Jump to content

Topic on Talk:PoolCounter/Flow

consistent hashing vs. primary-secondary server nodes

3
Adamw (talkcontribs)

I noticed that most of WMF's PoolCounter clients are set up to use a single server, and we were recommended to configure our client to use a primary server, then fallback to a secondary server if needed. This contradicts the documentation here for the servers variable, which mentions consistent hashing.

What is the preferred client implementation, primary with a fallback or consistent hashing across the two servers?

Legoktm (talkcontribs)
Adamw (talkcontribs)

Great, thanks for looking! I never should have doubted the documentation ;-)