Removing xloginsert_slots?

Started by Michael Paquieralmost 12 years ago5 messages

Michael Paquier

michael.paquier@gmail.com

almost 12 years ago

1 attachment(s)

Hi all,

The undocumented GUC called xloginsert_slots has been introduced by
commit 9a20a9b. It is mentioned by the commit that this parameter
should be removed before the release. Wouldn't it be a good time to
remove this parameter soon? I imagine that removing it before the beta
would make sense so now is perhaps too early... Either way, attached
is a patch doing so...
Regards,
--
Michael

Attachments:

xlog_insert_slot_removal.patchtext/x-diff; charset=US-ASCII; name=xlog_insert_slot_removal.patchDownload

*** a/src/backend/access/transam/xlog.c
--- b/src/backend/access/transam/xlog.c
***************
*** 69,74 **** extern uint32 bootstrap_data_checksum_version;
--- 69,76 ----
  #define PROMOTE_SIGNAL_FILE		"promote"
  #define FALLBACK_PROMOTE_SIGNAL_FILE "fallback_promote"
  
+ /* Number of slots for concurrent xlog insertions */
+ #define XLOG_INSERT_SLOTS		8
  
  /* User-settable parameters */
  int			CheckPointSegments = 3;
***************
*** 85,91 **** int			sync_method = DEFAULT_SYNC_METHOD;
  int			wal_level = WAL_LEVEL_MINIMAL;
  int			CommitDelay = 0;	/* precommit delay in microseconds */
  int			CommitSiblings = 5; /* # concurrent xacts needed to sleep */
- int			num_xloginsert_slots = 8;
  
  #ifdef WAL_DEBUG
  bool		XLOG_DEBUG = false;
--- 87,92 ----
*** a/src/backend/utils/misc/guc.c
--- b/src/backend/utils/misc/guc.c
***************
*** 2078,2094 **** static struct config_int ConfigureNamesInt[] =
  	},
  
  	{
- 		{"xloginsert_slots", PGC_POSTMASTER, WAL_SETTINGS,
- 			gettext_noop("Sets the number of slots for concurrent xlog insertions."),
- 			NULL,
- 			GUC_NOT_IN_SAMPLE
- 		},
- 		&num_xloginsert_slots,
- 		8, 1, 1000,
- 		NULL, NULL, NULL
- 	},
- 
- 	{
  		/* see max_connections */
  		{"max_wal_senders", PGC_POSTMASTER, REPLICATION_SENDING,
  			gettext_noop("Sets the maximum number of simultaneously running WAL sender processes."),
--- 2078,2083 ----
*** a/src/include/access/xlog.h
--- b/src/include/access/xlog.h
***************
*** 192,198 **** extern bool EnableHotStandby;
  extern bool fullPageWrites;
  extern bool wal_log_hints;
  extern bool log_checkpoints;
- extern int	num_xloginsert_slots;
  
  /* WAL levels */
  typedef enum WalLevel
--- 192,197 ----

andres@2ndquadrant.com

almost 12 years ago

In reply to: Michael Paquier (#1)

Re: Removing xloginsert_slots?

Hi,

On 2014-01-29 21:59:05 +0900, Michael Paquier wrote:

The undocumented GUC called xloginsert_slots has been introduced by
commit 9a20a9b. It is mentioned by the commit that this parameter
should be removed before the release. Wouldn't it be a good time to
remove this parameter soon? I imagine that removing it before the beta
would make sense so now is perhaps too early... Either way, attached
is a patch doing so...

I'd rather wait till somebody actually has done some benchmarks. I don't
think we're more clueful about it now than back when the patch went
in. And such benchmarking is more likely during beta, so...

Greetings,

Andres Freund

--
Andres Freund http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services

--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

robertmhaas@gmail.com

almost 12 years ago

In reply to: Andres Freund (#2)

Re: Removing xloginsert_slots?

On Wed, Jan 29, 2014 at 8:22 AM, Andres Freund <andres@2ndquadrant.com> wrote:

Hi,

On 2014-01-29 21:59:05 +0900, Michael Paquier wrote:

The undocumented GUC called xloginsert_slots has been introduced by
commit 9a20a9b. It is mentioned by the commit that this parameter
should be removed before the release. Wouldn't it be a good time to
remove this parameter soon? I imagine that removing it before the beta
would make sense so now is perhaps too early... Either way, attached
is a patch doing so...

I'd rather wait till somebody actually has done some benchmarks. I don't
think we're more clueful about it now than back when the patch went
in. And such benchmarking is more likely during beta, so...

Well, it's either got to go away, or get documented, IMHO.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

andres@2ndquadrant.com

almost 12 years ago

In reply to: Robert Haas (#3)

Re: Removing xloginsert_slots?

On 29. Januar 2014 20:51:38 MEZ, Robert Haas <robertmhaas@gmail.com> wrote:

On Wed, Jan 29, 2014 at 8:22 AM, Andres Freund <andres@2ndquadrant.com>
wrote:

Hi,

On 2014-01-29 21:59:05 +0900, Michael Paquier wrote:

The undocumented GUC called xloginsert_slots has been introduced by
commit 9a20a9b. It is mentioned by the commit that this parameter
should be removed before the release. Wouldn't it be a good time to
remove this parameter soon? I imagine that removing it before the

beta

would make sense so now is perhaps too early... Either way, attached
is a patch doing so...

I'd rather wait till somebody actually has done some benchmarks. I

don't

think we're more clueful about it now than back when the patch went
in. And such benchmarking is more likely during beta, so...

Well, it's either got to go away, or get documented, IMHO.

Yes, all I am saying is that I'd like to wait till things have calmed down a bit, so it actually makes sense to run bigger benchmarks. I don't think removing the option is that urgent.

Andres

--
Please excuse brevity and formatting - I am writing this on my mobile phone.

Andres Freund http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services

--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

simon@2ndQuadrant.com

over 11 years ago

In reply to: Andres Freund (#4)

Re: Removing xloginsert_slots?

On 29 January 2014 20:53, Andres Freund <andres@2ndquadrant.com> wrote:

On 29. Januar 2014 20:51:38 MEZ, Robert Haas <robertmhaas@gmail.com> wrote:

On Wed, Jan 29, 2014 at 8:22 AM, Andres Freund <andres@2ndquadrant.com>
wrote:

Hi,

On 2014-01-29 21:59:05 +0900, Michael Paquier wrote:

The undocumented GUC called xloginsert_slots has been introduced by
commit 9a20a9b. It is mentioned by the commit that this parameter
should be removed before the release. Wouldn't it be a good time to
remove this parameter soon? I imagine that removing it before the

beta

would make sense so now is perhaps too early... Either way, attached
is a patch doing so...

I'd rather wait till somebody actually has done some benchmarks. I

don't

think we're more clueful about it now than back when the patch went
in. And such benchmarking is more likely during beta, so...

Well, it's either got to go away, or get documented, IMHO.

Yes, all I am saying is that I'd like to wait till things have calmed down a bit, so it actually makes sense to run bigger benchmarks. I don't think removing the option is that urgent.

I do not want this removed until we have reasonable evidence that the
correct number is "8", and that it is useful for both small, large and
every other kind of config.

We may find evidence it is useful to be able to alter this in the
field and decide to keep it.

I suggest we maintain a "Request for Beta Tests" list, so people are
aware that they can (and should) test this, but it is not necessarily
functionality we would like to keep in the future.

--
Simon Riggs http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services

--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers