You are browsing a read-only backup copy of Wikitech. The live site can be found at wikitech.wikimedia.org
Incident documentation/2021-06-15 Eqsin network
document status: in-review
At 09:23 UTC, alerts indicated connectivity issues to the Eqsin cluster in Singapore. At 09:31 UTC, @Ema deployed a DNS change to depool the Eqsin cluster. This diverted most of its assigned traffic to Ulsfo, and some to Esams. At 09:35 UTC traffic started recovering, with traffic back to regular levels at 09:45 UTC. The 15-minute window is attributed to DNS caches expiring (e.g. at ISPs and on client devices). The connectivy issues were resolved later that day, and at 18:50 UTC @CMooney repooled the Eqsin cluster, with traffic back to regular levels in Eqsin by 19:00 UTC.
Impact: For about 35 minutes from 09:20 to 09:45 UTC, the wikis were largely unreachable from countries normally served by the Singapore DC (including India, Hong Kong, and Japan).
- Wikimedia DNS: DC geo map
- Grafana: Navigation Timing by Continent 2021-06-15
- Grafana: Traffic volume by DC 2021-06-15