> On Jun 22, 2020, at 8:13 AM, Flaris Feller <flaris.fel...@gmail.com> wrote:
> 
> Hello all,
> 
> When using Postgres 9.6.15 on "CentOS Linux release 7.5.1804 (Core)" on Intel 
> x86_64 I noticed "invalid memory alloc request size" error at PostgreSQL logs.
> This is the postgresq.log file's fragment of log where the error was found.
> 
> 2020-06-22 00:29:18 BRT [16987]: [1-1] db=bxs,user=postgres ERRO:  invalid 
> memory alloc request size 18446744073709551613
> 2020-06-22 00:29:18 BRT [16987]: [2-1] db=bxs,user=postgres COMANDO:  COPY 
> public.cham_chamada....
> 
> PostgreSQL was installed using official PGDG repository through yum.
> This is the list of postgresql.conf settings used in the cluster.
> 
> listen_addresses = '*'
> log_destination = 'stderr' 
> logging_collector = on 
> log_directory = 'pg_log' 
> log_filename = 'postgresql-%a.log' 
> log_truncate_on_rotation = on 
> log_rotation_age = 1d 
> log_rotation_size = 0 
> autovacuum = off
> datestyle = 'iso, mdy'
> lc_messages = 'pt_BR'                   
> lc_monetary = 'pt_BR'                   
> lc_numeric = 'pt_BR'                    
> lc_time = 'pt_BR'                               
> default_text_search_config = 'pg_catalog.portuguese'
> max_connections=150
> shared_buffers=2GB
> effective_cache_size=4GB
> work_mem=13981kB
> maintenance_work_mem=256MB
> log_min_duration_statement = 300000
> log_line_prefix = '%t [%p]: [%l-1] db=%d,user=%u '
> log_lock_waits = on
> timezone='America/Sao_paulo'
> log_timezone = 'Brazil/East'
> min_wal_size = 1GB
> max_wal_size = 2GB
> 
> Looking at previous list's messages I've found this could be a data 
> corruption issue and I've followed the recommended procedures.
> But even removing the corrupted records, doing a vacuum full and re-indexing 
> the table the problem keep going recurrently.
> So I would like some guidance to find the root cause of the table corruption 
> on the database.
> Any help would be appreciated.
> 
> Thanks in advance,
> Flaris Feller.

I there an application generating this value perhaps?
1111111111111111111111111111111111111111111111111111111111111101
Looks a little like a signed/un-signed mismatch


Reply via email to