Re: Large shared_buffer stalls WAS: proposal: Set effective_cache_size to greater of .conf value, shared_buffers

Josh Berkus <josh@agliodbs.com> schrieb:

>What's notable is that sometimes it's just *one* of the replicas which
>goes into paralysis.  If the master gets this issue though, the
>replicas
>experience it soon afterwards.  Increasing wal_buffers from 16GB to
>64GB
>seems to make this issue happen less frequently, but it doesn't go away
>entirely.  

Youre talking about MB, not GB here, right?

Andres

-- 
Please excuse brevity and formatting - I am writing this on my mobile phone.

Andres Freund                       http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services