Commitdc88460

committed

Doc: Review for "Optionally prefetch referenced data in recovery."

Typos, corrections and language improvements in the docs, and a few incode comments too.Reported-by: Justin Pryzby <pryzby@telsasoft.com>Discussion:https://postgr.es/m/20210409033703.GP6592%40telsasoft.com

1 parent49fb4e6 commitdc88460Copy full SHA for dc88460

File tree

+10

-10

lines changed

+10

-10

lines changed

Lines changed: 1 addition & 1 deletion

Original file line number	Diff line number	Diff line change
`@@ -3621,7 +3621,7 @@ include_dir 'conf.d'`
`3621`	`3621`	`pool after that. However, on file systems with a block size larger`
`3622`	`3622`	`than`
`3623`	`3623`	`<productname>PostgreSQL</productname>'s, prefetching can avoid a`
`3624`		`- costly read-before-write whenablocks are later written.`
	`3624`	`+ costly read-before-write when blocks are later written.`
`3625`	`3625`	`The default is off.`
`3626`	`3626`	`</para>`
`3627`	`3627`	`</listitem>`

Lines changed: 1 addition & 3 deletions

Original file line number	Diff line number	Diff line change
`@@ -816,9 +816,7 @@`
`816`	`816`	`prefetching mechanism is most likely to be effective on systems`
`817`	`817`	`with <varname>full_page_writes</varname> set to`
`818`	`818`	`<varname>off</varname> (where that is safe), and where the working`
`819`		`- set is larger than RAM. By default, prefetching in recovery is enabled`
`820`		`- on operating systems that have <function>posix_fadvise</function>`
`821`		`- support.`
	`819`	`+ set is larger than RAM. By default, prefetching in recovery is disabled.`
`822`	`820`	`</para>`
`823`	`821`	`</sect1>`
`824`	`822`

Lines changed: 7 additions & 5 deletions

Original file line number	Diff line number	Diff line change
`@@ -31,12 +31,14 @@`
`31`	`31`	`* stall; this is counted with "skip_fpw".`
`32`	`32`	`*`
`33`	`33`	`* The only way we currently have to know that an I/O initiated with`
`34`		`- * PrefetchSharedBuffer() hasthat recovery will eventually call ReadBuffer(),`
`35`		`- *and perform a synchronous read. Therefore, we track the number of`
	`34`	`+ * PrefetchSharedBuffer() hascompleted is to wait for the corresponding call`
	`35`	`+ *to XLogReadBufferInRedo() to return. Therefore, we track the number of`
`36`	`36`	`* potentially in-flight I/Os by using a circular buffer of LSNs. When it's`
`37`		`- * full, we have to wait for recovery to replay records so that the queue`
`38`		`- * depth can be reduced, before we can do any more prefetching. Ideally, this`
`39`		`- * keeps us the right distance ahead to respect maintenance_io_concurrency.`
	`37`	`+ * full, we have to wait for recovery to replay enough records to remove some`
	`38`	`+ * LSNs, and only then can we initiate more prefetching. Ideally, this keeps`
	`39`	`+ * us just the right distance ahead to respect maintenance_io_concurrency,`
	`40`	`+ * though in practice it errs on the side of being too conservative because`
	`41`	`+ * many I/Os complete sooner than we know.`
`40`	`42`	`*`
`41`	`43`	`*-------------------------------------------------------------------------`
`42`	`44`	`*/`

Lines changed: 1 addition & 1 deletion

Original file line number	Diff line number	Diff line change
`@@ -2774,7 +2774,7 @@ static struct config_int ConfigureNamesInt[] =`
`2774`	`2774`	`{`
`2775`	`2775`	`{"wal_decode_buffer_size",PGC_POSTMASTER,WAL_ARCHIVE_RECOVERY,`
`2776`	`2776`	`gettext_noop("Maximum buffer size for reading ahead in the WAL during recovery."),`
`2777`		`-gettext_noop("This controls the maximum distance we can read aheadn the WAL to prefetch referenced blocks."),`
	`2777`	`+gettext_noop("This controls the maximum distance we can read aheadin the WAL to prefetch referenced blocks."),`
`2778`	`2778`	`GUC_UNIT_BYTE`
`2779`	`2779`	`},`
`2780`	`2780`	`&wal_decode_buffer_size,`

Comments

(0)