Similarly to the post of a couple of weeks back relating about the [new memory allocation routine] (/postgresql-2/postgres-9-5-feature-highlight-allocation-routine-no-oom/) able to give a plan B route in case of OOM, here is a follow-up commit adding more infrastructure in the same area but this time for some widely-used memory allocation routines:

[log_autovacuum_min_duration] (https://www.postgresql.org/docs/devel/static/runtime-config-autovacuum.html#GUC-LOG-AUTOVACUUM-MIN-DURATION) is a system-wide parameter controlling a threshold from which autovacuum activity is logged in the system logs. Every person who has already worked on looking at a system where a given set of table is bloated has for sure already been annoyed by the fact that even a high value of log_autovacuum_min_duration offers no guarantee in reducing log spams of not-much-bloated tables whose autovacuum runtime takes more than the threshold value, making its activity being logged (and this is after working on such a lambda system that the author of this feature wrote a patch for it). Postgres 9.5 is coming with a new feature allowing to control this logging threshold at relation level, feature introduced by this commit:

This week, I wanted to share something that may impact many users of Postgres, with this commit changing a behavior in a binary utility that had for a long time been the same default:

This week the following commit has landed in PostgreSQL code tree, introducing a new feature that will be released in 9.5:

A nice feature extending the usage of [pgbench] (https://www.postgresql.org/docs/devel/static/pgbench.html), in-core tool of Postgres aimed at doing benchmarks, has landed in 9.5 with this commit:

In Postgres, [full-page writes] (https://www.postgresql.org/docs/devel/static/runtime-config-wal.html#RUNTIME-CONFIG-WAL-SETTINGS). which are in short complete images of a page added in WAL after the first modification of this page after a checkpoint, can be an origin of WAL bloat for applications manipulating many relation pages. Note that full-page writes are critical to ensure data consistency in case particularly if a crash happens during a page write, making perhaps this page made of both new and old data.

When working on testing WAL compression, I developed a simple hack able to calculate the amount of CPU used by a single Postgres backend process during its lifetime using getrusage invoked at process startup and shutdown. This thing is not aimed for an integration into core, still it may be useful for people who need to measure the amount of CPU used for a given set of SQL queries when working on a patch. Here is the patch, with no more than 20 lines:

The last months have showed a couple of vulnerabilities in openssl, so sometimes it is handy to get a status of how SSL is used on a given instance. For this purpose, there is a nice tool called sslyze that can help scanning SSL usage on a given server and it happens that it has support for the SSLrequest handshake that PostgreSQL embeds in its protocol (see [here] (https://www.postgresql.org/docs/devel/static/protocol-flow.html) regarding the message SSLrequest for more details).

Up to Postgres 9.4, when a node in recovery checks for the availability of WAL from a source, be it a WAL stream, WAL archive or local pg_xlog and that it fails to obtain what it wanted, it has to wait for a mount of 5s, amount of time hardcoded directly in xlog.c. 9.5 brings more flexibility with a built-in parameter allowing to control this interval of time thanks to this commit:

Not later than this week a bug regarding [pg_dump] (https://www.postgresql.org/docs/devel/static/app-pgdump.html) and compression with zlib when dumping data has been reported [here] (https://www.postgresql.org/message-id/20150217153446.2590.24945@wrigleys.postgresql.org).