# Confluence PDF Export has changed again - and Forge PDF exports are failing **URL:** \ **Category:** Confluence Cloud\ **Tags:** pdf-export, forge\ **Created:** [September 23, 2026, 11:19pm UTC](https://community.developer.atlassian.com/t/confluence-pdf-export-has-changed-again-and-forge-pdf-exports-are-failing/102857 "2026-09-23T23:19:04Z")\ **Posts on this page:** 19\ **Page:** 1
**Author:** ![Chris\_at\_DigitalRose](https://sea1.discourse-cdn.com/atlassiandeveloper/user_avatar/community.developer.atlassian.com/chris_at_digitalrose/32/41070_2.png) [@Chris\_at\_DigitalRose](https://community.developer.atlassian.com/u/Chris_at_DigitalRose)\ **Post date:** [September 23, 2026, 11:19pm UTC](https://community.developer.atlassian.com/t/confluence-pdf-export-has-changed-again-and-forge-pdf-exports-are-failing/102857/1 "2026-09-23T23:19:04Z")
Hi all, we just noticed today that Confluence PDF export has changed again. There is a brand new UI experience as a popup dialog instead of a separate page, and no countdown clock. Not sure when this change was released. ![image](https://global.discourse-cdn.com/atlassiandeveloper/original/3X/b/c/bc49ab95ece36f5680de6b121b0023199fc447c9.png) And with this change our new Forge native macros are failing to export to PDF. (The PDF export runs for 3+ minutes and the PDF contains a spinner image only for each macro). The old app version (with connect macros) still exports to PDF successfully. Atlassian team, where should we be looking to get notification of breaking changes to the PDF Export component in Confluence? Status Pages are all green… ---
**Author:** ![nathanwaters](https://sea1.discourse-cdn.com/atlassiandeveloper/user_avatar/community.developer.atlassian.com/nathanwaters/32/38195_2.png) [@nathanwaters](https://community.developer.atlassian.com/u/nathanwaters)\ **Post date:** [September 24, 2026, 12:12am UTC](https://community.developer.atlassian.com/t/confluence-pdf-export-has-changed-again-and-forge-pdf-exports-are-failing/102857/2 "2026-09-24T00:12:19Z")
> [@Chris\_at\_DigitalRose](#): > > Status Pages are all green… Genuinely no point in having a status page when they [refuse](https://community.developer.atlassian.com/t/12-hour-api-outage-with-no-incident-report/102601) to report incidents on there. ---
**Author:** ![Chris\_at\_DigitalRose](https://sea1.discourse-cdn.com/atlassiandeveloper/user_avatar/community.developer.atlassian.com/chris_at_digitalrose/32/41070_2.png) [@Chris\_at\_DigitalRose](https://community.developer.atlassian.com/u/Chris_at_DigitalRose)\ **Post date:** [September 24, 2026, 10:15pm UTC](https://community.developer.atlassian.com/t/confluence-pdf-export-has-changed-again-and-forge-pdf-exports-are-failing/102857/3 "2026-09-24T22:15:47Z")
Update. Tested again today, and the PDF export worked on 2 sites with small pages containing 1 or 2 macros on the page. But a test page with 17 macros failed to render (spinner only), after taking 90 seconds to return the PDF. Note: We using the return null for PDF that Atlassian came up with as a ‘solution’, so it may be related to: [PDF Export Fails - something went wrong](https://community.developer.atlassian.com/t/pdf-export-fails-something-went-wrong-on-pages-with-many-forge-macros-whose-adfexport-returns-null-for-pdf/102886) which reports more problems with multiple macros per page. PS I do miss the PDF time counter that was removed from the new UX. The static modal feels slower somehow… ---
**Author:** ![VladymyrPostnov](https://avatars.discourse-cdn.com/v4/letter/v/439d5e/32.png) [@VladymyrPostnov](https://community.developer.atlassian.com/u/VladymyrPostnov)\ **Post date:** [September 28, 2026, 12:22pm UTC](https://community.developer.atlassian.com/t/confluence-pdf-export-has-changed-again-and-forge-pdf-exports-are-failing/102857/4 "2026-09-28T12:22:38Z")
Chris, I got curious and checked the Atlassian status pages for 22-26 Sept. Nothing about Confluence at all, and nothing mentioning PDF since August. Your outage on 21 Aug did get an incident, but it went up at 07:16 UTC. From your post that was about 3 hours in. I’ve started collecting cases like this next to what Atlassian actually posted, Nathan’s 12h one is the first. Mind if I add yours? Rough start and end time is enough. ---
**Author:** ![WingSzeNg](https://avatars.discourse-cdn.com/v4/letter/w/34f0e0/32.png) [@WingSzeNg](https://community.developer.atlassian.com/u/WingSzeNg)\ **Post date:** [September 28, 2026, 5:54pm UTC](https://community.developer.atlassian.com/t/confluence-pdf-export-has-changed-again-and-forge-pdf-exports-are-failing/102857/5 "2026-09-28T17:54:20Z")
Chris, can you open a ticket at [https://developer.atlassian.com/support](https://developer.atlassian.com/support)? We asked our internal PDF team investigated this but the issue wasn’t reproducible. Please open a support case and allow them access to your site so the team can investigate further. Thanks. ---
**Author:** ![Chris\_at\_DigitalRose](https://sea1.discourse-cdn.com/atlassiandeveloper/user_avatar/community.developer.atlassian.com/chris_at_digitalrose/32/41070_2.png) [@Chris\_at\_DigitalRose](https://community.developer.atlassian.com/u/Chris_at_DigitalRose)\ **Post date:** [September 28, 2026, 10:39pm UTC](https://community.developer.atlassian.com/t/confluence-pdf-export-has-changed-again-and-forge-pdf-exports-are-failing/102857/6 "2026-09-28T22:39:09Z")
@WingSzeNg thank you for reaching out. Tried PDF export again today and the first attempt at an x20 export never returned. But then 3 more attempts at x20 macros worked well, returning in ~40 seconds. We’ve been struggling with the inconsistency of PDF exports on Forge for months. 6 weeks ago we went through a series of tests after switching to the ADF null Atlassian workaround, and we thought everything was working and stable. And now last week there is brand new UX and the exports are failing/slow again. It feels like there is a continuous stream of unannounced changes for the Confluence PDF export. Where are these changes published so we (as partners) know when to retest? We are just looking for a bit of a heads up when major changes are made so we can help verify that our apps still work… ---
**Author:** ![WingSzeNg](https://avatars.discourse-cdn.com/v4/letter/w/34f0e0/32.png) [@WingSzeNg](https://community.developer.atlassian.com/u/WingSzeNg)\ **Post date:** [September 30, 2026, 10:58pm UTC](https://community.developer.atlassian.com/t/confluence-pdf-export-has-changed-again-and-forge-pdf-exports-are-failing/102857/7 "2026-09-30T22:58:36Z")
Hi Chris — thank you for the additional detail. I’m sorry for the uncertainty and repeated retesting this has caused. The [Confluence public release notes](https://community.atlassian.com/release-notes?products=CONFLUENCE&limit=20) are the intended place to follow Confluence changes and their rollout schedule. In this case, the new export dialog was treated as UI polish and wasn’t considered for broader communication. Your feedback shows that even a UI-focused change can prompt partners to retest critical integrations, and the owning team is notified and they will take that into account when considering communications for future changes. The PDF engineer investigated but they haven’t established that the UX change caused the Forge PDF export failures. To investigate that properly, please open a case through [Developer Support](https://developer.atlassian.com/support/) and, if possible, provide site access, affected page IDs, approximate export timestamps, macro counts, and examples of successful and failed attempts. That will let the team correlate your tests with service logs and determine what is actually failing. Thank you again for raising both the reliability problem and the communications gap. ---
**Author:** ![Chris\_at\_DigitalRose](https://sea1.discourse-cdn.com/atlassiandeveloper/user_avatar/community.developer.atlassian.com/chris_at_digitalrose/32/41070_2.png) [@Chris\_at\_DigitalRose](https://community.developer.atlassian.com/u/Chris_at_DigitalRose)\ **Post date:** [September 30, 2026, 11:06pm UTC](https://community.developer.atlassian.com/t/confluence-pdf-export-has-changed-again-and-forge-pdf-exports-are-failing/102857/8 "2026-09-30T23:06:44Z")
Thanks @WingSzeNg, we will raise a case when we have a reproducible item. But obviously more happened than just the UX change. And unsolicited UX feedback from our team. Losing the timer from the PDF generation feels like a downgrade and makes it feel slower. ---
**Author:** ![VladymyrPostnov](https://avatars.discourse-cdn.com/v4/letter/v/439d5e/32.png) [@VladymyrPostnov](https://community.developer.atlassian.com/u/VladymyrPostnov)\ **Post date:** [October 1, 2026, 5:01pm UTC](https://community.developer.atlassian.com/t/confluence-pdf-export-has-changed-again-and-forge-pdf-exports-are-failing/102857/9 "2026-10-01T17:01:54Z")
If a UI change can break an export workflow, the release notes are the wrong bar for it. The practical problem is that there was no way to tell, on the day it happened, whether the new dialog was a deliberate rollout or a fault. Would it be possible to list export dialog changes in the developer changelog as well, even when the backend is untouched? ---
**Author:** ![VladymyrPostnov](https://avatars.discourse-cdn.com/v4/letter/v/439d5e/32.png) [@VladymyrPostnov](https://community.developer.atlassian.com/u/VladymyrPostnov)\ **Post date:** [October 1, 2026, 5:02pm UTC](https://community.developer.atlassian.com/t/confluence-pdf-export-has-changed-again-and-forge-pdf-exports-are-failing/102857/10 "2026-10-01T17:02:45Z")
The missing progress timer is more than cosmetic for us too. Without it there is no way to tell a slow export from a stalled one, so the only option is to wait and guess. When you do open the case, it would help to know whether your failures were on large pages specifically, since that is the one pattern we could compare against. ---
**Author:** ![Chris\_at\_DigitalRose](https://sea1.discourse-cdn.com/atlassiandeveloper/user_avatar/community.developer.atlassian.com/chris_at_digitalrose/32/41070_2.png) [@Chris\_at\_DigitalRose](https://community.developer.atlassian.com/u/Chris_at_DigitalRose)\ **Post date:** [October 1, 2026, 6:28pm UTC](https://community.developer.atlassian.com/t/confluence-pdf-export-has-changed-again-and-forge-pdf-exports-are-failing/102857/11 "2026-10-01T18:28:12Z")
Hi @VladymyrPostnov. Anecdotally we did see more failures and timeouts last week where we were testing 5+ macros on the same page then with pages with fewer macros. Our macros are small and inline so the Confluence pages themselves are not large, rendered PDF’s are rarely more than 2 or 3 pages long. Chris ---
**Author:** ![VladymyrPostnov](https://avatars.discourse-cdn.com/v4/letter/v/439d5e/32.png) [@VladymyrPostnov](https://community.developer.atlassian.com/u/VladymyrPostnov)\ **Post date:** [October 2, 2026, 11:42am UTC](https://community.developer.atlassian.com/t/confluence-pdf-export-has-changed-again-and-forge-pdf-exports-are-failing/102857/12 "2026-10-02T11:42:06Z")
Thanks Chris, that is a useful split. If it tracks with macro count and not page size or PDF length, that points at a per-macro render step hitting a concurrency or timeout cap, rather than anything to do with payload size. Two things that would narrow it down. When it fails, do you get an outright export error, or a PDF that comes back with blank or placeholder slots where the macros should be? And does the same page succeed if you just retry it a few minutes later? If retries fix it, that is a load pattern. If the same page fails every time at 5 macros and passes at 4, that is a hard cap and much easier to take to Atlassian. ---
**Author:** ![WingSzeNg](https://avatars.discourse-cdn.com/v4/letter/w/34f0e0/32.png) [@WingSzeNg](https://community.developer.atlassian.com/u/WingSzeNg)\ **Post date:** [October 6, 2026, 5:21pm UTC](https://community.developer.atlassian.com/t/confluence-pdf-export-has-changed-again-and-forge-pdf-exports-are-failing/102857/13 "2026-10-06T17:21:07Z")
Thank you for sharing the feedback. I have passed along both points—the feedback on the PDF export progress bar behavior and the request for better visibility in the developer changelog—directly to the relevant engineering and platform teams. We completely understand how frustrating it is when underlying platform changes or refactoring unexpectedly impact your apps. Improving communication and visibility around changes that impact the developer ecosystem is something we take seriously, and we are taking this feedback on board. ---
**Author:** ![VladymyrPostnov](https://avatars.discourse-cdn.com/v4/letter/v/439d5e/32.png) [@VladymyrPostnov](https://community.developer.atlassian.com/u/VladymyrPostnov)\ **Post date:** [October 6, 2026, 5:52pm UTC](https://community.developer.atlassian.com/t/confluence-pdf-export-has-changed-again-and-forge-pdf-exports-are-failing/102857/14 "2026-10-06T17:52:55Z")
Thanks for passing it on. One thing worth connecting: today’s changelog entry about static macros moving to Preview says static macro content is now automatically included in PDF and Word exports, which means the export pipeline itself has been under active change. Can you confirm whether that work touched the regular PDF export path in the window when the failures started? If it did, a line in the changelog saying so would have saved everyone here a week of guessing. ---
**Author:** ![VladymyrPostnov](https://avatars.discourse-cdn.com/v4/letter/v/439d5e/32.png) [@VladymyrPostnov](https://community.developer.atlassian.com/u/VladymyrPostnov)\ **Post date:** [October 11, 2026, 9:13pm UTC](https://community.developer.atlassian.com/t/confluence-pdf-export-has-changed-again-and-forge-pdf-exports-are-failing/102857/15 "2026-10-11T21:13:15Z")
Chris, there are numbers on this now in 102886. Rolf Eriksen and Mihai got a near constant stop at about 240 s with a Custom UI macro whose adfExport returns null for pdf, failing at 12 and 18 instances while 1 and 8 passed, and the same pages exported in roughly 35 s once adfExport was dropped or returned real ADF. So it reads as a timeout on the null fallback path rather than a cap on macro count. It is filed as ECOHELP-169006. If your macro declares adfExport only to return null for pdf, dropping it should get you the native render, and that is the first thing worth testing. ---
**Author:** ![Chris\_at\_DigitalRose](https://sea1.discourse-cdn.com/atlassiandeveloper/user_avatar/community.developer.atlassian.com/chris_at_digitalrose/32/41070_2.png) [@Chris\_at\_DigitalRose](https://community.developer.atlassian.com/u/Chris_at_DigitalRose)\ **Post date:** [October 13, 2026, 5:57am UTC](https://community.developer.atlassian.com/t/confluence-pdf-export-has-changed-again-and-forge-pdf-exports-are-failing/102857/16 "2026-10-13T05:57:04Z")
@VladymyrPostnov thank you for additional test info. A lot of our customers use the MS Word or email exports so unfortunately we need to keep adfExport enabled and depend on the null hack to get Native PDF. Maybe Atlassian will reconsider adding a manifest option for output type equivalent to what connect could easily do. It would be far more efficient if the manifest could say pdfExport=native. The macros could then be simply rendered in puppeteer instead of having to wait to spin up all these lambda’s that say if type=pdf return null. ---
**Author:** ![VladymyrPostnov](https://avatars.discourse-cdn.com/v4/letter/v/439d5e/32.png) [@VladymyrPostnov](https://community.developer.atlassian.com/u/VladymyrPostnov)\ **Post date:** [October 13, 2026, 2:25pm UTC](https://community.developer.atlassian.com/t/confluence-pdf-export-has-changed-again-and-forge-pdf-exports-are-failing/102857/17 "2026-10-13T14:25:26Z")
That makes sense, Word and email exports need the ADF path, so dropping it is not an option for you. A manifest flag like pdfExport=native would skip the lambda round trip completely. Is there an ECO or DEVHELP ticket for that request yet? If you share the key I can add the 240 s numbers from 102886 to it. ---
**Author:** ![Chris\_at\_DigitalRose](https://sea1.discourse-cdn.com/atlassiandeveloper/user_avatar/community.developer.atlassian.com/chris_at_digitalrose/32/41070_2.png) [@Chris\_at\_DigitalRose](https://community.developer.atlassian.com/u/Chris_at_DigitalRose)\ **Post date:** [October 14, 2026, 3:18am UTC](https://community.developer.atlassian.com/t/confluence-pdf-export-has-changed-again-and-forge-pdf-exports-are-failing/102857/18 "2026-10-14T03:18:52Z")
This pdf native rendering was raised as an issue under connect equivalence under a Forge ticket: [Jira](https://ecosystem.atlassian.net/browse/FRGE-2093) Atlassian resolved it with the return null “workaround” instead of changing the manifest to bypass adf for pdf. It wasn’t ideal or efficient with the wasted function callbacks, but it worked (at low macro volumes) and we were able to use this approach for forge. I am not aware of any follow-up tickets to improve upon this. Maybe the instability now documented in the current pdf export could anchor a fix in the manifest. Static declaration of native PDF rendering (or not) of macros would simplify the puppeteer chromium logic. ---
**Author:** ![VladymyrPostnov](https://avatars.discourse-cdn.com/v4/letter/v/439d5e/32.png) [@VladymyrPostnov](https://community.developer.atlassian.com/u/VladymyrPostnov)\ **Post date:** [October 14, 2026, 9:17am UTC](https://community.developer.atlassian.com/t/confluence-pdf-export-has-changed-again-and-forge-pdf-exports-are-failing/102857/19 "2026-10-14T09:17:26Z")
That return null workaround explains a lot, thanks. If the export now breaks above a certain macro count, that would be a strong case for the manifest flag you describe. Do you know roughly how many macros on a page it takes before the export starts failing?