open |
| Waiting | Blocked on a change window, a vendor or another team |
| Done | Merged, deployed and checked |
Set a work-in-progress limit on "In progress", so the board shows when the team is spread across too many half-finished upgrades. Our Jira kanban board guide covers columns and limits in detail. For the long-running platform debt that never quite fits a sprint, see Jira tech debt.
Keep incident response in the tool built for paging. What belongs in Jira is the follow-up: the action items from the post-incident review, each as a work item with an owner and a due date, so they do not evaporate once the incident channel goes quiet.
The recurring DevOps chores #
This is the calendar a typical platform or DevOps team carries. Every row is work that happens whether or not anybody created a ticket for it:
| Chore | Typical cadence | What goes wrong without a ticket |
|---|---|---|
| Dependency and base image upgrades | Every two weeks | Upgrades pile up until one is a migration |
| Secret and API key rotation | Every 90 days, or per policy | Keys outlive the people who created them |
| Backup restore drill | Monthly | Backups run for months without anyone restoring one |
| TLS certificate and domain expiry check | Monthly | An expired certificate takes a service down |
| Cloud cost review | Monthly | Idle resources keep billing |
| Flaky test triage | Weekly | The team learns to ignore red builds |
| Alert and dashboard review | Quarterly | Alerts fire for things nobody acts on |
| Runbook review | Quarterly | The runbook is wrong on the night it is needed |
| Disaster recovery or game day exercise | Twice a year | Recovery steps are untested |
| Access review for production and CI | Quarterly | Leavers keep access; an auditor finds it first |
Each one is small, and that is why it slips. A chore with a ticket has an owner, a due date and a record that it was done. A chore without one depends on whoever happens to remember.
The record matters more than it looks. When a security questionnaire asks whether production access is reviewed every quarter, a saved filter showing four resolved review tickets, each with a date, an assignee and notes, is the answer.
Three ways to create the recurring tickets #
| Clone by hand | Jira Automation scheduled rule | A scheduling app | |
|---|---|---|---|
| Who creates the ticket | A person, every cycle | A rule | A rule in the project |
| Setup per chore | None | A rule, often with a cron expression | A rule, in a form |
| "Last Friday of the month" | Whoever remembers | Needs cron | Chosen in the form |
| Weekends and holidays | Whoever notices | Handled in the rule's logic, if at all | Set once for the project |
| When the owner is on leave | Not created | Created on time | Created on time |
| Cost | Engineer time | Counts toward your plan's automation usage | App licence |
Clone by hand #
Keep a master ticket per chore with the steps in the description and clone it each cycle. It costs nothing and is fine for a team with two or three chores and one reliable owner. It fails in the week of a big release, which is usually the week the cost review gets skipped.
A Jira Automation scheduled rule #
Most DevOps engineers reach for this first, and it fits a simple cadence well. Atlassian's trigger reference, as of October 2026, says you can "run the flow at a fixed rate (for example, every 7 days), or use a Cron expression for more complex schedules." A weekly flaky test triage is a one-line rule. "The last Friday of the month" or "every second Tuesday" needs cron, and some cadences cannot be written in cron at all; our Jira cron expression guide covers the syntax and those gaps.
Two things to know before you build a dozen of these. Each run counts toward your plan's automation usage, explained in Jira Automation pricing. And the same Atlassian page says:
Scheduled flows that reach a Failure status for 10 consecutive executions will disable automatically.
For a weekly rule, that is ten weeks of failures before it switches itself off, and then silence. For a quarterly rotation reminder, it can mean years of nobody noticing. If you go this way, put "check the automation audit log" on the list of chores. Our Jira automation examples article walks through where scheduled rules tend to break.
A scheduling app #
Recurring Work Items for Jira is our app, and recurring routine work is the job it was built for. A DevOps chore becomes a rule in the project:
- Turn the app on for the platform project and choose who may schedule.
- Set the working week and add public holidays, so a rule can move off a day the team is not working.
- Create a rule per chore. The schedule options are every N days, chosen weekdays every N weeks, a day of the month every N months, or a weekday position such as the second Tuesday or the last Friday of the month.
- Name each ticket with tokens, for example
Rotate CI deploy keys {{now.monthName}} {{now.year}}, so every cycle has its own title in search results. - Put the steps in the description or as subtasks, and set the assignee, labels and a due date as an offset from the run.
It uses no Jira Automation quota, because the scheduling is the app's own. Every created or failed run is written to the rule's history with its reason, so "why is there no restore drill ticket this month?" has an answer you can read. A rule is scoped to the one project it lives in, so the platform team's schedules never create tickets in anyone else's project. It is free for up to 10 users.
A starter set of DevOps rules #
Adjust the cadence to your own policy. Add a routine label to every rule so the filters below can find them.
- Fortnightly dependency upgrade, every second Monday, due that Friday. Subtasks: application dependencies, base images, CI runner images.
- Weekly flaky test triage, every Monday, due Tuesday. Description: the CI view that lists the slowest and least reliable tests, and the rule "fix, quarantine or delete".
- Monthly backup restore drill, on the first Wednesday, due the following Friday. Description: which database to restore this month, where to restore it, and how to confirm the data.
- Monthly certificate and domain expiry check, on the 1st. Description: renew anything expiring in the next 60 days.
- Monthly cloud cost review, on the last Friday of the month. Description: the billing view to read, and where to record what was shut down.
- Quarterly secret rotation, every three months, assigned to each secret's owner.
- Quarterly runbook and alert review, every three months on the first Monday.
When a chore has several fixed steps owned by different people, use subtasks rather than one long description. See where checklist steps should live and Jira subtask template for the trade-offs. Our Jira patch management guide covers the monthly patch cycle in the same way.
Reporting on the routine #
Three saved filters answer the questions a platform lead is asked most:
- What routine work is overdue?
project = PLAT AND labels = routine AND due < now() AND statusCategory != Done - Did every secret rotation happen this year?
project = PLAT AND summary ~ "rotate" AND resolved >= startOfYear() - What is blocked on someone else?
project = PLAT AND status = Waiting ORDER BY updated ASC
Replace PLAT with your project key. Subscribe to the first one so the overdue list arrives by email every Monday. The Jira weekly report guide shows how to turn filters like these into a short report.
When the manual way is enough #
If your team has a couple of chores and one person who owns them without fail, a calendar reminder and a cloned ticket are enough. If you already run Jira Automation rules comfortably inside your plan's usage and your cadences fit a fixed rate, a scheduled rule for the weekly triage is the natural next step. A scheduling app earns its place when the chore list grows past what anyone wants to maintain as cron expressions, when holidays keep producing tickets on days nobody is in, or when an auditor wants proof that each cycle happened.
For a wider look at the options, see Jira recurring tasks, and for the IT side of the house, Jira for IT operations.
Frequently asked questions #
Is Jira a DevOps tool? Jira tracks the work rather than running it. With the work item key in branch names and commits, and your CI tool connected, a work item shows the code, builds and deployments that delivered it.
How do I link commits to a Jira work item? Put the key, in capital letters, in the branch name and at the start of each commit message. The commits then appear in the work item's development panel.
How do I schedule recurring DevOps tasks in Jira? Clone a master ticket each cycle, create a Jira Automation scheduled rule, or use a scheduling app such as Recurring Work Items for Jira, where a rule like "last Friday of the month" is chosen in a form.