mited to
hitting this bug (when Valgrind memcheck is used)?
Are you satisfied that I have adequately described steps to reproduce?
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
it for Robert to comment on
what I've said before acting.
We should be more ambitious about adding test coverage to tuplesort.c.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
scussion with Tom.
> I don't know, either. You read both Robert and me indicating that this bug
> fix will be better with a test, and nobody has said that a test-free fix is
> optimal. Here's your chance to obliterate that no-tests precedent.
I'm happy that I can do at least
x27;t add
tests! There is scarcely any example of anyone deliberately adding
test coverage to that file. I don't understand why it is so.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
ck).
Those two cases were not exactly comparable in perhaps several other
ways, but even still my sense is that that this can be at least
partially explained by md.c bottlenecks. This is something that we'll
need to confirm through profiling. Hopefully it's just this one
bottleneck.
On Fri, Jul 1, 2016 at 12:30 PM, Peter Geoghegan wrote:
> Checkout my gensort tool from github. Build the C tool with "make".
> Then, from the working directory:
>
> ./postgres_load.py -m 250 --skew --logged
> psql -c "CREATE INDEX segfaults on sort_test_skew
ess needs to occur following any movement of
the tuple. Whereas, in all other cases there is no need to do
something similar, as it happens.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
mory. Clearly that needs to be
updated if the tuple is moved, since it always points to an offset
into the same tuple. I thought that that explanation was simple.
[1] https://github.com/petergeoghegan/gensort
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.or
involved or
difficult question, then no need to answer now.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
master. Also some comment policing.
Are you planning to pick this up again, Andres?
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
tuples using memmove()), because their particular
variety of caller tuple happens to itself contain a pointer to
palloc()'d memory. Attached patch fixes this use-after-free bug.
--
Peter Geoghegan
From aa68ed9e3f952bc19aa3bc8e31d0e216e13e0fae Mon Sep 17 00:00:00 2001
From: Peter Geoghegan
is to be least important (except for the
> documentation update, but that's pending further from Peter Geoghegan
> about exactly what he thinks needs to be changed).
I got sidetracked on that, in part due to investigating a bug in the
9.6 external sort work. I'll have more informa
27;s really hard to prove. It may be that it doesn't matter there,
because working on fixing that yields obvious benefits. But, as you
suggest, we should consider that that might not always be true. That's
really tricky, and has historically been the kind of thing we've
managed very badly.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
t_putdatum(), and now tuplesort_putindextuplevalues()).
I believe that this has value. All the extra boilerplate code misleads.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
rtainly not
opposed to that.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
#x27;m well aware of that issue. This is the same reason why we don't
have any regression test coverage of external sorts. I don't think
that that's good enough.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make ch
indexes. I broke hash index tuplesort builds in a similar
way at one point, too. The slightest bit of regression test coverage
would have caught either bug, I believe. I think that some minimal
regression tests should be added, because evidently they are needed.
--
Peter Geoghegan
--
Sent via pgsql-
milar considerations around memory management).
--
Peter Geoghegan
From 8b2c0d3b1844caaf144971f488201e806c7fcc5e Mon Sep 17 00:00:00 2001
From: Peter Geoghegan
Date: Thu, 23 Jun 2016 18:51:37 -0700
Subject: [PATCH] Remove dead COPYTUP routine in tuplesort.c
Commit 9f03ca915 made copytup_index() dead
nd for this.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
like work_mem and temp file space. This obviously
> isn't ideal, but it's what could be done by the ship date.
Where would that be documented, though? Would it need to be noted in
the case of each such GUC?
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@p
lity statements. It's just
clear that it's not appropriate to use max_parallel_workers_per_gather
within utility statements, even though something like that will be
needed.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to you
9.6.
I am in favor of having something similar to
max_parallel_workers_per_gather for utility statements like CREATE
INDEX. That will need a cost model, at least where the DBA isn't
explicit about the number of workers to use.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgs
xamine the issue more closely.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
ability is a design
goal that I'd be willing to give up other things for.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
On Wed, May 18, 2016 at 12:01 PM, Peter Geoghegan wrote:
>> I think for 9.6 we just have to document this issue. In the next
>> release, we could (and might well want to) try to do something more
>> clever.
>
> Works for me. You may wish to update comments within fd.c
so disregard max_parallel_degree if you
so choose. My guess is that DBAs will frequently want to do so.
I'm not attached to any of this, but that's what I've come up with to
date. Possibly the index storage parameter concept has baggage
elsewhere, which comes up when we later go to pa
; available slots.
That never occurred to me. I'd say it could be a serious problem, that
should be fixed promptly.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
cores to I use?
Well, it depends on the planner, of course. But when constrained only
by the availability of worker processes, then your example could use 3
cores -- the 2 remaining parallel workers, plus the leader itself.
> Presumably the leader isn't counted towards max_parallel_workers
actually corresponds to
> leader + (N-1) workers internally.
>
> I think we can hope that developers are going to be less confused about that
> than users.
+1.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscript
On Tue, May 31, 2016 at 11:09 AM, Peter Geoghegan wrote:
> The only reason I favor defining parallel_degree = 1
I meant "redefining max_parallel_degree =1 to effectively disable
parallel query", of course.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (
and Oracle (as
long as we match those semantics). Also, I think that number of cores
used is a more important consideration for users than the number of
workers used.
Users will definitely be confused about workers used vs. cores used,
but I don't think that any proposal fixes that.
--
ut
the number of cores being the primary consideration.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
. Similarly,
parallel aggregate does not really have much for the leader process to
do other than act as a worker.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
ate in parallel query does need to be user-visible. Worker
processes are a commodity (i.e. the user must consider
max_worker_processes).
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
; which appears to be a 9.6 regression, presumable fallout from the path
> restructuring.
It's surprising that SQL Smith didn't catch something with such simple
steps to reproduce.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make
ition, was essential for
something like LWLocks. The paper made me aware that it's perfectly
acceptable practice to make informed trade-offs around fairness for
what it calls "latches" (what we'd call LWLocks). Maybe that was
obvious to you, but it wasn't to
/gray/papers/Convoy%20Phenomenon%20RJ%202516.pdf
I could believe that there's a case to be made for per-LWLock fairness
characteristics, which may be roughly what Andres meant.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
are initialized there)
seems a little contrived. A single entry point for nbtsort.c works
better.
Short term, I guess the best solution is to just have a pgrminclude
directive in both files.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make
On Mon, May 23, 2016 at 4:41 PM, Tom Lane wrote:
> Pushed, thanks.
Thanks.
If BTREE_BUILD_STATS needs a "tcopprot.h pgrminclude ignore" within
nbtree.c, then ISTM that the similar include directive within
nbtsort.c ought to receive the same treatment.
--
Peter Geoghegan
--
S
; The important thing is to have at least one buildfarm animal using it,
> which there soon will be.
I manually enable it sometimes. And, I've pointed out bugs that that
would have caught to other hackers during review a couple of times.
That's my view. I'm not going to make a fuss
dfarm animal or two to add the new #define, which is not exactly
> a huge cost.
You also have to be aware of the new thing, which a lot of hackers
will not be, if awareness of COPY_PARSE_PLAN_TREES is anything to go
by.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hacker
The attached patch fixes the BTREE_BUILD_STATS build. Looks like
65c5fcd353a859da9e61bfb2b92a99f12937de3b broke this. That commit was
made back in January, so no backpatch is required.
--
Peter Geoghegan
From 06d2150ff20dfa3e6fbc579a9e41ff3f6487 Mon Sep 17 00:00:00 2001
From: Peter Geoghegan
is what
> to use to enable the debug test. I propose "#ifdef COPY_PARSE_PLAN_TREES"
> since that enables similar sanity checking for other parts of
> backend/nodes/, and we do have at least one buildfarm member using it.
That's what I was thinking, too. No need to keep it separate.
--
Pe
x27;t think it would be
difficult to find a choke point for that. Perhaps there is some
subtlety I've missed, since what I propose seems too obvious. FWIW, I
don't think it would much matter if this debug code was not reliably
executed for every possible SelectStmt. Just limiting it to top
cycles. A noticeable omission has a
cost: it gets noticed. Given this code path is likely to hardly ever
be hit, this mechanical approach seemed best. That's all.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subs
On Sat, May 21, 2016 at 5:03 PM, Peter Geoghegan wrote:
> On Sat, May 21, 2016 at 4:28 PM, wrote:
>> ERROR: XX000: unrecognized node type: 920
>> LOCATION: raw_expression_tree_walker, nodeFuncs.c:3410
>>
>> I expected the query run successfully and return one row wi
ox.
Obviously this will be slower than an equivalent plain pgbench SELECT,
but the difference may be smaller than you expect.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
will be very difficult without better testing.
This idea also enables complementary techniques, like interpolation
search that can degrade to binary search.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.
Non-boolean options would be problematic as well.
While the spelling needs work, I like the general idea.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
erification would be a
somewhat generic operation, with AM-specific code invoked as
appropriate.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
On Wed, May 18, 2016 at 8:52 AM, Jeff Janes wrote:
> How about going with something that says more about why we are doing
> it, rather than trying to describe in one or two words what it is
> doing?
>
> VACUUM (FORENSIC)
>
> VACUUM (DEBUG)
>
> VACUUM (LINT)
+1
--
; clever.
Works for me. You may wish to update comments within fd.c at the same time.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
On Tue, May 17, 2016 at 3:33 PM, Peter Geoghegan wrote:
> Fundamentally, since temporary_files_size enforcement simply
> piggy-backs on low-level fd.c file management, without any
> consideration of what the temp files contain, it'll be hard to be sure
> that parallel workers wil
piggy-backs on low-level fd.c file management, without any
consideration of what the temp files contain, it'll be hard to be sure
that parallel workers will not have issues. I think it'll be far
easier to fix the problem then it would be to figure out if it's
possible to get away
ives no special treatment for parallel query. Enforcement of
temp_file_limit will therefore not work *per-session*.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
On Fri, May 6, 2016 at 9:12 PM, Peter Eisentraut
wrote:
> All committed.
Thanks!
This should no longer be referenced in the 9.6 release notes. It
should just appear in the next batch of point releases. Tom has an
sgml comment in the draft 9.6 release notes to that effect.
--
Peter Geoghe
ay let allocatedMemory go negative for a period, which would now be
broken.
If you did make this exact error, you would not be the first. If it
isn't actually broken, perhaps you should still make this change,
simply on general principle. I'd like to hear other opinions on that,
uld be even better if Jeff's approach to testing
was followed as an example by other people, but I wouldn't bet on it
ever happening. It requires real persistence and deep understanding to
do well.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org
On Thu, May 5, 2016 at 7:42 PM, Tom Lane wrote:
> Peter Geoghegan writes:
>> I think that there could stand to be some consolidation among the
>> items that I authored.
>
> After thinking a bit, I merged all the abbreviated-keys stuff including
> the ordered-set-aggrega
ature as follows:
"""
Improve the speed of sorting of varchar, text, and numeric fields via
"abbreviated" keys (Peter Geoghegan, Andrew Gierth, Robert Haas)
"""
Users only cared that there was a nice optimization added to sorting
that affects only the type
seems to be a lack of urgency about it, and given that it's
moderately complicated, that tends to mean it will keep getting put
off. I did notice that you have an sgml comment about it, but the
wording isn't optimistic.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsq
can shed some light. I focused exclusively on the
correctness of the core mechanism. It simply didn't occur to me to
check for problems of the type you describe, that affect the system
even when the feature is disabled. I didn't check because Kevin is a
committer, that I assumed wouldn
this attitude to
reverting patches. It should not be seen as a big personal failure,
because it isn't. Stigmatizing reverts incentivizes behavior that
leads to bad outcomes.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make cha
On Sat, Apr 30, 2016 at 4:26 PM, Bruce Momjian wrote:
> On Sat, Apr 30, 2016 at 04:23:00PM -0700, Peter Geoghegan wrote:
>> Maybe there is a place to emphasize this change in the release notes.
>> I don't really want to make it about the external sort feature,
>> thoug
does not disadvantage external sorts is as much about enabling
HashAggregates as it is about enabling internal sorts.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
g
the full benefit of the hardware). Getting this right could allow
unsharded Postgres to go a lot further (parallelism is the final part
of this), which could make all the difference for some users.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
changing that, since I think that you
have commensurate skill. But, in order to be successful in these
projects, patience and resolve are very important.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
gt; instead of removing on your say-so.
I don't think that that's an efficient use of anyone's time. If
somebody proposes a patch with functionality that needs to merge HLL
states, then they can add an implementation at that time.
--
Peter Geoghegan
--
Sent via pgsql-hackers maili
On Tue, Apr 26, 2016 at 4:41 PM, Robert Haas wrote:
> I'm not prepared to commit this over the objection offered by Tomas
> Vondra on that thread.
You don't want to remove buggy code that is currently unused simply
because it might be useful to have that functionality in the fu
The function hyperLogLogMerge() is faulty [1]. It has no current
callers, though. I propose that we rip it out, as in the attached
patch.
[1]
http://www.postgresql.org/message-id/CAM3SWZT-i6R9JU5YXa8MJUou2_r3LfGJZpQ9tYa1BYxfkj0=c...@mail.gmail.com
--
Peter Geoghegan
From
xec-over-ssl-connection-can-fail-on
There are numerous problem reports like this that are easily found
using Google. I think that there are probably a variety of unpleasant
interactions and symptoms. Crashes are one rarer symptom, seen in
certain scenarios only (crashes are not described in the link ab
On Fri, Apr 8, 2016 at 11:43 AM, Peter Geoghegan wrote:
> I'll do so soon. I was waiting on Peter E to take me up on the offer.
Attached is a series of patches for each supported release branch. As
discussed, I would like to target every released version of Postgres
with this bugfix.
my complaint is entirely about the
baggage that the term "degree of parallelism" happens to have, and how
effectively that has been managed.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
istent
with that of other major systems. Since the GUC's behavior isn't going
to change now, the terminology should change. I'm fine with that. I'm
not particularly concerned with the specifics of some new terminology,
as long as it's consistent with the idea of auxiliary
In short, the existing
ordering isn't really so arbitrary.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
And so, parallelism
*will* still be used on Postgres. That's a pretty significant
difference.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
ll keep things tolerable for the 1000-database
> case, but the performance regression remains jarring. I think we should not
> release 9.6 with pg_dump performance as it stands today.
As someone that is responsible for many such clusters, I strongly agree.
--
Peter Geoghegan
--
t it
again with fresh eyes, but offhand having no safe interlock for the
well established RecordFreeIndexPage() call for GIN seems like an
implausibly obvious bug.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
ing work_mem/maintenance_work_mem may fail to improve
performance only because sorts then become more I/O bound. When in
doubt, testing is advised. A balance may need to be found, if only to
avoid wasting memory.
Thoughts?
[1] https://queue.acm.org/detail.cfm?id=1563874
--
Peter Geogh
get.
Those went into amcheck's regression tests, and so have yet to be
committed.
I think that your approach there has plenty to recommend it, at least
where requirements are more complicated.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make
On Wed, Apr 20, 2016 at 2:49 PM, Robert Haas wrote:
> Committed. But I think the regression test needs more thought, so I
> left that out.
It would be nice if there was a fuzz testing infrastructure that
verified that parallel plans produce the same answer as serial plans.
--
Peter Geo
not critically important if it gets to
that). Rarely, the message will be "I'm deeply concerned about the
*lasting* repercussions of having merged this patch". And so, yes, I
think that we might want to be clearer about looking for nuances like
that.
[1] http://www.scottaaronson.co
ident I am in my
prediction, and how scary in absolute terms I consider the scariest
patches to be.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
On Thu, Apr 14, 2016 at 4:42 PM, Peter Geoghegan wrote:
> * We should look into using the ucol_nextSortKeyPart() API:
>
> http://userguide.icu-project.org/collation/architecture#TOC-Partial-sort-keys
Another more rich API we could immediately put to good use is the ICU
strcoll() var
pecial strxfrm() (which only actually
needs 8 bytes for abbreviated keys) a lot faster. I'd be interested to
see how your Russian text example does with this extra optimization.
We should not be surprised that this kind of support exists within
ICU, because abbreviated keys are actually quite an o
Attached patch removes obsolete comment from fmgr.c. The code the
comment refers to was already removed by commit
5ea86e6e65dd2da3e9a3464484985d48328e7fe3.
--
Peter Geoghegan
remove_obsolete_comment.patch
Description: Binary data
--
Sent via pgsql-hackers mailing list (pgsql-hackers
aptuple.c:881
The memory appears to be freed -- CLOBBER_FREED_MEMORY stuff does this.
You might try using Valgrind, with --track-origins, which will show
where the memory was first allocated.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make cha
bout that being
"small enough to fit in L2 cache". I'm pretty sure that that's still
true at least one time over with the latest Raspberry Pi model, so it
should be revisited.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
On Wed, Apr 13, 2016 at 1:02 AM, Magnus Hagander wrote:
> It's fairly common to see a lot of "Incomplete startup packet" in the
> logfiles caused by monitoring or healthcheck connections.
I've also seen it caused by port scanning.
--
Peter Geoghegan
--
Sent via
of the
TRACE_SORT crud. That's all.
I'll put it on my long-term TODO list.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
to log_executor_stats, which is
categorized as "Statistics / Monitoring". I guess it isn't worth
changing now, though.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
the
associated crud. Just having that as an option suggests that there is
some possible upside to disabling trace_sort, which is clearly not
true. I will write a patch doing this if there are no objections. I
think this is justifiable as clean-up for 9.6.
--
Peter Geoghegan
--
Sent via pgsql-ha
modules
> have rather little test coverage, so they're poor as benchmarks.
+1.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
t
from this patch.
I am reminded of this blog post, written by a friend and former
co-worker: https://brandur.org/postgres-queues
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
should be moved to the next CF, unless a committer
wants to pick it up today.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
On Fri, Apr 8, 2016 at 11:42 AM, Tom Lane wrote:
> Seems like a reasonable thing to do, but somebody would have to do the
> legwork to produce back-branch patches.
I'll do so soon. I was waiting on Peter E to take me up on the offer.
--
Peter Geoghegan
--
Sent via pgsql-hackers m
On Fri, Apr 8, 2016 at 11:20 AM, Peter Geoghegan wrote:
> That seems reasonable. I'm glad we finally got this done. Thanks.
Are we going to backpatch this?
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscripti
djusted this so that we check for n<0 after SSL_read and
> SSL_write, as you had in the frontend code but not in the backend code.
That seems reasonable. I'm glad we finally got this done. Thanks.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.or
ers to put all
this together in more detail, which I believe has value. It suggests
that increasing work_mem or maintenance_work_mem can have almost no
downside provided you don't need that memory for anything else, which
is true.
I will be glad to see this through. Thanks for your help with this, Rober
tead of just 0?
Because the variable is declared as a raw unsigned long, IIRC. It
hardly matters.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
701 - 800 of 3472 matches
Mail list logo