Real-time status of Spalce-hosted services and customer-facing platforms.
Last updated 16 June 2026 at 08:42 UTC
Brief read-only window (~5 min) expected during failover.
A misconfigured WAF rule promoted during a routine policy update started rejecting requests with multipart bodies above 8 MB at our Lagos edge. About 11% of API traffic in the region returned 502 for 38 minutes. We rolled the rule back, drained the affected pop, and added a synthetic check for large-payload requests on every edge. Customers with retries on 5xx saw no data loss.
Queue lag on our Frankfurt webhook worker pool reached 47 minutes at peak after an unexpected burst of compounding retries from a single customer integration backfilling six months of events. No deliveries were lost — all caught up by 18:30 GMT. We've added per-tenant rate caps on the retry path and a circuit breaker that pages on-call when lag crosses ten minutes.
A cache eviction storm on our primary auth Redis cluster pushed median token-refresh latency from 14 ms to 1.2 s for twenty-three minutes during peak EU traffic. No failed authentications, but downstream services saw timeouts. The eviction was triggered by a memory-pressure miscalculation after a node replacement; we've sized the cluster to absorb a one-node failure without crossing the eviction watermark.
Scheduled PostgreSQL minor-version upgrade failed over to the standby cleanly, but DNS propagation to our connection pooler took longer than expected. Eight minutes of read-only mode visible to a small slice of dashboard users. Maintenance window was announced 96 hours in advance. We've cut the DNS TTL on the pooler endpoint from 60s to 5s for future upgrades.
Our upstream MoMo aggregator in Ghana experienced a national outage affecting all merchants. Spalce-routed transactions queued in our retry buffer and replayed automatically as service was restored — no payments were lost. We're adding a secondary aggregator as a hot standby this quarter so that single-provider outages no longer impact customer payment flows.
Subscribe to incident notifications via email or webhook from the link below.
Get notified the moment something changes — by email, webhook, or RSS.
