Baptiste LHOSTE wrote:
> We are still trying to fix our issue and we found following logs :
>
> 2013-01-17 09:55:01 CET LOG: automatic vacuum of table
> "flows.public.agg_t1213_incoming_a6_dst_port_and_proto_f5": index scans: 1
> pages: 0 removed, 136547 remain
> tuples: 0 removed, 4044679 remain
> system usage: CPU 3.21s/5.21u sec elapsed 2005.15 sec
>
> 2013-01-17 10:12:50 CET LOG: automatic vacuum of table
> "flows.public.agg_t1214_outgoing_a1_src_net_f5": index scans: 1
> pages: 59886 removed, 37828 remain
> tuples: 1645338 removed, 3750874 remain
> system usage: CPU 3.01s/4.08u sec elapsed 1015.75 sec
>
> How is it possible that first task took twice the time of the second without any tuples to remove ?
Maybe more of the first table is cached.
Maybe contention on the table or the I/O system in general
varied between the two VACUUMs.
Yours,
Laurenz Albe