I am a newbie user of a small Postgres DB app running on a older Intel server (Dual P3-733, 512MB RAM, 2 x SCSI drive RAID 1 array for the Redhat 7.2 OS, 3 x SCSI drive RAID 5 array for Postgres, log files and the Java based app.
I know this configuration is not optimal as the WAL files arent on their own disk but it has worked fine for 4 months until now.
The client portion of the Java app has now started losing connection to Postgres at exactly 10:00pm several nights a week for 1-3 hr duration. It appears that the WAL files are being written to disk at this time which causes the DB to become unavailable. Once the write to disk is finished all workstations can once again access the Postgres DB.
We have no DBA so I am left with trying to find a solution myself. All of the WAL settings in the Postgres conf file are remarked out so I guess it is using system defaults for fsync wal_buffers etc.
Q1: Why 10:00pm ? I assume there must be a time setting in Postgres somewhere as there are no Cron jobs scheduled in Linux.
Q2: How do I prevent losing connection to Postgres ?
Thanks,
Gruntmann
I know this configuration is not optimal as the WAL files arent on their own disk but it has worked fine for 4 months until now.
The client portion of the Java app has now started losing connection to Postgres at exactly 10:00pm several nights a week for 1-3 hr duration. It appears that the WAL files are being written to disk at this time which causes the DB to become unavailable. Once the write to disk is finished all workstations can once again access the Postgres DB.
We have no DBA so I am left with trying to find a solution myself. All of the WAL settings in the Postgres conf file are remarked out so I guess it is using system defaults for fsync wal_buffers etc.
Q1: Why 10:00pm ? I assume there must be a time setting in Postgres somewhere as there are no Cron jobs scheduled in Linux.
Q2: How do I prevent losing connection to Postgres ?
Thanks,
Gruntmann