abdujaparov schrieb am 24.09.2015 um 14:58:
> IO
> CPU
> WAIT
> CONCURRENCY
> CLUSTER
> CONFIGURATION
> NETWORK
> etc
>
> and for each type of waits the wait event for example
>
> cursor mutex x or hw contention or enq TX contention or undo segment
> extention or log file switch
>
> I mentioned wait classes and wait events of Oracle because I know it.
In general Postgres doesn't expose any "wait events" similar to the way Oracle does it.
And due to the difference in architecture many of them wouldn't exist in Postgres anyway.
e.g. everything around contention in the the shared pool, simply because Postgres doesn't
have the concept of a shared pool.
Also Postgres does not have an "undo tablespace", so there can't be any "undo segment extension"
The same is true for the "temp" tablespace - Postgres doesn't have that either the
way Oracle has it.
Cluster wait events also don't exist in Postgres because there is no RAC
Log file switches also don't exist. The most similar thing would probably be a checkpoint.
You can monitor them, by logging them:
http://www.postgresql.org/docs/current/static/runtime-config-logging.html#GUC-LOG-CHECKPOINTS
Obviously there are similar "waits" as the those that are covered by "enq: XX contention", but
they are not persisted anywhere (no ASH in Postgres), so you can only check this in a limited way
if you monitor pg_locks on a regular basis.
It would be better if you told us what underlying problem you are trying to solve, then
we can maybe suggest the corresponding tools or system views.
