High CPU Utilization for PostgreSQL Postmaster Process on Unitrends UEB

[Reid Cooper]

I was looking at the SAR report on my UEB box and noticed that continuously, all day long, the Postmaster process is using a tremendous amount of CPU. Is this normal? Is there something that I should do to adjust this?

Here is the output from top, these are the top three values:

31631 postgres  25   0 1415m 276m  30m R 100.0  2.3   1:10.26 postmaster
31732 postgres  15   0 1165m  10m 9264 R 69.9  0.1   0:02.10 postmaster
31734 postgres  19   0 1331m 498m 332m R 58.2  4.1   0:01.75 postmaster

Those seem like extremely high values.

[Reid Cooper]

If I do a "top -c" instead, this is the additional info that I get:

32336 postgres  25   0 1570m 400m 3248 R 100.0  3.3   0:29.98 postgres: postgres bpdb 127.0.0.1(39763) BIND
32358 postgres  25   0 1334m 1.0g 843m R 99.7  8.4   0:04.38 postgres: postgres bpdb 127.0.0.1(39772) SELECT
32383 postgres  16   0 1165m 8304 6604 R  6.6  0.1   0:00.20 postgres: postgres bpdb 127.0.0.1(39773) SELECT

[Reid Cooper]

You can see that there are not just three hung up processes, there are new processes spawning regularly. The database is just very busy.

[Jaguar]

Yup, pretty normal. Postgres is the database. Hashing never stops, baby.

Unitrends is constantly preparing itself for need to do an autosynth, and is working on the dedup and inline hashing and everything else all the time. It's pretty normal for postsgres to be pegged.