Frontend memory issue

[zeckli]

Checklist

• Investigate root cause and fix it
• Enlist all relevant things into monitoring plan

Refs

• Discussion
• Discussion
• Report (might related)

Story

• Story: review & planning #418

[robertu7]

scaled up to 4-8 instances from 2-8 last week, no alert until now.

[tx0c]

problems found:

• ElasticBeanstalk reports little visibility in CloudWatch; (for alerts, it only has HealthCount, not http status code reporting,)
• https://auth.matters.news/ still serving, mostly forged non existed urls like /wp/admin attack with 500 internal errors
• Warning: data for page "/..." is 141 kB which exceeds the threshold of 128 kB; looks like have duplicates in > JSON.stringify( JSON.parse(document.getElementById("__NEXT_DATA__").textContent) .props .apolloState .data ) .length
  135162 <= 135KB JSON data
  • for page "/"
  • for page "/tags/[tagId]"
  • for page "/[name]/collections/[collectionId]"
• auth.matters.news triggered updated hostname to apiUrl: 'https://server.matters.news/graphql

[tx0c]

mitigation in thematters/matters-web#4291