/lists.mysql.com/mysql
> >> To unsubscribe:http://lists.mysql.com/[EMAIL PROTECTED]
> >>
> >>
> >
> >
> >--
> >MySQL General Mailing List
> >For list archives: http://lists.mysql.com/mysql
> >To unsubscribe:http://lists.mysql.com/[EMAIL P
Confused about how ALTER TABLE foo ADD col2... operates.
On an otherwise idle server, iostat 1 shows the disk being written to at a
rate of 15MB/sec (its capacity is about 50MB/sec), but the ibdata-autoextend
file only grows at a rate of 500kb/sec.
12GB table (w/ 1.5GB index) took 6 h
> What is the definition of strmov in C? It seems to be a
> MYSQL library sprecific function in m_string.h and is used
> in some of the documentations examples.
In MySQL 3.23.x, it's defined in strings/strmov.c
#if !defined(MC68000) && !defined(DS90)
char *strmov(register char *dst, register c
> >>>Does Red Hat have some kind of userland address space hack that
> >>>we're not aware of?
> >>
> >>Do you have any special kernel config options that you did not use before?
> > the thread stacks are 2MB apiece (bf601000-bf80 is
> > 2093056 bytes, or 2044kB)! Yet:
> >
> > # mysql -e 'sho
0:00 0
bf601000-bf80 rwxp 1000 00:00 0
bfffa000-c000 rwxp b000 00:00 0
the thread stacks are 2MB apiece (bf601000-bf80 is
2093056 bytes, or 2044kB)! Yet:
# mysql -e 'show variables' | grep thread_stack
thread_stack196608
It seems like the setting does nothing for
Debian and the MySQL AB 3.23.58
binary around 245 threads, linux 2.4.23 custom kernel. (3GB of RAM,
not that it matters)
Are we missing a setting?
Does Red Hat have some kind of userland address space hack that
we're not aware of?
--
Michael Bacarella24/7 phone: 1-6
home_dir = /dblog
> > innodb_log_arch_dir = /dblog
> > innodb_log_archive=0
> >
> > set-variable = innodb_log_files_in_group=3
> > set-variable = innodb_log_file_size=1.5G
> > set-variable = innodb_log_buffer_size=8M
> >
> > innodb_flush_log_
max_connections=415
> set-variable = table_cache=256
> set-variable = sort_buffer=4M
> set-variable = key_buffer=128M
> set-variable = thread_cache_size=415
> set-variable = max_connect_errors=5
> set-variable = thread_stack=256k
--
Michael Bacarella24/7 phone: 1-646-641-8662
Netgraft Corporation http://netgraft.com/
--
MySQL General Mailing List
For list archives: http://lists.mysql.com/mysql
To unsubscribe:http://lists.mysql.com/[EMAIL PROTECTED]
et-variable = wait_timeout=360
set-variable = tmp_table_size=800
set-variable = max_connections=415
set-variable = table_cache=256
set-variable = sort_buffer=4M
set-variable = key_buffer=128M
set-variable = thread_cache_size=415
set-variable = max_connect_errors=5
set-variable
First we cut to the chase with a resolved stack trace from
the most recent crash:
0x80c23d5 handle_segfault__Fi + 425
0x40022f54 _end + 935506260
0x822cdef btr_search_build_page_hash_index + 4771
0x82281c3 btr_search_info_update_slow + 919
0x8213f9e btr_cur_search_to_nth_level + 3154
0x81e9dce row
accident, maybe it was a sign of technical
brilliance--ignoring marketing demands to achieve technical superiority.
Regardless, it's to a great effect.
I suspect all of the clever and hard work in the world won't get
MS SQL Server to do the unconventional.
Also, insert obligatory benefits of
64(320, F_SETFL, O_RDWR|O_NONBLOCK) = 0
setsockopt(320, SOL_IP, IP_TOS, [8], 4) = 0
setsockopt(320, SOL_TCP, TCP_NODELAY, [1], 4) = 0
The 10.0.0.x connects are from our web servers. Wish I could've seen
what killed off all of the threads.
Nothing in the logs.
So, anyone have any idea what
in log 'dbms2-bin.319' at
position 68555844
Is the error disconnect/reconnect not sync-safe?
[1] The watchdog script checks to make sure the that a frequently
updated table has a row with timestamp younger than 5 minutes on
the slave.
--
Michael Bacarella24/7
Names of vendors who are happy to provide servers
applicable for high load Linux/MySQL. Willing to
do custom configurations.
Anyone?
--
Michael Bacarella24/7 phone: 1-646-641-8662
Netgraft Corporation http://netgraft.com/
Finger email address for public key
> does anyone know what it means when it says state = NULL doing the "show
> processlist"?
The client is connected but not doing anything. It could be
either preparing to send a query, between queries, or finished
with its last query and hasn't disconnected yet.
-M
--
MySQL General Mailing Li
,
(struct hostent*) 0, (char*) 0, 0, (struct hostent **) 0, &skr);
;
return 0;
}
Solution? Tell the user there is no C++ compiler if there is no C++ compiler.
--
Michael Bacarella 24/7 phone: 646 641-8662
Netgraft Corporation http://netgraft.com/
&q
Are you sure it would help? MySQL is damn fast at taking connections.
When we converted to mod_perl I made a mental note to switch
to persistent connections. It turns out that it still
ran fast enough even though every hit results in an RDBMS
connection. And at this point we're doing 5M of them p
nsactions or running in auto-commit mode?
Most of the time we're in auto-commit mode. It's uncommon that we
do explicit BEGIN/COMMIT. (That was your question right?)
Anyway, checking out the sar logs, it looks like MySQL has to resort to
reading disk more to serve its queries. Ti
= table_cache=256
set-variable = sort_buffer=4M
set-variable = key_buffer=128M
set-variable = thread_cache_size=350
set-variable = max_connect_errors=5
Could anyone suggest anything?
--
Michael Bacarella | Netgraft Corp
| 545 Eighth Ave #401
Systems Analysis | New Yo
?
How can I make them go away? :)
--
Michael Bacarella | Netgraft Corp
| 545 Eighth Ave #401
Systems Analysis | New York, NY 10018
Technical Support | 212 946-1038 | 917 670-6982
Managed Services | http://netgraft.com
ur vendors do this too, but its a good idea to
do it after UPS hands you the box too, for reasons that
should be obvious.
--
Michael Bacarella | Netgraft Corp
| 545 Eighth Ave #401
Systems Analysis | New York, NY 10018
Technical Support | 212 946-1038 | 917 670-69
routine stuff like table optimizations, adding keys, etc. If a database is
> reliable, things like that shouldn't happen. Comments like those in the
> MySQL manual scared me.
We've never lost data. Our database server has crashed hard from OS failures
and suffered plenty of
e.
> http://zdnet.com.com/2100-1104-977958.html
So why no mention on the MySQL.COM site? That rather bugs me. In contrast,
sites for products like Apache or Bind are very clear about current/past
security issues.
Is MySQL.COM the wrong place?
--
Michael Bacarella | Netgraft Corp
ng
the imagespace, is the cost. If each node were $5000 USD, and 20
nodes solved the problem, you're looking at $100,000 USD in hardware
costs. Isn't that like what a typical Oracle support contract
costs in a year?
--
Michael Bacarella | Netgraft Corp
| 545 Eighth Ave
letes.
Creating a thread often involves an expensive system call.
The thread cache instructs MySQL to keep N threads around,
even if they're not actively running queries, thereby eliminating
thread creation overhead.
Setting this value higher had a noticeable effect for me.
Results may vary.
--
trying to
> authenticate and all client machines report a cannot connect error message.
>
> We went to 3.23.51 instead a little while ago and it seems to be working
> normally. Any ideas about what might be causing this? Is this a known
> issue?
--
Michael Bacarella | Netgraf
is cooperative, that is, it probably checks during each phase of processing
whether or not it has been killed, and if it has, to safely clean up.
Forcibly terminating a thread where it has no choice will not
allow cleanup routines to run, which could cause undefined behavior
or at the very least r
#x27;t something to undertake lightly.
Make sure some egomaniac isn't pushing for it just because
of prejudices. Migrating a system such as this simply because
of lack of subselects sounds insane (how have you coped
up until now?)
Are there other factors?
--
Michael Bacarella | Netgraft C
t COUNT(*) from Log WHERE LogIP="10.0.0.22" ORDER BY LogID
> LIMIT 100;
--
Michael Bacarella | Netgraft Corp
| 545 Eighth Ave #401
Systems Analysis | New York, NY 10018
Technical Support | 212 946-1038 | 917 670-6
Is there anything in the error log?
See mysql/data/hostname.err
-M
On Tue, Jul 09, 2002 at 03:26:17PM -0500, Steve Mertz wrote:
> I keep getting "Lost connection to MySQL server during query"
>
> marcos:mysql-3.23.51 {117} mysqladmin version
> mysqladmin: connect to server at 'localhost' faile
in fact, these are common UNIX semantics.
If a child process terminates (on Linux these are threads), they are
marked defunct/zombie until the parent process gets around to "reaping"
them. Which is why they're absent from a second process listing.
See fork(2) and wait(2), waitpid
o avail. I reverted to 3.23.50.
> >
> > At first I thought it was my fault so I decided not to report it
> > until I did some further research.
>
> What OS are you using? And did you use the pre-built binaries or
> build from source?
Stock Red Hat 7.2. Pre-built mysql-max bi
results ?
> >
> > cd sql-bench
> > perl run-all-tests --log
> >
> > The file I am interested in is the summary file named 'output/RUN-*'
> >
> > Regards,
> > Monty
>
>
>
> --
and is busy almost 24 hours a day, so there's no real slow
period where the maintenance scripts run with less contention issues.
I can think of plenty of better workarounds, but I wanted to solicit input before
implementing something that I may not have to.
Suggestions? ;)
--
Michael Bacarella
oves the firewall rule. The select will block until innodb
startup is complete, and in the meantime clients will keep trying
other servers.
Ick, I feel unclean just writing that. /me washes his hands.
--
Michael Bacarella | Netgraft Corporation
| 545 Eighth Ave #401
Syst
yer instead of all of your scripts.
If you want to pitch it to management, just say "XMLRPC!" until their
ears bleed.
--
Michael Bacarella | Netgraft Corporation
| 545 Eighth Ave #401
Systems Analysis | New York, NY 10018
Technical Support | 212 946-1038 | 917
rect in assuming that each COMMIT generates disk I/O and will not
return until it's written to the log? Whereas setting:
innodb_flush_log_at_trx_commit=0
will cause COMMIT to return immediately, and hit the disk much less
frequently with the understanding that in a crash we would be
o
-Wwrite-strings -Woverloaded-virtual -Wextern-inline -Wsign-promo -Wreorder
-Wctor-dtor-privacy -Wnon-virtual-dtor -felide-constructors -fno-exceptions -fno-rtti
-mcpu=pentiumpro -O3 -fno-omit-frame-pointer' CXX=gcc
--
Michael Bacarella | Netgraft Corporation
| 545 Eighth Ave #40
in your
WHERE clause? For example:
UPDATE foo SET flag = 0 WHERE foo_id = foo_id;
foo_id = foo_id of course matches all.
Not much room for that kind of error in this example, but
perhaps in other scripts that deal with the table in question?
--
Michael Bacarella | Netgraft Corporation
ql.com/ (the list archive)
>
> To request this thread, e-mail <[EMAIL PROTECTED]>
> To unsubscribe, e-mail <[EMAIL PROTECTED]>
> Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
>
--
Michael Ba
less useful. Is there a way to switch to a particular language
only for this function or only for a specific session?
Just want to make sure I'm not duplicating any effort.
sql,table,query
--
Michael Bacarella | Netgraft Corporation
| 545 Eighth Ave #401
Systems Analy
(either put the hostname in /etc/hosts or use an IP
address instead of a name) if your mysql server is not localhost..
--
Michael Bacarella | Netgraft Corporation
| 545 Eighth Ave #401
Systems Analysis | New York, NY 10018
Technical Sup
ual)
>http://lists.mysql.com/ (the list archive)
>
> To request this thread, e-mail <[EMAIL PROTECTED]>
> To unsubscribe, e-mail <[EMAIL PROTECTED]>
> Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
>
--
Michael Bacarella | Netgraft C
.
Seems like I wasn't the only one who needed it. The response to
this article has been overall very positive.
More MySQL success stories would be much appreciated.
--
Michael Bacarella | Netgraft Corporation
| 545 Eighth Ave #401
Systems Analysis | New York, NY
3.23.50 is referenced in the documentation but not available
for download. Should it be there?
Don't mean to nag, but I'm itching for that autoextend feature.
PS. mysql,query,table
--
Michael Bacarella | Netgraft Corporation
| 545 Eighth Ave #401
Systems Analy
as where is that table being
> written to disk (I could check the amt of free disk space).
Try adjusting tmp_table_size.
This is a FAQ.
--
Michael Bacarella | Netgraft Corporation
| 545 Eighth Ave #401
Systems Analysis | New York, NY 10018
Technical Support | 212 9
we could probably
get away with it without too much of a hassle. We're years away
from this, if ever, though.
We keep expecting to hit a limit any day now, but MySQL has continued to
perform beyond our expectations for years. We haven't even switched
to InnoDB tables yet.
--
count(*) |
> +--+
> | 35395 |
> +--+
> 1 row in set (0.00 sec)
>
> Any ideas what might be causing this?
>
> Here's the version:
> [root@s2 tauren]# mysql -V
> mysql Ver 11.15 Distri
does nothing because I am unable to load mysqld
> >again so that I can restart it using the --skip-grant-tables option.
> >
> >Can someone help me understand which of the pid's is the actual one, and
> >how I can kill it?
--
Michael Bacarella | 545 Eighth Ave #401
does nothing because I am unable to load mysqld
> >again so that I can restart it using the --skip-grant-tables option.
> >
> >Can someone help me understand which of the pid's is the actual one, and
> >how I can kill it?
--
Michael Bacarella | 545 Eighth Ave #401
oi8_ru --prefix=/usr/local i386--freebsd4.4
> Perl: This is perl, version 5.005_03 built for i386-freebsd
>
> -
> Before posting, please check:
> http://www.mysql.com/manual.php (the manual)
>http:
re user friendly OS.
I was amazed at how hard it was for me to install, considering that
I can put together Linux systems completely from scratch.
YMMV.
---
Michael Bacarella <[EMAIL PROTECTED]>
Technical Staff / System Development,
New York Connect.Net, Ltd.
--
dn't have experience with Java, for example).
You should not only evaluate their current worth but any potential
future worth. If they're enthusiastic, a lack of industry experience
may become a non-issue after they've been working there for 3 months.
But then again, based on the po
.23 introduced a turbo switch. It's
> just off by default").
>
--
Michael Bacarella <[EMAIL PROTECTED]>
Technical Staff / System Development,
New York Connect.Net, Ltd.
-
Before posting, please che
It's
just off by default").
Thanks.
--
Michael Bacarella <[EMAIL PROTECTED]>
Technical Staff / System Development,
New York Connect.Net, Ltd.
-
Before posting, please check:
http://www.mysql.com/manual.
michael.bacarella.com/research/mysqlmyths.html>
is a document I've been maintaining (with the help of numerous others)
whose purpose is to alleviate such misunderstandings of MySQL.
I doubt all of the cluephone in the world wouldn't convince the
person responsible for that quote, tho
nothing to do with the domain, but they
would suffer on behalf of someone else's actions.
--
Michael Bacarella <[EMAIL PROTECTED]>
Technical Staff / System Development,
New York Connect.Net, Ltd.
-
Before posting
; start, I'd say use non-free software. That can be UNIX still, but not
> free-unix. Sybase, DB2 or Oracle for a database, and JSP seem to be the
> only "enterprise" solutions around for UNIX. But judging by your
> question you don't seem to be stuck with UNIX
testimonials.
Some questions rewritten.
Copyright information and distribution terms. Yay.
Hopefully, I've caught all of the spelling mistakes.
If anyone has any FUD they'd like to see added, let me know.
--
Michael Bacarella <[EMAIL PROTECTED]>
Technical Staff / System Devel
tial Integrity
> | > MySQL does not support RI, anyone has good suggestion that do this in
> the
> | > program ?
> | > As this would be a nightmare if I have 50 detail tables to update
> | > programmatically.
--
Michael Bacarella <[EMAIL PROTECTED]>
Technical Staff / Syste
enerally, it's hard to place one as
being better than the other when the choices are FreeBSD or Linux.
Linux vs say, Solaris on the other hand is much different. Especially where
MySQL or Apache are concerned. [Solaris being the bad choice on that hardware]
--
Michael Bacarella <[EM
gt;
> >
>
>
> -
> Before posting, please check:
>http://www.mysql.com/manual.php (the manual)
>http://lists.mysql.com/ (the list archive)
>
> To request this thread, e-mail <[EMA
licate key error.
A bail out scheme after say, 10 tries wouldn't be a bad idea.
--
Michael Bacarella <[EMAIL PROTECTED]>
Technical Staff / System Development,
New York Connect.Net, Ltd.
-
Before posting, please check:
te all feedback.
--
Michael Bacarella <[EMAIL PROTECTED]>
Technical Staff / System Development,
New York Connect.Net, Ltd.
-
Before posting, please check:
http://www.mysql.com/manual.php (the manual)
http://lists.mysql.com
http://netgraft.com/~mbac/research/mysqlmyths.html
--
Michael Bacarella <[EMAIL PROTECTED]>
Technical Staff / System Development,
New York Connect.Net, Ltd.
-
Before posting, please check:
http://www.mysql.com/manual.
ase system.
[ PS. I don't own the site, I just implemented it ]
--
Michael Bacarella <[EMAIL PROTECTED]>
Technical Staff / System Development,
New York Connect.Net, Ltd.
-
Before posting, please check:
http://www
a perl expert so I cannot speak to that. However, I doubt that
> it's effortless or trivial.
Why would it be easier to write said layer if it was inside the DBMS
instead of outside?
--
Michael Bacarella <[EMAIL
ySQL itself have to do this?
My naive understanding is that it would be effortless to write a module
in perl that translates the data returned by the DBI to XML.
--
Michael Bacarella <[EMAIL PROTECTED]>
Technical Staff / System Devel
RL where I can read more about doing this?
We use a seperate mysql database for one of our sites and we
_DON'T_ use persistant connections. The overhead of connecting
once at the start of each script never seemed that significant
to us.
YMMV.
--
Michael Bacarella <[EMAIL PROTEC
69 matches
Mail list logo