A system’s perspective – lessons from failures
summary
Reports issue: a report issues thousands of separate SQL reads because invisible fields force one query per cell and one search times out after 20 minute timeout; replacing repeated calls with stored procedures collapses many queries into one and the report loads. An installed plugin calls GC.Collect and triggers 117 second-generation garbage collections every five minutes, forcing a temporary disable and then a vendor fix; memory pressure rises to 128 GB of RAM on servers. Location search logic breaks after a database swap until load tests include location queries.
AI-written summary · about
chapters (6)
- 0:00Start
- 4:53Normal Work
- 5:38Resolving Issues and Production
- 16:18Change the Frequency of Ldap Synchronization
- 33:39How To Measure Something Not on Production
- 35:05What Are the Typical Bottlenecks or Problems
related talks
12mUsing Dynatrace and Kibana to Identify Bad Queries and Optimise Them
19mKeeping High-Traffic Systems Running When Things Go Wrong
01h 09mSherlock Homepage - A detective story about running large web services
30mAvoiding software fails: few metrics to improve application reliability
26mThe Sounds of Silence: Lessons from an 18 hour API outage
30mThe Meat of It
32mTackling Slow Queries: A Practical Approach to Prevention and Correction
37mNavigating Managed Databases for Over a Decade as a Non-DBA