about your DNS
setup; or maybe this is a plain old bug that needs to be reported to
Apple.
regards, tom lane
---(end of broadcast)---
TIP 1: subscribe and unsubscribe commands go to [EMAIL PROTECTED]
There's not enough information here to give any useful answer.
A replicatable test case would be nice, or at least show us the queries
that are taking a long time and the table schemas they are working on.
regards, tom lane
---(end of broadcast
not
on one of those platforms.
regards, tom lane
---(end of broadcast)---
TIP 4: Don't 'kill -9' the postmaster
AFAIR.
Vadim, what do you think?
regards, tom lane
---(end of broadcast)---
TIP 4: Don't 'kill -9' the postmaster
except for case, since the Postgres parser is usually
case-insensitive.
regards, tom lane
MAX_MESSAGE_LEN - 2);
return 0;
}
Possibly the documentation needs updated, but all I can find is:
: Caveats
:
: The query buffer is 8192 bytes long, and queries over that length will
: be rejected.
Is there another place that says the wrong thing?
regards, tom lane
re release.
regards, tom lane
.
regards, tom lane
. Assuming you want to leave the
old one where it is, just make sure you specify non-default settings
for --with-pgport and --prefix while configuring the new code.
regards, tom lane
script read the file...
regards, tom lane
nfiguration data.
Good point, but it was the second item on the message
override line:
101 INFO "Server started"
A
-+
Oops, I missed that...
regards, tom lane
REATE USER
doesn't undo the changes to the flat file, either.
So, unless someone is feeling inspired to go rework the way the pg_shadow
stuff is handled, I don't think it's a good model to emulate.
regards, tom lane
early
in the 7.1 cycle instead, and see if anyone reports problems.
regards, tom lane
?) pg_dump against a 7.0 database. The older
pg_dump doesn't do any version checking, and the first it realizes
it's got trouble is when it hits incompatibilities in the system
table layout.
Check your PATH to see where pg_dump is coming from...
regards, tom lane
to punctuation.
That's pretty bizarre (not to say difficult to believe). What LOCALE
setting are you running the postmaster in?
regards, tom lane
be
a planner bug), and the ordering you're getting is just whatever happens
to be in the underlying table. Does EXPLAIN show that the query is
being done with an explicit sort?
regards, tom lane
ch is quite possible if
the executable and .bki files don't match.
In any case, it's a little bit silly to be installing 6.5.1 today.
I'd suggest removing all trace of Postgres from your machine and then
making a fresh installation of Postgres 7.0.2 (or wait a couple days
for 7.0.3).
regards, tom lane
it
with one --- no -S switch, send stdout and stderr to a file ---
and do the dump again.)
regards, tom lane
to get to 1000 ...
regards, tom lane
ke install
regards, tom lane
is that
-N and -B are postmaster switches not backend switches. They don't go
into the argument of -o.
regards, tom lane
Dronamraju Rajesh [EMAIL PROTECTED] writes:
Is it a bug in memory management of postgres
Possibly, but you're going to have to get a lot more specific about
what you're doing and seeing before anyone will be able to tell.
regards, tom lane
ing around?). Look for leftover processes, and also see ipcs
and ipcrm to clean up the old shmem segment by hand (or try our
ipcclean script).
regards, tom lane
client is disconnecting without bothering
to send the "I'm done" message. It's pretty harmless from the DB's
point of view. Do you have a client that crashes a lot?
regards, tom lane
d (see src/backend/utils/Gen_fmgrtab.sh). I suspect problems
generating fmgr.h --- possibly your machine has no awk(1), or a broken
awk?
regards, tom lane
t be superuser, else they will
bypass all the permissions checks anyway.
regards, tom lane
log, I hope, and not using -S or
sending the messages to /dev/null.)
regards, tom lane
you can do something like
SELECT oid FROM foo OFFSET n-1 LIMIT 1;
DELETE FROM foo WHERE oid = value-just-determined;
to zap the tuple without trying to examine its corrupted column.
Repeat if necessary until all tables are clean ...
regards, tom lane
.)
There is a pg_dumplo utility in current sources, a bit late :-(.
I have attached a copy of it that is modified to work with 7.0,
and I think it will work with 6.5, but I don't have a running 6.5
server to check with. Test it out and let us know ...
regards, tom lane
it is, the more I'd tend to think it's
the entry-drop bug.
The bug is fixed (at long last) for 7.1, but in prior versions the only
good workaround is not to vacuum system tables while there are other
active backends.
regards, tom lane
.
regards, tom lane
how a bug can go
unnoticed for years and then two people pop up with it in a week.
You can change "/\name/" to "/name/" in src/backend/catalog/genbki.sh,
then re-make install in that directory; or just use the latest nightly
snapshot.
regards, tom lane
uch notice is not appearing.
But I need indices on this table.
Don't use hash indexes; use the default index type (btree). hash
indexes are subject to deadlock under concurrent insertions. Besides,
they do nothing that you can't do with a btree.
regards, tom lane
. Yes, it's a bug ... it's fixed for 7.1.
regards, tom lane
.
(Actually, I'd suggest using the latest nightly snapshot.)
regards, tom lane
f roundoff errors like that when the wrong compiler
flags are used on certain hardware platforms. What is your hardware,
and how did you build or obtain the Postgres executables? What Postgres
version is this, anyway?
regards, tom lane
disks. My preferred OS, HPUX, can do this easily; I believe
the same thing can be done with some Linux filesystems but don't know
the details. (Any help out there?)
regards, tom lane
following error -
ERROR: Function 'oid8types(oidvector)' does not exist
This second error message looks like you are running a 6.5 psql
against a 7.0 database server. Possibly the pg_dump problem is of
the same making (ie, 6.5 pg_dump) --- check your PATH.
regards, tom lane
Best way out of that is to temporarily set pg_hba.conf to use
non-password authentication ('trust' or maybe 'ident'). Then
you can connect without the password, update the password field
in pg_shadow, and finally restore pg_hba.conf.
regards, tom lane
it then we'd break a lot of other useful cases.
regards, tom lane
. Note however that the COUNT() function currently
uses an int4 counter, so will give bogus results for more than 2G rows.
Maximum number of indexes on a table?
No limit.
regards, tom lane
.
regards, tom lane
something else?
regards, tom lane
alhost, ie, use a local TCP connection instead of a Unix
socket. Set up 127.0.0.1 as ident sameuser and socket connections as
reject in pg_hba.conf.
regards, tom lane
t help you today).
regards, tom lane
=?iso-8859-1?Q?Lo=EFc=20TREGOU=CBT?= [EMAIL PROTECTED] writes:
Tom Lane wrote:
=?iso-8859-1?Q?Lo=EFc=20TREGOU=CBT?= [EMAIL PROTECTED] writes:
raslog=# create index user2000 on log2000(username);
FATAL 1: btree: failed to add item to the page in _bt_sort (2)
How long are the usernames
=localhost at the client.
regards, tom lane
where ctid = '(226,18)';
Note that the TID will change if you update the row!
regards, tom lane
oo big
60/60 nodes(%e), 0/2500 positions(%p), 1/500 (%n), 0 transitions, 0/1000
packed char classes(%k), 0/2000 packed transitions(%a), 0/3000 output
slots(%o)
Don't use lex, but flex. Or read the man page for lex about how to
increase its internal table sizes.
regards, tom lane
comparing performance with -F, I'm not sure exactly how it
stacks up.
regards, tom lane
... and the apparent
loss-of-data problem as even better reason. The number of bugs we
fixed between 6.5.* and 7.0.* is, um, large.
regards, tom lane
the WAL log as
well as the data files, but I'm not convinced of that... trying to
get some profile info ...
regards, tom lane
-B 64 is certainly too small to get reasonable performance.
I got roughly twice the tps reading (pgbench -t 1000, with -F) at -B 1024.
regards, tom lane
change up
to 2. In other words, your kernel may be rounding the delay up to
the next multiple of a clock tick, which might be 10 milliseconds.
That would explain a 50-tps limit real well...
BTW, have you tried pgbench with multiple clients (-c) rather than just
one?
regards, tom lane
, at least, 7.1 is
not slower, even with -F. (Given zero commit_delay, anyway ;-))
regards, tom lane
"lt" [EMAIL PROTECTED] writes:
I have tried add query=4 in pq_options file,
but nothing recorded in syslog.
Are you starting the postmaster with -S ? That suppresses its log...
regards, tom lane
of the smp case to be lower than the one of the up case
Hm, did you set commit_delay to zero? What are the other postmaster
parameters (especially -B) ?
regards, tom lane
the system
will notice that scanning the double-column index is more expensive, and
will avoid using it unless there's good reason.
regards, tom lane
of the creating user, for reasons that
should be apparent after some thought.
regards, tom lane
to report that you have a broken installation. What
would have been helpful is some information about your platform,
compiler, configuration options selected, etc.
regards, tom lane
PS: Hey Marc, why didn't majordomo bounce this? I thought we had
a 40K posting limit...
, tom lane
Compiler: gcc 2.95.2
Configuration Options: none, used defaults
That doesn't sound particularly risky. Odd ...
regards, tom lane
is or what he has to do to fix it. It'd be
better if the problem couldn't occur in the first place. If you really
need a view on pg_shadow, can you postpone creating it until you are run
as superuser?
regards, tom lane
and will eliminate all of these pushups: just store the values
naturally.
If there really is some good reason to use scaled integers, you should
build yourself a scaled-integer datatype.
regards, tom lane
, it might be a good idea to be sure you are saving the file with
Unix-ish newlines, not DOS-ish. Not sure whether our pg_hba reader code
will treat \r as whitespace or not.
regards, tom lane
.
Or possibly /home/lange/pgsql/arch_dep, not sure whether exec_prefix
works in 7.0. But either way you do not want to invoke it from the
source tree. Doing so will confuse it about where to find the postmaster.
regards, tom lane
ioned
last). There's probably a core file left from the crashed backend;
can you get a stack backtrace from it with gdb?
regards, tom lane
with the postmaster log (stdout stderr)? How high
do you have the debug level set?
regards, tom lane
---(end of broadcast)---
TIP 5: Have you checked our extensive FAQ?
http://www.postgresql.org/users-lounge/docs/faq.html
there already is a shm segment lying around that needs to be
removed? See "ipcs" and "ipcrm".
7.1 will remove or work around a conflicting old shm segment, but I'm
not sure that 7.0 will.
regards, tom lane
.
regards, tom lane
---(end of broadcast)---
TIP 6: Have you searched our list archives?
http://www.postgresql.org/search.mpl
7
6701
4321
3043
(10 rows)
regression=# select unique1 from tenk1 limit 3,5;
unique1
-
8009
5057
6701
(3 rows)
regression=#
What PG version are you running?
regards, tom lane
---(end of broadcast)---
rab an RPM from somewhere (if so, where)?
Have you tried running the regression tests on your installation?
regards, tom lane
---(end of broadcast)---
TIP 3: if posting/reading through Usenet, please send an appropriate
subscr
"John W Cunningham" [EMAIL PROTECTED] writes:
How do I run the regression tests?
See
http://www.postgresql.org/users-lounge/docs/7.0/postgres/regress.htm
regards, tom lane
---(end of broadcast)---
TIP 3:
processing, a
good way to avoid the whole problem is to build the software without
locale support.
regards, tom lane
---(end of broadcast)---
TIP 1: subscribe and unsubscribe commands go to [EMAIL PROTECTED]
tables, depending on how well your filesystem copes with lots of files
in a directory.
regards, tom lane
---(end of broadcast)---
TIP 1: subscribe and unsubscribe commands go to [EMAIL PROTECTED]
ut I don't see why you shouldn't
trust it for local.
Kerberos auth might be another answer, but I don't know enough about it
to comment intelligently.
regards, tom lane
---(end of broadcast)---
TIP 4: Don't 'kill -9' the postmaster
on system indexes; it's a tad
arcane.
regards, tom lane
---(end of broadcast)---
TIP 3: if posting/reading through Usenet, please send an appropriate
subscribe-nomail command to [EMAIL PROTECTED] so that your
message can g
Jie Liang [EMAIL PROTECTED] writes:
Dose later version such as 7.0.3 resolved this problem??
Possibly. Try it and see. You should be running 7.0.3 anyway,
just on general principles.
regards, tom lane
---(end of broadcast
.
regards, tom lane
---(end of broadcast)---
TIP 1: subscribe and unsubscribe commands go to [EMAIL PROTECTED]
backupfile.
regards, tom lane
---(end of broadcast)---
TIP 1: subscribe and unsubscribe commands go to [EMAIL PROTECTED]
t 'kill -9' the postmaster
Actually, as of 7.1RC1 that issue is gone away; it's still not
considered good practice to kill -9 any postgres process, but
at least you can't shoot yourself in the foot that way anymore.
regards, tom lane
---(end of
live,
then this is a known bug (date-to-timestamp conversion is off one hour
on such dates). It's fixed for 7.1.
regards, tom lane
---(end of broadcast)---
TIP 3: if posting/reading through Usenet, please send an appropriate
7.0.3, if
not experimenting with 7.1.
regards, tom lane
---(end of broadcast)---
TIP 2: you can get off all lists at once with the unregister command
(send "unregister YourEmailAddressHere" to [EMAIL PROTECTED])
Mathijs Brands [EMAIL PROTECTED] writes:
PS_USE_CLOBBER_ARGV -is- used on Solaris. I just checked.
Hm. But 7.1 postgres backends fail to change their ps display?
Does sendmail change its ps display on your machine?
regards, tom lane
---(end
postmaster. Use ipcs and
ipcrm to check and clean up.
7.1 uses a slightly different approach that hopefully avoids this
particular failure scenario...
regards, tom lane
---(end of broadcast)---
TIP 5: Have you checked our
mage's etc. Is there a script or such which does do the job?
7.1 pg_dump can handle large objects, but in prior versions you have to
use some additional code. Attached is some code that will work in 7.0.
(I think it works in 6.5 too, but have not tested.)
regards, tom lane
pg
forget 7.0.* and go straight
to 7.1.
regards, tom lane
---(end of broadcast)---
TIP 5: Have you checked our extensive FAQ?
http://www.postgresql.org/users-lounge/docs/faq.html
re issuing?
Also, I notice that when I try to use config as the table name the
$PGDATA/config file does NOT get created,
It'd be $PGDATA/base/config/config
regards, tom lane
---(end of broadcast)---
TIP 5: Have you checked our
version are you using?
regards, tom lane
---(end of broadcast)---
TIP 1: subscribe and unsubscribe commands go to [EMAIL PROTECTED]
up. If there's a 384M shared-memory block
in the system, how can there be 197M of kernel disk buffers (not to
mention the kernel and user programs)?
regards, tom lane
---(end of broadcast)---
TIP 1: subscribe and unsubscribe
and see how things change...
regards, tom lane
---(end of broadcast)---
TIP 3: if posting/reading through Usenet, please send an appropriate
subscribe-nomail command to [EMAIL PROTECTED] so that your
message can get through
it had a temp table
created, then you'd see the leftover tables. Offhand I think you won't
be allowed to delete 'em unless you restart the postmaster with the
allow-system-table-mods switch.
I'm using pgsql 6.5.3 on Linux 2.2
Time to upgrade...
regards, tom lane
bangh [EMAIL PROTECTED] writes:
I am using 6.5 too. The reason is some data type are
gone. e.g. datetime. If do so, I have to modify my application.
?? This still works:
regression=# create table foo (f1 datetime);
CREATE
regards, tom lane
temp tables.
BTW, you should consider updating to a newer release in hopes of getting
rid of the bug that's allowing the temp tables to stick around...
regards, tom lane
---(end of broadcast)---
TIP 2: you can get off all
versions do provide the relation name instead of number
in that error message, btw.
regards, tom lane
---(end of broadcast)---
TIP 5: Have you checked our extensive FAQ?
http://www.postgresql.org/users-lounge/docs/faq.html
like
the postmaster does not see that environment variable.
regards, tom lane
---(end of broadcast)---
TIP 4: Don't 'kill -9' the postmaster
?
regards, tom lane
---(end of broadcast)---
TIP 5: Have you checked our extensive FAQ?
http://www.postgresql.org/users-lounge/docs/faq.html
. You can certainly raise that limit, but I don't
know just how that's done on FreeBSD.
regards, tom lane
---(end of broadcast)---
TIP 4: Don't 'kill -9' the postmaster
Reiner Dassing [EMAIL PROTECTED] writes:
Tom Lane wrote:
That shouldn't happen. Could we see the dump file? (Just the schema
commands, not the data itself...)
Thank you for your help. Certainly I will give this short schema:
Hm. I replicated your data out to a million lines
There is no way to disable WAL. However, you might try running with
fsync turned off if you don't care about data integrity.
regards, tom lane
---(end of broadcast)---
TIP 2: you can get off all lists at once
1 - 100 of 3437 matches
Mail list logo
