On Mon, Apr 24, 2017 at 8:43 AM, Mark Watson <mark.wat...@jurisconcept.ca> wrote:
> Good day all, > > > > I just noticed an anomaly regarding the logging. I have my logging set up > as follows: > > log_filename = 'postgresql-%d.log' > > log_truncate_on_rotation = on > I don't see "log_rotation_age" and/or "log_rotation_size" here [1] and at least one needs to be set in order to enable actual rotation; the "truncate" option simply tells PostgreSQL what to do when encountering a file with the same name during the rotation process. log_rotation_age apparently has under-documented intelligence since I would expect a server that starts up mid-hour and uses a 60 minute rotation to rotate mid-hour as well so the log would contain 1 hours worth of data but the leading hours would be different. The examples in log_truncate_on_rotation indicate that this isn't the case. I have not tested reality or read the source. This is on Windows 10, 64-bit > > PostgreSQL 9.2.2, compiled by Visual C++ build 1800, 64-bit > > (EnterpriseDB installer) > > > > Note that this is not a major concern on my end; postgres 9.6.2 has > otherwise been running flawlessly. > > > Um...you're reporting a very outdated 9.2 release in the supposed copy-paste job above but claiming 9.6.2 ... [1] https://www.postgresql.org/docs/9.6/static/runtime-config-logging.html#RUNTIME-CONFIG-LOGGING-WHERE David J.
