Jobleads Releases Review (26.10)
Infrastructure Team

Bot attack

  • Problem: Since the beginning of April, we have faced heavy crawling. It has led to slowdowns.

 

  • Solution: We identified the source of traffic and took countermeasures using Cloudflare.

 

  • Effect: In the last 24h, 1.2 Mo requests were blocked. Night/weekend problems are mostly gone. We identified some weak spots in our infrastructure and planned improvements & load testing.

Redis desync

  • Problem: Due to extensive load and networking problems, some of our Redis clusters ended up in a broken state.

 

  • Solution: We recreated some of the master sets, prepared a command for potential future fixes, and planned infrastructure changes to prevent repetition. We turned on the eviction policy.

 

  • Effect: No Redis Cluster problem since fix.

K8 cluster upgrade

  • Problem: The K8 cluster in Corpex used a version from 3 years ago. Some features were missing.

 

  • Solution: A new cluster was created for Stage, and we migrated most of the services.

 

  • Effect: We will integrate autoscaling with KEDA. It will allow us to scale not only by CPU / RAM usage. All other new K8S features are available too. We learned a lesson before the production cluster update.

More QA machines

  • Problem: We reached a moment when all QA environments are occupied.

 

  • Solution: We added 6 more environments.

Other

  • Recruitment
  • Team support
  • Support for a new deployment scenario in Job Seekers
  • Fix for EKS eviction policy
  • Additional registration and conversion metrics for Grafana
  • Messenger support in Mezzio
  • Data retention policy in Langfuse
  • New blog deployment support
Made with Slides.com