- What changed for administrators
- For integrations
- Performance
- Cloud migration
- Fixes
- Upgrade checklist
6.4 ships in two builds with the same functionality: one for Jira 10.x and one for Jira 11.3. Pick the build that matches your Jira version on the Marketplace page - the app version, the settings and the REST API are identical in both.
The release carries the results of the 2026 Data Center review. Most of it is invisible, but six changes are visible from the outside: they change how the app is called, what it refuses to do, and who sees a field value. Read the two sections below before you upgrade.
GET call to the app's REST API now needs an XSRF token, including calls authenticated with a Personal Access Token or Basic auth. See REST API - XSRF and the X-Atlassian-Token header.
What changed for administrators
Field value access is enforced
The field configuration tab Performance is now called Value access, and it does what its name says: the roles, groups and users listed there are the only ones who get the field value. Until 6.4 the lists limited real-time sorting in JQL but the value itself could still reach a user who was not on them.
What to do: open every field configuration that has entries on that tab and check the lists are the audience you want. Lists that were filled in before the upgrade start to take effect after the first save of the configuration - open the configuration, review the tab, save.
See Time in status - field configuration.
Manual control of timers and stopwatches needs Browse Projects
A user who starts, pauses, stops or restarts a counter must now have the Browse Projects permission on the issueâs project. This is checked for the buttons on the issue, for the mass action by JQL and for the REST API.
The groups and roles you set for manual control now replace the old âwhoever can edit the issueâ rule instead of widening it: if the configuration names groups or roles, only their members can drive the counter, even if other users can edit the issue.
See Manual control for Timers and Stopwatches.
Stopped timers no longer send âless than N leftâ
A timer in the stopped state no longer emits the âX% passedâ / âN minutes leftâ notification. Before 6.4 a stopped timer could still notify about a deadline that nothing was counting towards any more.
Notifications for running timers are unchanged. See Timer config.
The JQL selection limit no longer takes 0 as âunlimitedâ
jqlCountMaxResult (the Count limit setting) stops accepting 0 as âno limitâ:
0or an empty value now means the default of 1000 issues.- No selection processes more than 100 000 issues, and that ceiling also applies to the users in Ignore count limit. The exception list raises the limit up to the ceiling, it no longer removes it.
What to do: if you relied on 0 for scripts, system users or reporting tools, set an explicit number. See Performance settings - JQL.
Oversized table reports fail instead of running
A table report whose JQL selects more issues than the limit now finishes with an error - more than N issues - instead of being calculated on a truncated or oversized selection. Narrow the report JQL or raise the limit.
Reports left hanging by a node restart can be recovered with POST /reportQueue/recover (see For integrations).
See Reports.
WebSudo on the appâs administration pages
The appâs administration screens are now behind Jiraâs WebSudo: Jira asks for your password again when you open them if your administrator session has expired. Nothing to configure - if WebSudo is disabled on your instance, nothing changes for you.
For integrations
Every non-GET REST call needs an XSRF token
Any request to /rest/jibrok/timeinstatus/* that is not a GET must carry X-Atlassian-Token: no-check (or an atl_token parameter in the query string). This includes scripts that authenticate with a Personal Access Token or with Basic auth - the token is required regardless of how the caller is authenticated.
curl -X POST \
-H "Authorization: Bearer $JIRA_PAT" \
-H "X-Atlassian-Token: no-check" \
-H "Content-Type: application/json" \
-d '{"issueKey": "TEST-1", "customFieldId": 10000}' \
"https://jira.example.com/rest/jibrok/timeinstatus/1.0/timerManager/start"
The full rule, the alternatives and the error you get without the header: REST API - XSRF and the X-Atlassian-Token header.
Five settingsForAll* endpoints moved from GET to POST
Five endpoints whose path starts with settingsForAll write settings, and were reachable with a GET. They now accept POST only, with the XSRF token. Their parameters are unchanged. A GET on them stops working after the upgrade - see the method table for how to migrate a call.
Manual control takes a JSON body
start, pause, stop and restart for timers and stopwatches take a JSON body instead of form parameters. The parameters themselves are the same (issueKey or issueId, plus customFieldId), and the request needs Content-Type: application/json alongside the XSRF token. See the request format.
Recovering stuck reports
POST /reportQueue/recover returns reports that were left in the queue - for example by a node restart in the middle of a generation - to a state where they can be generated again. Like every other non-GET call it needs the XSRF token.
500 responses no longer describe the failure
The body of a 500 response is now generic. Integrations that parsed the message out of an error body have to fall back to the response code; the details stay in the Jira log on the server.
Performance
The same review cut repeated work out of the hot paths:
| Path | Before | After |
|---|---|---|
| Issue history read per calculation | 5 reads | 1 read |
| License check per request | 7 checks | 1 check |
realTime* JQL functions |
ran a nested search | no nested search |
You do not have to change any setting to get this - it applies after the upgrade. The performance settings keep the same meaning.
Cloud migration
The migration payload goes to the production Forge environment
The JCMA app migration now delivers the appâs settings to the PRODUCTION environment of the Cloud app - the one installed from the Marketplace. Before 6.4 the payload was addressed to the appâs development environment, so a migration could report success while the production Cloud app stayed empty. Upgrade to 6.4 before you migrate to Cloud.
See What JCMA transfers to Cloud - and what is recalculated from history for the full contents of the transfer and for the MigrationListenerSelfCheck line that tells you whether JCMA can see the app at all.
Fixes
6.4 carries 23 changes from the 2026 Data Center review. The ones that are not listed above do not change how the app is configured or called: they are internal hardening and clean-up of the same code paths. If one of your support cases is waiting on this release, ask support to confirm it is in 6.4.
Upgrade checklist
- Take the build for your Jira line (Jira 10.x or Jira 11.3).
- Add
X-Atlassian-Token: no-checkto every non-GETcall to the app in your scripts, Automation rules and reporting tools - see XSRF and the X-Atlassian-Token header. - Change the five
settingsForAll*calls fromGETtoPOST, and manual control calls to a JSON body. - Replace
0in Count limit with an explicit number. - Open and save the field configurations that have lists on the Value access tab.
- Check that the users who drive counters manually have Browse Projects.
- Re-check report JQL that selects more issues than the limit - those reports now fail instead of running.
- If a Cloud migration is on your roadmap, upgrade before you run JCMA - see What JCMA transfers to Cloud.
- Run a full re-index if you are also coming from a version older than the index change described in Install.

