Time/Place

Attendees 

Agenda

  1. Updates
  2. Using and Improving on the jmeter tests
    1. ..over LevelDB
    2. ..over MySQL
    3. ..over Postgres?
  3. Large tests have shown slowdown due to garbage collection
  4. Child-count issue:  Unable to locate Jira server for this macro. It may be due to Application Link configuration.

Minutes

Updates

  1. last met a month ago
  2. sprater is available to do some work
  3. elkiss has put in a PR to jmeter
    1. uploading binary files
    2. creating test data on-the-fly (100MB/sec)
    3. suggestion to create test files on disk before ingest
    4. current approach is creating data in-memory
    5. jmeter http-sampler appears to load complete file into memory.. this may not scale
    6. Nick to review PR from elkiss
  4. awoods to add docs for java_opts and GC tooling

Next steps for testing

  1. elkiss has large mem machines
  2. sprater has an 8GB vm .. maybe get it to 32GB
  3. elkiss is interested in looking into other tests
  4. noting that there may be usage patterns of ordering and indirect containers
  5. next call Mar 14th, second Monday of every month

Actions

  1. Ben: maybe memory leak issue
    1. inspecting different LDP models
    2. will do some heap-dump analysis
  2. elkiss:
    1. to flush out more tests
      1. may be able to run tests in the beginning of March
    2. interested in clustering: HA, scale-out perf. as second concern - high ingest
    3. mysql vs. leveldb tests
  3. sprater: fine-tooth comb current test
    1. over next 2-4 weeks,
      1. understanding tests
      2. understanding execution time
      3. reproduce GC error
  4. Nick: to test elkiss PR