https://bugs.kde.org/show_bug.cgi?id=516861
--- Comment #3 from [email protected] --- Additional observations regarding the intermittent first-attempt failure The issue persists with the following pattern: On multiple update cycles, the first attempt in Discover aborts without showing an explicit error in the GUI. A second attempt immediately afterwards usually completes successfully. Environment: KDE Linux Alpha (immutable, erofs based) Updates via plasma-discover → kde-linux-sysupdated (socket activated) Observed behavior: First attempt: Discover stops the update without GUI error. Sometimes only /v1/updatesize requests appear in the journal. In earlier runs, occasional broken pipe messages were seen. No consistent fatal error in the journal. Second attempt: Update proceeds normally. Example successful log excerpt: Calculated update size: 477843164 GET "/v1/updatesize?version=202603030254" 200 Using erofs as seed kde-linux_202603020254.erofs Plan with holes: 3625 segments for 92748 chunks Local 6504 MiB; 1815 segments | Remote 455 MiB; 1810 segments GET "/kde-linux/sysupdate/v2/kde-linux_202603030254_root-x86-64.erofs" 200 servers are shutting down Server exiting The service exits cleanly after successful completion. Hypothesis: This looks like a race condition between plasma-discover and kde-linux-sysupdated during initial index/cache preparation or connection handling. The first run may terminate early on the client side, while the second run works because index/seed state is already prepared. The issue is reproducible across multiple update cycles. Please let me know if additional debug logs are required. -- You are receiving this mail because: You are watching all bug changes.
