slony1-engine.git
15 years ago1. The update sl_subscribe statement was duplicated,
Steve Singer [Fri, 17 Sep 2010 14:49:43 +0000 (10:49 -0400)]
1. The update sl_subscribe statement was duplicated,
   the exact same SQL was listed twice.
2. Do not delete the sl_subscribe row for the backup node.
   We were finding that a slave could end up with data the
   backup node did not have when it became a master because
   the backup node was ignoring the sync because it didn't
   think it had a provider.

This reverts parts of d50fba024aac79020549973e98a4ea9d5bccdab5
(cherry picked from commit 7970189ec93faf3ee71cf49e97529980bbd219aa)

15 years agoFixes for win32 compiles.
Steve Singer [Mon, 13 Sep 2010 20:46:23 +0000 (16:46 -0400)]
Fixes for win32 compiles.
 -Try to include pgbindir in the path when we invoke the test program to check PQisthreadsafe so libpq.dll can
 be found.
-slon_terminate_worker is not used on win32 (slon does not fork)
-updating expected results(parsestatement test) to match what is currently generated on win32
(cherry picked from commit 6c7c91252f437e9dd5c86b6f2a0f92840c135b6d)

15 years agoFix for bug154.
Steve Singer [Mon, 13 Sep 2010 20:38:25 +0000 (16:38 -0400)]
Fix for bug154.

Call updateRelName from the ddlScript_complete_int function instead of ddlScript_complete
 because we want this function to be called on both subscribers and the origin.

This issue addresses something like renaming a table via EXECUTE SCRIPT where
sl_table will continue to have the old name on subscribers.
(cherry picked from commit dfd8816556017fb0a3316b799accf617907c9b62)

15 years agoReplacing calls to signal with sigaction.
Steve Singer [Mon, 13 Sep 2010 20:37:27 +0000 (16:37 -0400)]
Replacing calls to signal with sigaction.
AIX + Solaris have an implementation of
signal that resets the signal handler after it is called once.

Keep calls to signal() on CYGWIN
(cherry picked from commit 7a6c8c0dced9fed8f806dabe8c7e11fbd8d8b18c)

15 years agoAdditional attempt at bug #40.
Steve Singer [Mon, 13 Sep 2010 20:37:07 +0000 (16:37 -0400)]
Additional attempt at bug #40.

If libpq isn't already in the system shared library search
path directories then it might not be found when configure
tries to run the test program.

Build the path to libpq into the test executable via linker
flags.  The linker flags for this vary based on platform.
The logic was based on logic in postgresql/src/makefiles
in the Postgresql source tree.

Also perform a runtime check to make sure the runtime
libpq is thread-safe.
(cherry picked from commit 2ee266a20be37964b6aa0df1554a0b5998fbba4a)

15 years agoBug 149.
Steve Singer [Tue, 7 Sep 2010 17:11:23 +0000 (13:11 -0400)]
Bug 149.

Fix for when local listen fails to start properly.
It was observed that slon can wait for the local listen cond to be set where the local listener
thread stops on an error. In this case we want to signal the mutex but flag the local listener
as not having started. The main slon thread should then abort.

This patch had previously been reverted due to an improper merge. This
version should have been merged properly.
(cherry picked from commit 69b54a2625384010f4bee355fd49853a56d6cd6c)

15 years ago Fix for Bug 121.
Steve Singer [Tue, 7 Sep 2010 13:36:59 +0000 (09:36 -0400)]
Fix for Bug 121.

    Issue the subscribe request against the origin not the provider.
    We sometimes get situations where the subscribe request goes
    and gets issues on the provider and reaches the origin before
    it reaches the receiver.

    The origin generates the ENABLE_SUBSCRIPTION as being originated
    on the origin.  The ENABLE_SUBSCRIPTION can then reach the
    receiver before the SUBSCRIBE_SET reaches the receiver.
    This causes some of these node -1 errors.
(cherry picked from commit 0b59bd96e8eebbd97778185b7b29dfa696b56927)

15 years agoBug #40 - apply to HEAD
Christopher Browne [Mon, 30 Aug 2010 19:30:28 +0000 (15:30 -0400)]
Bug #40 - apply to HEAD

15 years agoAdd more files ignorable by git because they are generated as parts of builds
Christopher Browne [Fri, 27 Aug 2010 22:26:45 +0000 (18:26 -0400)]
Add more files ignorable by git because they are generated as parts of builds

15 years agoBug #115
Jan Wieck [Tue, 24 Aug 2010 21:17:00 +0000 (17:17 -0400)]
Bug #115

Cause the parser to generate an ERROR if a WAIT FOR EVENT
occurs inside of a TRY block.

In reality, the wait itself isn't the problem. It is the combination
of creating the event (sync, store whatever) and waiting for it
inside the same try block. Try blocks don't commit until they end,
so the event never gets committed and logically, the wait must wait
forever.
(cherry picked from commit 783eee0a78628e62ae081fbebd366f9d3fad5af8)

15 years agoFix for when local listen fails to start properly. It was observed that slon can...
Steve Singer [Tue, 13 Jul 2010 13:02:54 +0000 (09:02 -0400)]
Fix for when local listen fails to start properly. It was observed that slon can wait for the local listen cond to be set where the local listener thread stops on an error. In this case we want to signal the mutex but flag the local listener as not having started. The main slon thread should then abort.
(cherry picked from commit ba234cfe686f33fd10d2af1c557af66f8ec17b02)

15 years agoAdditional bug 132 fixes:
Steve Singer [Fri, 27 Aug 2010 18:17:33 +0000 (14:17 -0400)]
Additional bug 132 fixes:

        The worker will only restart when it can't get the nodelock if
        the worker is in a 'restart' state because it previously
        exited asking its watchdog to restart.

        To faciliate this the watchdog will no longer exec when it needs
        to restart its child but instead just call fork() again.

        Also removing more slon_log instances from the signal handler
        since they are no async safe

Additional fix for bug136:

-Do not use evcommand when calling reshape subscriber since this
function does not generate an event.

15 years agoOnly restart the slon during a reshape if sl_subscribe was
Steve Singer [Tue, 24 Aug 2010 13:28:14 +0000 (09:28 -0400)]
Only restart the slon during a reshape if sl_subscribe was
actually updated.

Bug 136 additional fixes
(cherry picked from commit 3c7b70e40bdb913c65e0c5499bc3ac25268da5d4)

15 years agoChange query looking for the need to "reshape subscription" to
Christopher Browne [Wed, 18 Aug 2010 16:22:00 +0000 (12:22 -0400)]
Change query looking for the need to "reshape subscription" to
return a count of nodes, rather than the list of nodes.

This is effectively a minor performance tweak.

15 years agoIf the nodelock can not be obtained then the worker will keep retrying instead
Steve Singer [Thu, 5 Aug 2010 14:45:35 +0000 (10:45 -0400)]
If the nodelock can not be obtained then the worker will keep retrying instead
of exiting.  Bug # 132 discussses issues where slon can't restart because
of a duplicate node lock.  The watchdog instance of slon can kill the worker
child and restart it.  The issue is that postgresql backend for the old
worker might still be around when the new worker tries to get the nodelock.

Prior to this patch we only had one shot at this, now we will keep retrying
until a user decides to kill this process or the other one goes away
(cherry picked from commit acd46819bad1613764708b138ebcfa895467ac51)

15 years agoInstall the signal handlers after the fork.
Steve Singer [Thu, 5 Aug 2010 15:55:43 +0000 (11:55 -0400)]
Install the signal handlers after the fork.
sighandler is intended to be used by the watchdog instance not the worker

This can be considered part of bug 139 (non async safe functions being called
by a signal handler)
(cherry picked from commit 173e7786ddc8ec2ab2b9299fc991573d34e816ba)

15 years agoReplace usage of alarm() for restarting slon with a KILL of the child
Steve Singer [Wed, 4 Aug 2010 12:17:41 +0000 (08:17 -0400)]
Replace usage of alarm() for restarting slon with a KILL of the child
and a sleep before restarting.

Numerous instances were observed where the parent/watchdog slon instance
would exit on a SIGALRM.  The various POSIX standards define rules about
what functions can be called by a signal handler.  We suspect not following
these guidelines was part of the problem.

15 years agoBug 136
Steve Singer [Fri, 27 Aug 2010 18:28:40 +0000 (14:28 -0400)]
Bug 136
    When a forwarding node goes away a dba might issue a subscribe set to reshape the
    cluster so the receiver gets the data from a different provider.
    It could be the case that the receiver never sees the SUBSCRIBE_SET event
    because it is only listening on events from the origin from the forwarder
    that has gone away.

    If we are reshaping a cluster then slonik will contact the receiver
    directly and run update sl_subscribe so the receiver can
    listen from the correct source.
(cherry picked from commit 0f8305660488d508df6a6ea96bf8ebf57e2e77ea)

Conflicts:

src/backend/slony1_funcs.sql

15 years agoWhen doing a fail node where the failed node is not a provider of the backup node
Steve Singer [Fri, 30 Jul 2010 19:55:17 +0000 (15:55 -0400)]
When doing a fail node where the failed node is not a provider of the backup node
we need to delete any subscribe entries that are used to forward data to the backup
node.  These will not be needed when the backup node becomes the new origin.
Until that happens they get in the way of the FAILOVER_SET message
that will be posted on the 'most ahead node' but iwth an origin of
the old origin from getting to this node.
(cherry picked from commit d50fba024aac79020549973e98a4ea9d5bccdab5)

15 years agoIf a origin node fails a subscriber might
Steve Singer [Thu, 29 Jul 2010 19:04:25 +0000 (15:04 -0400)]
If a origin node fails a subscriber might
not be listening for events from the new origin. The FAILOVER event gets
processed on the subscriber but nothing after might be since the
sl_listen rows are setup so all events from the new origin get routed
through the node that used to be the provider(even if a direct
path from the new origin to the subscriber exists).

On processing the FAILOVER event we will mark the old origin as no
inactive.  RebuildListenEntries will now not remove paths that are
dedundent if the (otherwise) preferred path flows through a disabled
node.
(cherry picked from commit 60566590d683b85733404ef290e6c1823c4c014c)

15 years agoBug 124
Jan Wieck [Fri, 27 Aug 2010 17:00:05 +0000 (13:00 -0400)]
Bug 124

Enforce that EVENT NODE is identical to EXECUTE ONLY ON node if
the latter is specified.

Also restore the session_replication_role in case ddlScript_prepare
changed it to local.
(cherry picked from commit c6de292bbb3fea189cece38d6bf5bf7c22a9fa92)

15 years agoBug #117
Jan Wieck [Wed, 25 Aug 2010 20:41:26 +0000 (16:41 -0400)]
Bug #117
Change logging call to omit the write to stdout if we've got and are using only syslog (use_syslog = 2)
(cherry picked from commit 9eb5bce628d7feafb422c6f4b7e65518c8f1bc22)

15 years agoBug 75
Jan Wieck [Tue, 24 Aug 2010 21:59:48 +0000 (17:59 -0400)]
Bug 75

Call db_commit_xact() at the end of slonik's db_connect() to make
the session settings for datestyle and replication role permanent.
(cherry picked from commit 3d8606973c4ed5790787cb8b87e662edbb00c3ee)

15 years agoFix for bug #138
Steve Singer [Mon, 23 Aug 2010 17:01:55 +0000 (13:01 -0400)]
Fix for bug #138
Fix synatx in autoconf m4 file to define GROFF
(cherry picked from commit b4f202e3f67e66a311a4a32c6392bb2be3bc05a2)

Conflicts:

RELEASE

15 years ago check fd for negatvie values. Bug 104 proper fix
Steve Singer [Mon, 23 Aug 2010 16:56:32 +0000 (12:56 -0400)]
 check fd for negatvie values. Bug 104 proper fix
(cherry picked from commit 65af4b6002e945c3df977d12b8613b98a379113a)

15 years agogenerated SGML should be ignored by git
Christopher Browne [Tue, 17 Aug 2010 17:12:33 +0000 (13:12 -0400)]
generated SGML should be ignored by git

15 years agoRemove reference to logswitch_weekly(), per email from TAKATSUKA Haruka <harukat...
Christopher Browne [Tue, 17 Aug 2010 17:11:49 +0000 (13:11 -0400)]
Remove reference to logswitch_weekly(), per email from TAKATSUKA Haruka <harukat@sraoss.co.jp> 2010-06-24

15 years agoBug #79 - add better documentation about configure options
Christopher Browne [Mon, 16 Aug 2010 21:24:19 +0000 (17:24 -0400)]
Bug #79 - add better documentation about configure options

15 years agoMaking ducttape scripts executable again.
Jan Wieck [Sat, 14 Aug 2010 03:45:01 +0000 (23:45 -0400)]
Making ducttape scripts executable again.

15 years agoRemoved definition of PGAUTODOC. The variable is set in Makefile.global
Jan Wieck [Fri, 13 Aug 2010 19:42:43 +0000 (15:42 -0400)]
Removed definition of PGAUTODOC. The variable is set in Makefile.global
by configure.

15 years agoMerge branch 'master', remote branch 'origin'
Jan Wieck [Fri, 13 Aug 2010 19:34:10 +0000 (15:34 -0400)]
Merge branch 'master', remote branch 'origin'

15 years agoMore removals of references to TABLE ADD KEY
Christopher Browne [Fri, 13 Aug 2010 19:29:48 +0000 (15:29 -0400)]
More removals of references to TABLE ADD KEY

15 years agocleanup event function HTML should be ignored
Christopher Browne [Fri, 13 Aug 2010 19:29:06 +0000 (15:29 -0400)]
cleanup event function HTML should be ignored

15 years agoTABLE ADD KEY was still indirectly referenced in the documentation in several places...
Christopher Browne [Fri, 13 Aug 2010 19:16:25 +0000 (15:16 -0400)]
TABLE ADD KEY was still indirectly referenced in the documentation in several places.  Cleaning that out

15 years agoMerge branch 'master' into bug-146-HEAD
Christopher Browne [Thu, 12 Aug 2010 17:25:20 +0000 (13:25 -0400)]
Merge branch 'master' into bug-146-HEAD

Conflicts:
src/backend/slony1_funcs.sql
tests/test1/generate_dml.sh

Merged HEAD changes

15 years agoMerge branch 'bug148-head'
Christopher Browne [Thu, 12 Aug 2010 17:08:13 +0000 (13:08 -0400)]
Merge branch 'bug148-head'

15 years agoBug 42:
Steve Singer [Wed, 11 Aug 2010 21:41:18 +0000 (17:41 -0400)]
Bug 42:

The problem is that postgresql_autodoc produces XML but the documentation is
processed using SGML.  SGML does not process the empty tag notation <xref .../>
correctly.  Instead it should be <xref ...> without slash.  I have attached a
patch that fixes up the XML to be valid SGML so the documentation comes out
right.

Patch by Peter Eisentraut

15 years agoMerge branch 'master' of ssh://git.postgresql.org/slony1-engine
Christopher Browne [Wed, 11 Aug 2010 21:36:39 +0000 (17:36 -0400)]
Merge branch 'master' of ssh://git.postgresql.org/slony1-engine

15 years agoRemoving admin guide entry for TABLE ADD KEY since the command
Steve Singer [Wed, 11 Aug 2010 21:13:36 +0000 (17:13 -0400)]
Removing admin guide entry for TABLE ADD KEY since the command
no longer exists.
See bug 147
(cherry picked from commit a2c1cd68c242bc8833b599bb7edb8aef730ed27f)

15 years agoBug 147
Steve Singer [Wed, 11 Aug 2010 21:12:23 +0000 (17:12 -0400)]
Bug 147
Remove TABLE ADD KEY from altperl scripts.
Slony-I 2.0.x does not support TABLE ADD KEY so we should not
be generating slonik scripts that try and use it.

Instead generate errors if we find instances where it is required
(cherry picked from commit 9d827a40bf3a1fdaee599628c439870c7d410848)

Conflicts:

RELEASE

15 years ago Fix for bug #127.
Steve Singer [Wed, 11 Aug 2010 21:00:38 +0000 (17:00 -0400)]
Fix for bug #127.
    Generation of syns on leaf nodes (nodes that are not set origins) were disabled some time ago. This reverts
    that commit and re-enables them
(cherry picked from commit 572581c4d80fd95206aed1910fb22d7119aa4be4)

15 years ago Fix for Bug #120.
Steve Singer [Wed, 11 Aug 2010 20:58:56 +0000 (16:58 -0400)]
Fix for Bug #120.
    Call enableNode as part of the CLONE NODE processing. This will start the worker thread for the cloned node.
(cherry picked from commit 0d1a293927805e5728c39f8000031bd698ab49c4)

15 years agoRemove sl_log_1/2 from list of tables to be vacuumed, per comment from Jan in bug #78
Christopher Browne [Wed, 11 Aug 2010 16:29:24 +0000 (12:29 -0400)]
Remove sl_log_1/2 from list of tables to be vacuumed, per comment from Jan in bug #78

15 years agoBug #78
Christopher Browne [Wed, 11 Aug 2010 16:23:52 +0000 (12:23 -0400)]
Bug #78
http://www.slony.info/bugzilla/show_bug.cgi?id=78

"Christopher Browne 2010-06-02 13:20:53 PDT
We should deprecate the "cleanup_deletelogs" parameter in 2.0.5 or so."

15 years agoBug #148
Christopher Browne [Tue, 10 Aug 2010 20:33:29 +0000 (16:33 -0400)]
Bug #148
When ShouldSlonyVacuumTable() asks if a table should be vacuumed, the first test should be:
  Does the table exist altogether???
If not, then there's surely no need to vacuum it
'

15 years agoBug #142 - Minimize security profile
Christopher Browne [Mon, 9 Aug 2010 19:59:19 +0000 (15:59 -0400)]
Bug #142 - Minimize security profile

The two places where superuser access is *truly required* have been
shifted to functions where they could readily be:
 a) Turned into security definers, and
 b) Locked down so only the slony management user(s) can run them

Also added documentation on this to "best practices."

15 years agoAdditional HTML files to be ignored as they are generated from doc tree
Christopher Browne [Mon, 9 Aug 2010 19:58:44 +0000 (15:58 -0400)]
Additional HTML files to be ignored as they are generated from doc tree

15 years ago Fix for bug # 122
Steve Singer [Fri, 6 Aug 2010 18:13:47 +0000 (14:13 -0400)]
Fix for bug # 122

    Print a better error message when using a receiver node id that does not exist
(cherry picked from commit 03db9e0b2c1625aafdaedbf6bf3730c8889cbb2f)

15 years agoBug #141 - removing alterTableRestore()
Christopher Browne [Thu, 5 Aug 2010 16:02:30 +0000 (12:02 -0400)]
Bug #141 - removing alterTableRestore()

- Applying this to HEAD

15 years agoMerge branch 'master' of git://git.postgresql.org/git/slony1-engine into bug-146...
Christopher Browne [Wed, 4 Aug 2010 20:14:00 +0000 (16:14 -0400)]
Merge branch 'master' of git://git.postgresql.org/git/slony1-engine into bug-146-HEAD

Conflicts:
config.h.in

15 years agoInitial swing at removing cleanup_deletelogs
Christopher Browne [Wed, 4 Aug 2010 20:13:19 +0000 (16:13 -0400)]
Initial swing at removing cleanup_deletelogs

15 years agoMake scripts executable
Christopher Browne [Wed, 4 Aug 2010 20:12:33 +0000 (16:12 -0400)]
Make scripts executable

15 years agoChanging version number
Christopher Browne [Wed, 4 Aug 2010 20:11:41 +0000 (16:11 -0400)]
Changing version number

15 years agoMake shell scripts executable - expected by tests
Christopher Browne [Wed, 4 Aug 2010 20:08:21 +0000 (16:08 -0400)]
Make shell scripts executable - expected by tests

15 years agoSet version number to 2.1.0 for HEAD
Christopher Browne [Wed, 4 Aug 2010 20:07:58 +0000 (16:07 -0400)]
Set version number to 2.1.0 for HEAD

15 years agoAdd more generated files to .gitignore
Christopher Browne [Wed, 4 Aug 2010 20:06:49 +0000 (16:06 -0400)]
Add more generated files to .gitignore

15 years agoAdd FAQ describing recent SSL-related problem experienced by Karl
Christopher Browne [Wed, 4 Aug 2010 18:58:31 +0000 (14:58 -0400)]
Add FAQ describing recent SSL-related problem experienced by Karl
Denninger

15 years agoDocumentation updates. Mostly typos and minor corrections.
Steve Singer [Wed, 4 Aug 2010 14:13:48 +0000 (10:13 -0400)]
Documentation updates. Mostly typos and minor corrections.
These changes are based on a list of issues submitted by
Tomonari Katsumata, NTT Open Source Software Center

Thanks

15 years agoFixed some broken logic in failedNode()
Steve Singer [Thu, 29 Jul 2010 15:37:44 +0000 (11:37 -0400)]
Fixed some broken logic in failedNode()

It apparently never supported failing over to nodes that are not
a direct subscriber to the failed node.

Jan
(cherry picked from commit 2705654ec3dcb0381afef591756170c6de8f8ce2)

Conflicts:

src/backend/slony1_funcs.sql

15 years agoUpdating version number to 2.0.4
Steve Singer [Wed, 28 Jul 2010 21:12:32 +0000 (17:12 -0400)]
Updating version number to 2.0.4

No longer generate warning of a unsupported version with Pg 9.0
(cherry picked from commit 2d645b3fa1716d649345e8edbb27b188d9505728)

Conflicts:

config.h.in
src/backend/slony1_funcs.sql
src/slonik/slonik.c

15 years agoBug #95 - http://www.slony.info/bugzilla/show_bug.cgi?id=95
Steve Singer [Wed, 28 Jul 2010 21:09:27 +0000 (17:09 -0400)]
Bug #95 - slony.info/bugzilla/show_bug.cgi?id=95

There are two calls to the postgres function subscribeSet_int(integer, integer,
integer, boolean, boolean) in slony1-engine/src/slonik/slonik.c

Both calls do not pass in the last boolean input (the new OMIT COPY parameter)
and hence you get errors.  We have only seen it when using slonik_store_node
but looking at the code it could also happen in slonik_failed_node.
(cherry picked from commit ba90f82da5778648e53d0d742b1fa89901a794e8)

Conflicts:

src/slonik/slonik.c

15 years agoAdd in OMIT COPY option to SUBSCRIBE SET in support of upgrading from
Steve Singer [Wed, 28 Jul 2010 21:06:12 +0000 (17:06 -0400)]
Add in OMIT COPY option to SUBSCRIBE SET in support of upgrading from
elder Slony-I versions.
(cherry picked from commit dc3f318983213f8fb90eec0545e60dff61f1f55d)

Conflicts:

doc/adminguide/adminscripts.sgml
doc/adminguide/intro.sgml
doc/adminguide/slonyupgrade.sgml
src/backend/slony1_funcs.sql
src/slon/remote_worker.c
src/slonik/parser.y
src/slonik/scan.l
src/slonik/slonik.c
src/slonik/slonik.h
tests/run_test.sh

15 years agoPerform a va_copy and use the copy in the loop that uses va_args.
Steve Singer [Wed, 28 Jul 2010 20:27:19 +0000 (16:27 -0400)]
Perform a va_copy and use the copy in the loop that uses va_args.
You are not allowed to use va_args multiple times.

Fix for bug # 55

Patch submitted by Paul Cannon
(cherry picked from commit f98048352362c990d51a694583bfaa78736cd33b)

Conflicts:

src/slon/misc.c

15 years agoUpdating supportted Pg versions
Steve Singer [Wed, 28 Jul 2010 20:21:22 +0000 (16:21 -0400)]
Updating supportted Pg versions
(cherry picked from commit 1a5b4d6e2df472a7ae0889661ff81f4c8afeb486)

Conflicts:

INSTALL

15 years agoImprove comment about "missing function" - the problem could be that
Steve Singer [Wed, 28 Jul 2010 20:20:19 +0000 (16:20 -0400)]
Improve comment about "missing function" - the problem could be that
libpq wasn't found, not just that you're on some grossly ancient version
of PostgreSQL
(cherry picked from commit 8e510f21fa35358a1b50c5c3c60486c6c56c6fc6)

15 years agoRemove direct uses of yyleng and instead use the function yyget_leng()
Steve Singer [Wed, 28 Jul 2010 20:19:26 +0000 (16:19 -0400)]
Remove direct uses of yyleng and instead use the function yyget_leng()
More recent versions of flex have changed yyleng from an int to a size_t.
By using the function we will let the compiler do the conversion if required
on the function result.

We also add an explicit object for scan.o instead of including it at the
bottom of parser.y because we need to generate and include a header for
the scanner (scan.h) so we have the explicit(and correct) definitions of
the yyyget_leng() function.

A autoconf check to ensure that a version of flex with yyget_leng() is
available has also been added.
(cherry picked from commit b44f967e2c62c60f5a2448ce1c4d5afd878922bf)

Conflicts:

RELEASE
configure
configure.ac
src/slonik/Makefile
src/slonik/parser.y
src/slonik/scan.l
src/slonik/slonik.h
src/slony_logshipper/Makefile
src/slony_logshipper/parser.y
src/slony_logshipper/scan.l

15 years agoBug 140
Steve Singer [Thu, 22 Jul 2010 20:53:14 +0000 (16:53 -0400)]
Bug 140

The perl module is actually called DBD:Pg not DBI:Pg
Reported by Peter Eisentraut

15 years agoFix compile issues on win32
Steve Singer [Mon, 19 Jul 2010 14:29:04 +0000 (10:29 -0400)]
Fix compile issues on win32
Patch by Hiroshi Saito

15 years agoRemove all CVS ID tags obsolete in Git
Christopher Browne [Fri, 16 Jul 2010 22:25:24 +0000 (18:25 -0400)]
Remove all CVS ID tags obsolete in Git

15 years agoRemove configure, as it is a generated file
Christopher Browne [Fri, 16 Jul 2010 16:54:17 +0000 (12:54 -0400)]
Remove configure, as it is a generated file

15 years agoAdd reference to Bucardo-related tool
Christopher Browne [Fri, 16 Jul 2010 15:35:53 +0000 (11:35 -0400)]
Add reference to Bucardo-related tool

Per email from Aleksey Tsalolikhin <atsaloli.tech@gmail.com> on
mailing list.

15 years agoRevise release checklist to indicate Git-oriented procedures.
Christopher Browne [Fri, 16 Jul 2010 15:25:24 +0000 (11:25 -0400)]
Revise release checklist to indicate Git-oriented procedures.

Steps that talk about distclean cleaning out .cvsignore have been
removed, as this seems to be nonsense:

 a) It doesn't seem to be actually done in the makefiles
 b) It's certainly obsolete, as we're not using CVS anymore

15 years agoRevise .cvsignore files --> .gitignore
Christopher Browne [Fri, 16 Jul 2010 15:14:49 +0000 (11:14 -0400)]
Revise .cvsignore files --> .gitignore

In many cases, existing .cvsignore files need only be renamed.

In a few cases, new .gitignore files needed to be created as there
wasn't a .cvsignore file (e.g. - for doc/adminguide, src/parsestatements).

Some were missing material (e.g. - src/slon didn't have enough in it).

15 years agoRevise text about the growth of communications costs to temper wording
Christopher Browne [Fri, 9 Jul 2010 18:40:13 +0000 (18:40 +0000)]
Revise text about the growth of communications costs to temper wording
that has tended to lead to people saying wild things about Slony-I
performance.

15 years agoschemadoc.xml is a generated file, so remove it from SCM
Christopher Browne [Fri, 9 Jul 2010 18:38:24 +0000 (18:38 +0000)]
schemadoc.xml is a generated file, so remove it from SCM

15 years agoChange rules to generate schemadoc.xml automatically, and change "make clean"
Christopher Browne [Fri, 9 Jul 2010 18:37:43 +0000 (18:37 +0000)]
Change rules to generate schemadoc.xml automatically, and change "make clean"
to remove it.

15 years agofixing output message per report by Gurjeet Singh
Steve Singer [Fri, 9 Jul 2010 13:32:05 +0000 (13:32 +0000)]
fixing output message per report by Gurjeet Singh

15 years agoFix script name in help text.
Devrim GUNDUZ [Fri, 9 Jul 2010 07:10:08 +0000 (07:10 +0000)]
Fix script name in help text.

15 years agorelease sl_config_lock before sleeping
Steve Singer [Wed, 7 Jul 2010 14:45:27 +0000 (14:45 +0000)]
release sl_config_lock before sleeping

See bug # 135, slon appeared to be waiting to get a sl_config_lock
that the same slon process was already holding.
slon kept sleeping waiting for a sync to catch up.
Not holding the lock when we sleep should allow
other threads to obtain the lock.

Merged from 2.0

15 years agoLog shipping fixes
Steve Singer [Wed, 7 Jul 2010 14:39:22 +0000 (14:39 +0000)]
Log shipping fixes
-
-Proper fix for checking for the proper return code to indicate no files
are queued for processing. This will allow log_shipper to sleep waiting for
more work when no files are ready for processing.  The previous commit was
the wrong commit.

Merged from 2.0

15 years agoLog shipping fixes
Steve Singer [Wed, 7 Jul 2010 14:33:28 +0000 (14:33 +0000)]
Log shipping fixes
-slony1_dump.sh now generates a .sql file that restores in replica mode
so triggers don't fire.  It will also truncate tables before restoring
to them
-Check for the proper return code to indicate no files are queued for
processing. This will allow log_shipper to sleep waiting for more work
when no files are ready for processing.

Merged from 2.0

15 years agoIn 8.4 and above use TRUNCATE ONLY instead of TRUNCATE when
Steve Singer [Mon, 5 Jul 2010 15:06:04 +0000 (15:06 +0000)]
In 8.4 and above use TRUNCATE ONLY instead of TRUNCATE when
cleaning out tables before a subscription.  TRUNCATE ONLY
behaves as TRUNCTE did in 8.3.

This fixes bug # 118 - subscribing to a set with an inherited
table  where the inheriting table has a lower tab_id than the
parent table deletes all of the data in the parent table

Merged from 2.0 branch

15 years ago-Added slonik_add_node altperl script
Steve Singer [Wed, 30 Jun 2010 14:37:47 +0000 (14:37 +0000)]
-Added slonik_add_node altperl script

-Slonik_drop_node now takes an event node

-Changes to the message printed in slon_kill.

Patch by Gurjeet Singh
Merged from 2.0 branch

15 years agoFix for bug #43
Steve Singer [Wed, 30 Jun 2010 14:04:57 +0000 (14:04 +0000)]
Fix for bug #43
Use get_set more consistently in the perl tools.
Patch by Peter Eisentraut.

Merged from 2.0

15 years agoFix for bug #104
Steve Singer [Mon, 28 Jun 2010 15:46:29 +0000 (15:46 +0000)]
Fix for bug #104
Check for NULL file descriptors before performing actions on them.

Merging from 2.0 STABLE branch

15 years agoMemory leak fixes submittedb by Ulrich Weber
Steve Singer [Tue, 1 Jun 2010 15:15:33 +0000 (15:15 +0000)]
Memory leak fixes submittedb by Ulrich Weber
http://lists.slony.info/pipermail/slony1-hackers/2010-April/000321.html

Merged from REL_2_0_STABLE

15 years agoDocumentation clarification, submitted by singh.gurjeet@gmail.com
Steve Singer [Tue, 25 May 2010 13:29:21 +0000 (13:29 +0000)]
Documentation clarification, submitted by singh.gurjeet@gmail.com

15 years agoClarifying documentation to be clear that 'DROP SCHEMA' must be
Steve Singer [Mon, 17 May 2010 18:21:00 +0000 (18:21 +0000)]
Clarifying documentation to be clear that 'DROP SCHEMA' must be
used to drop the slony schema.

15 years agoThe variable name is wrong (syntax error on execution). Switched to correct
Steve Singer [Mon, 17 May 2010 17:16:29 +0000 (17:16 +0000)]
The variable name is wrong (syntax error on execution).  Switched to correct
variable name.

Merged from REL_2_0_STABLE branch

15 years agoBug # 119
Steve Singer [Thu, 13 May 2010 19:41:40 +0000 (19:41 +0000)]
Bug # 119

cloneNodeFinish() called updateRelOid() that was looking at tab_id in
sl_sequence but the column is named seq_id

Report/patch from Chirag Dave (cdave@ca.afilias.info)

Merge from REL_2_0_STABLE

15 years agoGet rid of *some* references to gborg, and point them to new website.
Devrim GUNDUZ [Mon, 3 May 2010 05:56:17 +0000 (05:56 +0000)]
Get rid of *some* references to gborg, and point them to new website.

15 years agoUpdate supported platforms list, and update mailing list URL
Devrim GUNDUZ [Mon, 3 May 2010 05:52:03 +0000 (05:52 +0000)]
Update supported platforms list, and update mailing list URL

15 years agoThe patch for an issue reported on the mailing list where
Jan Wieck [Fri, 30 Apr 2010 23:15:55 +0000 (23:15 +0000)]
The patch for an issue reported on the mailing list where
tuples with large rows where causing data corruption (reported
as invalid utf8 characters) because the memory was freed while
it was still being used.

This issue was introduced with 1.176.2.8 on Feb 11 2010

15 years agoFix shebang, per #108.
Devrim GUNDUZ [Sun, 18 Apr 2010 09:12:26 +0000 (09:12 +0000)]
Fix shebang, per #108.

15 years agoFix sgml error.
Devrim GUNDUZ [Sat, 10 Apr 2010 12:39:42 +0000 (12:39 +0000)]
Fix sgml error.

15 years agoFixed some obsolete prerequisites.
David Fetter [Thu, 1 Apr 2010 17:26:54 +0000 (17:26 +0000)]
Fixed some obsolete prerequisites.

15 years agoAdd notes about bug #55 to test which exercises the fix
Christopher Browne [Wed, 31 Mar 2010 18:10:16 +0000 (18:10 +0000)]
Add notes about bug #55 to test which exercises the fix

15 years agoSPI_getbinval() was being passed a NULL, which PostgreSQL 8.5 no longer
Christopher Browne [Tue, 16 Feb 2010 16:58:22 +0000 (16:58 +0000)]
SPI_getbinval() was being passed a NULL, which PostgreSQL 8.5 no longer
accepts.  Add isnull variable in the relevant code block so there's a place
to stow the bool result.

Reason for this noted by Alvaro Herrera - thanks!

15 years agoPort fixes in 2.0 to HEAD...
Christopher Browne [Fri, 12 Feb 2010 17:16:05 +0000 (17:16 +0000)]
Port fixes in 2.0 to HEAD...

1.  Inapropos number of parms for insert to confirmations table

2.  Change UPDATEs that break as of 9.0

15 years agoBug #106 - change error message for session configuration failure
Christopher Browne [Thu, 11 Feb 2010 23:09:47 +0000 (23:09 +0000)]
Bug #106 - change error message for session configuration failure