Re: Gigantic load average spikes

On 3/31/19 8:23 AM, rihad wrote:
> Postgres 10.3 On freebsd 10.3 is almost idle, disk i/o about 5-10%, 
> number running processes about 1-3, cpu about 90% idle, then we run a 
> i/o heavy job like "refresh materialized view", cpu & disk i/o still not 
> maxed out, all of a sudden the number of running processes increases to 
> about 250-300 for a second, which increases load averages that we 

Do you know what the processes are?

Are there any messages in the Postgres log from that time period that 
might shed light?

> periodically poll to see if everything's alright. Why is that? Disk i/o 
> is far from being maxed out, the actual number of running processes is 
> really small.

You are referring to the between spikes process count?

> 
> 
> 
> 


-- 
Adrian Klaver
adrian.klaver@aklaver.com