Sympa HR Outage History

**What happened** * We experienced delays in updating data in Sympa. This meant that when new information was added—such as a new employee—it did not always appear in the system immediately. Some updates were processed within seconds or minutes, while others took longer than expected. Additionally, some customers may have experienced slowness when using reporting features. **Why it happened** * The delays were caused by an unusually high load in the system. This increased volume of updates created a backlog, meaning that new information had to wait longer than usual before being processed and displayed. **What we did** * Increased system capacity to handle the higher load * Scaled up our reporting infrastructure to improve performance * Monitored the system closely to reduce the backlog of updates * Continued to track performance to ensure the situation stabilized

**What happened** During the night of April 9–10, 2026, the calculation engine ran with an updated version that included stricter formula validation. This stricter validation caused formulas that previously produced results to now yield empty values in the report database. Specifically, three categories of formulas were affected: * Formulas that used the `+` operator for string concatenation — all `+` operations were treated as arithmetic * Formulas that compared fields against values of incompatible types \(e.g. boolean compared to null, approval compared to a number\) — previously the engine silently ignored these, now it rejected them * Formulas that contained inline comments using the `#` character **Why it happened** We upgraded the calculation engine to improve diagnostics and enable deeper investigation of calculation issues. The new engine included stricter type validation rules. While the intent was to improve robustness, the stricter rules rejected formulas that the previous engine had accepted. These edge cases were not sufficiently covered by the existing regression tests and were not caught before deployment. **What we did** * **Apr 10, 11:57 EEST** — Root cause identified: three categories of formulas were failing under the new stricter validation * **Apr 10, 15:29 EEST** — Fix deployed by rolling back the calculation services to the previous version; all reported cases corrected * **Apr 10, 21:21 EEST** — Fix confirmed stable after monitoring; incident resolved ‌ **The fixes for the identified edge cases and dedicated regression tests are planned as follow-up work.**

**What happened** On April 1, 2026, starting at 09:07, customers experienced delays in seeing their data after saving. **Why it happened** The issue was caused by a limitation in the system responsible for handling background processing. It reached its capacity and could not process new and ongoing operations efficiently. In addition, the system did not release memory quickly enough during high usage, which caused delays before saved data was applied. **What we did** We took the following actions to resolve the issue: * Improved how the system releases resources, making it more responsive during high usage * Optimized memory usage to improve overall stability These changes were fully implemented by 10:43, and the disruption was fully resolved by 12:34.

**What happened:** * Customers experienced issues with the eSign \(Scrive\) service, preventing some signing requests from being processed successfully * The incident was time-limited but resulted in a number of affected documents requiring manual review **Why it happened:** * The root cause was identified as a configuration error on our partner’s \(Scrive\) side * The issue was resolved once identified, restoring normal service **What we did:** * Worked closely with our partner to identify and resolve the issue as quickly as possible * Monitored the service to ensure stability after the fix was deployed * Identified affected documents and provided customers with a tool to efficiently review and reprocess them where needed

**What happened** On March 23rd, some customers experienced an increased number of error messages \("Something unexpected happened"\) and slower system performance. **Why it happened** The issue was caused by an unusually high number of internal requests being processed at the same time. This led to temporary congestion in one of our backend services, which impacted system stability. **What we did** We resolved the situation by increasing system capacity, which restored normal performance. We are now working on improvements to prevent similar situations in the future. These include better handling of unnecessary requests, improved monitoring, and optimizations to reduce system load during peak times. We apologize for any inconvenience caused and appreciate your patience.

**What happened** * Table rows -copy option has disappeared, preventing the copying of tablerow -data to new row. **Why it happened** * This issue was inadvertently introduced to production with the update deployed last week \(week 10 2026\). **What we did** * We have fixed the issue, and added additional tests to prevent this situation from occurring again.

#### What happened Some users occasionally experienced slower than usual loading times and "content not found" errors when accessing profile pages. #### Why it happened We have identified a suspected issue with an underlying caching component that may be causing delays and occasional failures when loading profile page content. Investigation is ongoing. #### What we did * Identified a potential root cause in the caching layer * Working on migrating to an improved caching solution to address the issue * Continuing to monitor to ensure the issue is fully resolved

**What happened** * CDN problems caused customers not being able to login. **Why it happened** * Azure related issue, detected. **What we did** * A fix planned that will solve these issues in the future. This cdn issue is already known in our frontend team.