On Mon, Jun 23, 2014 at 09:28:06PM -0500, Rene Romero Benavides wrote: > It depends on the database transactional activity, > observe how many new wal files are generated during a period equivalent to what > it takes to do your base backup. I would set it to twice that number. Take into > account that a checkpoint is issued at the beginning of the process. If you're > lazy just try setting it to something very high such as 256 or more to prevent > wal files being recycled during the process. > > > 2014-06-23 2:12 GMT-05:00 J rgen Fuchsberger <juergen.fuchsberger@xxxxxx>: > > Hi all, > > Can I do a consistent file-system-level backup using the following > procedure: > > 1) SELECT pg_start_backup(...) > 2) rsync postgres data dir to another server > 3) SELECT pg_stop_backup() > 4) rsync pg_xlog directory > > >From what I understand this should be similar to running pg_basebackup > using the -x parameter, correct? One caveat seems to be that > wal_keep_segments should be set "high enough". Can anybody tell what > "high enough" usually is? I am coming late to this thread, but it seems easier for the user to set archive_command to something meaningful during start/stop backup, and set it to /bin/true at other times. I realize they can't turn archive_mode on/off without a restart. -- Bruce Momjian <bruce@xxxxxxxxxx> http://momjian.us EnterpriseDB http://enterprisedb.com + Everyone has their own god. +
