MyISAM RAID Tables in the MySQL-MAX Binaries?
Are RAID tables supposed to be enabled in the -MAX binaries? They appear not to be (in 4.0.1), and that surprises me a bit. It thought that the point of the -MAX binaries was to enable all the bells and whistles by default. Jeremy -- Jeremy D. Zawodny, [EMAIL PROTECTED] Technical Yahoo - Yahoo Finance Desk: (408) 349-7878 Fax: (408) 349-5454 Cell: (408) 685-5936 MySQL 3.23.47-max: up 18 days, processed 582,964,719 queries (361/sec. avg) - 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 [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Re: MyISAM RAID Tables in the MySQL-MAX Binaries?
Are RAID tables supposed to be enabled in the -MAX binaries? They appear not to be (in 4.0.1), and that surprises me a bit. I thought RAID tables were retired in MySQL 4. It was only a split for the data files anyhow, not the index files, so it was incomplete at best for getting around the lack of large file support in some (now old) OSes. For speed, RAID is best handled by the OS or hardware. Sincerely, Steven Roussey http://Network54.com/?pp=e - 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 [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Re: MyISAM RAID Tables in the MySQL-MAX Binaries?
On Mon, Feb 25, 2002 at 11:21:26PM -0800, Steven Roussey wrote: Are RAID tables supposed to be enabled in the -MAX binaries? They appear not to be (in 4.0.1), and that surprises me a bit. I thought RAID tables were retired in MySQL 4. Hmm. The have_raid variable is still there in SHOW VARIABLES. It was only a split for the data files anyhow, not the index files, so it was incomplete at best for getting around the lack of large file support in some (now old) OSes. I agree. For speed, RAID is best handled by the OS or hardware. And that's basically what I just finished writing. :-) But I also had to write that in order you get RAID, you *must* recompile MySQL. I'm not sure if that's by design or a simple omission. Jeremy -- Jeremy D. Zawodny, [EMAIL PROTECTED] Technical Yahoo - Yahoo Finance Desk: (408) 349-7878 Fax: (408) 349-5454 Cell: (408) 685-5936 MySQL 3.23.47-max: up 18 days, processed 584,411,183 queries (361/sec. avg) - 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 [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Backing up raid tables
Well, I successfully solved our 2 gig file limit problem with RAID tables without too much hassle, despite a dearth of documentation. However, it has introduced another significant problem. It would seem mysqldump isn't smart enough to look for raided tables. Taking the server down to do nightly backups is a very unfavorable solutions with orders coming in globally at all times. Is there a) a way to get mysqldump to work with raided tables or b) another way to do tape backups while retaining file integrity and not taking mysqld down? Thanks, Age --- Adriano Age Manocchia [EMAIL PROTECTED] Cornell University AIM/Yahoo: BigRedWhat http://www.hockey.cornell.edu/age ICQ: 5962736 --- The capacity to blunder slightly is the real marvel of DNA. Without this special attribute, we would still be anaerobic bacteria, and there would be no music. -Lewis Thomas - 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 [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Re: raid tables
On Tue, Jun 19, 2001 at 04:49:58AM +0200, Tonu Samuel wrote: On Mon, 18 Jun 2001, Jeremy Zawodny wrote: I'd use a combination of my.cnf entries and SQL. The my.cnf entry (or entries) woudl list the available directories, like others have suggested. Then, using SQL, you could: * Mark a table to be spread out at creation time vis some attribute. * Possibly do it after the fact using ALTER TABLE. Maybe you are right. I am thinking about security and this is a not the best idea if we introduce a way to write data anywhere on the disks on user request. Indeed. Meanwhile any change in cnf file requires restart of MySQL which is not always possible :(. A while back (at last year's database summit), Monty made some rumblings about maybe making it possible to modify many (most?) of the server variables without an actual restart. Is that that on the horizon for MySQL 4.x, or did it become more difficult that originally thought? If it does happen, I'll start working on the MySQL self-tuning add-on so that folks don't have to play with their my.cnf files as much... Really. I think it'd be fun to build and VERY useful. Then MSSQL wouldn't be the only [advertised] self-tuning database server in the market. :-) Jeremy -- Jeremy D. Zawodny, [EMAIL PROTECTED] Technical Yahoo - Yahoo Finance Desk: (408) 349-7878Fax: (408) 349-5454Cell: (408) 439-9951 MySQL 3.23.29: up 2 days, processed 20,171,403 queries (79/sec. avg) - 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 [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Re: Important bug in RAID tables
On Tue, 19 Jun 2001, Patrizio Bruno wrote: BTW, why did you change the chunksize? By default it is 256000 and I tested it for speed with many different values. 256000 seemed to be best value. If you have any experience showing that I was wrong, please share it with me. I was testing different chunksize looking for the best on SGI's XFS, I will let you know some better value than 256000, if one exists :) I got this result when I first put lower vales like 4kB and 8kB and found that RAID takes too much of CPU time. When using RAID, instead of single syscalls open/read/write/seek/tell/close mysql calls then many times. Syscalls are very expensive in time in many operating systems and I got no slowdown when I increased raid chunk value to be so high as 256000kBytes. I think this is the good value as such amount of data can be fitted into hard disk hardware cache and can speed up writing do disk many times. Because this I do not suggest to extend it also. Default value should be good for most cases. -- For technical support contracts, goto https://order.mysql.com/ __ ___ ___ __ / |/ /_ __/ __/ __ \/ /Mr. Tonu Samuel [EMAIL PROTECTED] / /|_/ / // /\ \/ /_/ / /__ MySQL AB, Security Administrator /_/ /_/\_, /___/\___\_\___/ Hong Kong, China ___/ www.mysql.com - 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 [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Re: Important bug in RAID tables
I can't reproduce the error using a different database configuration and I can't use the same database schema to make my test (the machine is now in production state), so I attached to this mail te sql file used to create the tables, this database is used by ASPseek 1.2.0, you can download it at www.aspseek.org, and the problem occurred indexing http://lanazione.monrif.net (almost 3 millions documents, so there were almost 3 millions records in the table 'urlword') - Patrizio Bruno DADA spa / Special Projects Viale Giovine Italia 50122 Firenze Italy tel +39 055200211 fax +39 0552478143 PGP PublicKey available at: http://www.keyserver.net/en/ - # This SQL script will create the necessary tables for ASPSeek. # It should be used for MySQL only. # To create the tables, use something like # mysql -uasp -pmypassword aspseek tables.sql # #drop table wordurl; #drop table wordurl1; #drop table urlword; #drop table urlwords00; #drop table urlwords01; #drop table urlwords02; #drop table urlwords03; #drop table urlwords04; #drop table urlwords05; #drop table urlwords06; #drop table urlwords07; #drop table urlwords08; #drop table urlwords09; #drop table urlwords10; #drop table urlwords11; #drop table urlwords12; #drop table urlwords13; #drop table urlwords14; #drop table urlwords15; #drop table sites; #drop table stat; #drop table robots; #drop table subsets; #drop table spaces; #drop table tmpurl; #drop table wordsite; #drop table citation; #drop table countries; #drop table cache; create table wordurl(word tinyblob not null, word_id integer auto_increment primary key, urls blob, urlcount integer, totalcount integer, unique index(word(64))) PACK_KEYS=1 DELAY_KEY_WRITE=1; create table wordurl1(word tinyblob not null, word_id integer not null, urls blob, urlcount integer, totalcount integer, unique index(word(64)), unique index(word_id)); create table urlword(url_id integer auto_increment primary key, site_id integer not null, deleted tinyint DEFAULT 0 NOT NULL, url varchar(128) not null, next_index_time INT NOT NULL, status int(11) DEFAULT '0' NOT NULL, crc char(32) DEFAULT '' NOT NULL, last_modified varchar(32) DEFAULT '' NOT NULL, etag varchar(48) DEFAULT '' NOT NULL, last_index_time INT NOT NULL, referrer int(11) DEFAULT '0' NOT NULL, tag int(11) DEFAULT '0' NOT NULL, hops int(11) DEFAULT '0' NOT NULL, redir integer, origin integer, unique index(url), index(next_index_time), index(hops,next_index_time), index crc (site_id,crc(8))) DELAY_KEY_WRITE=1 RAID_TYPE=STRIPED RAID_CHUNKS=8 RAID_CHUNKSIZE=153600 PACK_KEYS=1; create table urlwords00(url_id integer NOT NULL, deleted tinyint DEFAULT 0 NOT NULL, wordcount integer, totalcount integer, content_type varchar(48) DEFAULT '' NOT NULL, charset varchar(32) DEFAULT '' NOT NULL, title varbinary(128) DEFAULT '' NOT NULL, txt varbinary(255) DEFAULT '' NOT NULL, docsize int(11) DEFAULT '0' NOT NULL, keywords varchar(255) DEFAULT '' NOT NULL, description varbinary(100) DEFAULT '' NOT NULL, lang varchar(2) DEFAULT '' NOT NULL, words longblob, hrefs longblob, unique index(url_id)) DELAY_KEY_WRITE=1 RAID_TYPE=STRIPED RAID_CHUNKS=10 RAID_CHUNKSIZE=153600; create table urlwords01(url_id integer NOT NULL, deleted tinyint DEFAULT 0 NOT NULL, wordcount integer, totalcount integer, content_type varchar(48) DEFAULT '' NOT NULL, charset varchar(32) DEFAULT '' NOT NULL, title varbinary(128) DEFAULT '' NOT NULL, txt varbinary(255) DEFAULT '' NOT NULL, docsize int(11) DEFAULT '0' NOT NULL, keywords varchar(255) DEFAULT '' NOT NULL, description varbinary(100) DEFAULT '' NOT NULL, lang varchar(2) DEFAULT '' NOT NULL, words longblob, hrefs longblob, unique index(url_id)) DELAY_KEY_WRITE=1 RAID_TYPE=STRIPED RAID_CHUNKS=10 RAID_CHUNKSIZE=153600; create table urlwords02(url_id integer NOT NULL, deleted tinyint DEFAULT 0 NOT NULL, wordcount integer, totalcount integer, content_type varchar(48) DEFAULT '' NOT NULL, charset varchar(32) DEFAULT '' NOT NULL, title varbinary(128) DEFAULT '' NOT NULL, txt varbinary(255) DEFAULT '' NOT NULL, docsize int(11) DEFAULT '0' NOT NULL, keywords varchar(255) DEFAULT '' NOT NULL, description varbinary(100) DEFAULT '' NOT NULL,
Re: Important bug in RAID tables
I can't reproduce the error using a different database configuration and I can't use the same database schema to make my test (the machine is now in production state), so I attached to this mail te sql file used to create the tables, this database is used by ASPseek 1.2.0, you can download it at www.aspseek.org, and the problem occurred indexing http://lanazione.monrif.net (almost 3 millions documents, so there were almost 3 millions records in the table 'urlword') - Patrizio Bruno DADA spa / Special Projects Viale Giovine Italia 50122 Firenze Italy tel +39 055200211 fax +39 0552478143 PGP PublicKey available at: http://www.keyserver.net/en/ - # This SQL script will create the necessary tables for ASPSeek. # It should be used for MySQL only. # To create the tables, use something like # mysql -uasp -pmypassword aspseek tables.sql # #drop table wordurl; #drop table wordurl1; #drop table urlword; #drop table urlwords00; #drop table urlwords01; #drop table urlwords02; #drop table urlwords03; #drop table urlwords04; #drop table urlwords05; #drop table urlwords06; #drop table urlwords07; #drop table urlwords08; #drop table urlwords09; #drop table urlwords10; #drop table urlwords11; #drop table urlwords12; #drop table urlwords13; #drop table urlwords14; #drop table urlwords15; #drop table sites; #drop table stat; #drop table robots; #drop table subsets; #drop table spaces; #drop table tmpurl; #drop table wordsite; #drop table citation; #drop table countries; #drop table cache; create table wordurl(word tinyblob not null, word_id integer auto_increment primary key, urls blob, urlcount integer, totalcount integer, unique index(word(64))) PACK_KEYS=1 DELAY_KEY_WRITE=1; create table wordurl1(word tinyblob not null, word_id integer not null, urls blob, urlcount integer, totalcount integer, unique index(word(64)), unique index(word_id)); create table urlword(url_id integer auto_increment primary key, site_id integer not null, deleted tinyint DEFAULT 0 NOT NULL, url varchar(128) not null, next_index_time INT NOT NULL, status int(11) DEFAULT '0' NOT NULL, crc char(32) DEFAULT '' NOT NULL, last_modified varchar(32) DEFAULT '' NOT NULL, etag varchar(48) DEFAULT '' NOT NULL, last_index_time INT NOT NULL, referrer int(11) DEFAULT '0' NOT NULL, tag int(11) DEFAULT '0' NOT NULL, hops int(11) DEFAULT '0' NOT NULL, redir integer, origin integer, unique index(url), index(next_index_time), index(hops,next_index_time), index crc (site_id,crc(8))) DELAY_KEY_WRITE=1 RAID_TYPE=STRIPED RAID_CHUNKS=8 RAID_CHUNKSIZE=153600 PACK_KEYS=1; create table urlwords00(url_id integer NOT NULL, deleted tinyint DEFAULT 0 NOT NULL, wordcount integer, totalcount integer, content_type varchar(48) DEFAULT '' NOT NULL, charset varchar(32) DEFAULT '' NOT NULL, title varbinary(128) DEFAULT '' NOT NULL, txt varbinary(255) DEFAULT '' NOT NULL, docsize int(11) DEFAULT '0' NOT NULL, keywords varchar(255) DEFAULT '' NOT NULL, description varbinary(100) DEFAULT '' NOT NULL, lang varchar(2) DEFAULT '' NOT NULL, words longblob, hrefs longblob, unique index(url_id)) DELAY_KEY_WRITE=1 RAID_TYPE=STRIPED RAID_CHUNKS=10 RAID_CHUNKSIZE=153600; create table urlwords01(url_id integer NOT NULL, deleted tinyint DEFAULT 0 NOT NULL, wordcount integer, totalcount integer, content_type varchar(48) DEFAULT '' NOT NULL, charset varchar(32) DEFAULT '' NOT NULL, title varbinary(128) DEFAULT '' NOT NULL, txt varbinary(255) DEFAULT '' NOT NULL, docsize int(11) DEFAULT '0' NOT NULL, keywords varchar(255) DEFAULT '' NOT NULL, description varbinary(100) DEFAULT '' NOT NULL, lang varchar(2) DEFAULT '' NOT NULL, words longblob, hrefs longblob, unique index(url_id)) DELAY_KEY_WRITE=1 RAID_TYPE=STRIPED RAID_CHUNKS=10 RAID_CHUNKSIZE=153600; create table urlwords02(url_id integer NOT NULL, deleted tinyint DEFAULT 0 NOT NULL, wordcount integer, totalcount integer, content_type varchar(48) DEFAULT '' NOT NULL, charset varchar(32) DEFAULT '' NOT NULL, title varbinary(128) DEFAULT '' NOT NULL, txt varbinary(255) DEFAULT '' NOT NULL, docsize int(11) DEFAULT '0' NOT NULL, keywords varchar(255) DEFAULT '' NOT NULL, description varbinary(100) DEFAULT '' NOT NULL,
raid tables
Is there a way to tell MySQL to automatically spread its tables over a number of disks without going through the trouble of symbolic linking the chunks to the different disks? Thanks, Jamie Krasnoo [EMAIL PROTECTED] - 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 [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Re: raid tables
Isn't this something that RAID O does for you on a larger scale anyway? I would expect that most would have some sort of raid on their mutliple drive system. On Monday 18 June 2001 01:52 am, you wrote: On Sun, Jun 17, 2001 at 11:48:04PM -0700, Jamie Krasnoo wrote: Is there a way to tell MySQL to automatically spread its tables over a number of disks without going through the trouble of symbolic linking the chunks to the different disks? There isn't yet, no. It'd certainly be a nice addition, though, wouldn't it? Jeremy - 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 [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Re: raid tables
Hi Samuel I would do this in sql. I thinks this would be much more convenient to administrate. It would be great if you would see these informations also in with the command eg: show database storage and you can see as return, where the files are located. Thanks Philip On Mon, 18 Jun 2001 12:13:50 +0200 (CEST) Tonu Samuel [EMAIL PROTECTED] wrote: On Mon, 18 Jun 2001, Jamie Krasnoo wrote: Yes. It definitely would. I didn't know Yahoo uses MySQL. What would be great for an interim is if a script could determine the space available and suggest the settings to create the table. I see that by setting RAID_CHUNKS you could tell how many directories MySQL should make. The script could evenly distribute those directories and make the links and directories for MySQL. Sent: Sunday, June 17, 2001 11:53 PM To: Jamie Krasnoo Cc: MySQL Subject: Re: raid tables On Sun, Jun 17, 2001 at 11:48:04PM -0700, Jamie Krasnoo wrote: Is there a way to tell MySQL to automatically spread its tables over a number of disks without going through the trouble of symbolic linking the chunks to the different disks? There isn't yet, no. It'd certainly be a nice addition, though, wouldn't it? As author of this RAID code I would like to know, how did you see this in your mind: - As MySQL command DISTRIBUTE TABLES ? - As some kind of external script? - Other way? I would like to implement this feature but currently I do not see an smooth way to do it. Maybe something like LINK RAID DIR 1 to /mnt/bigdisk ? -- For technical support contracts, goto https://order.mysql.com/ __ ___ ___ __ / |/ /_ __/ __/ __ \/ /Mr. Tonu Samuel [EMAIL PROTECTED] / /|_/ / // /\ \/ /_/ / /__ MySQL AB, Security Administrator /_/ /_/\_, /___/\___\_\___/ Hong Kong, China ___/ www.mysql.com - 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 [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php -- Besten Dank Philip Markwalder == Markwalder Philip Consultant [EMAIL PROTECTED] Trivadis AG Kanalstrasse 5 CH-8152 Glattbrugg Tel.: +41- 1-808 70 20 Fax : +41- 1-808 70 21 Mobile: +41-79-445 77 87 http://www.trivadis.com == - 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 [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Re: raid tables
On Mon, Jun 18, 2001 at 03:56:32AM -0700, Jamie Krasnoo wrote: [mysqld] datadir=/data1;/data2;/data3 - or - datadir=/data1:/data2:/data3 As it stands now, both will produce an error and MySQL will not start. Other ways could be like: RAID_DIRS=/data1:/data2:/data3 Well, I'll throw my thoughts in as well... I'd use a combination of my.cnf entries and SQL. The my.cnf entry (or entries) woudl list the available directories, like others have suggested. Then, using SQL, you could: * Mark a table to be spread out at creation time vis some attribute. * Possibly do it after the fact using ALTER TABLE. As for the exact syntax, I don't know what might be best. Jeremy -- Jeremy D. Zawodny, [EMAIL PROTECTED] Technical Yahoo - Yahoo Finance Desk: (408) 349-7878Fax: (408) 349-5454Cell: (408) 439-9951 MySQL 3.23.29: up 2 days, processed 16,991,404 queries (79/sec. avg) - 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 [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Re: Important bug in RAID tables
On Mon, 18 Jun 2001, Patrizio Bruno wrote: select max(field) from table returns the greatest field's value in the first file of the RAID table (successive files are not considered), adding a where clause (for ex. select max(field) from table where field2 = 200) the problem seems not to exist. I don't know if the bug is limited to the 'max()' function, but, for the use I make of mysql, this is a big problem. I think this is impossible. MySQL-s higher lever doesn't know anything about RAID and because this cannot return rows based on location in files. You did not supplied enough information to reproduce the bug or even understand its nature. From current niformation I believe that you have index file broken in smoe reason and because this MySQL of course can return wrong result not depenging if it is raided or not. Run REPAIR TABLE urlword and see if problem persists. If yes, give me a repeatable testcase by mysqldump-ing this table and giving clear example on which exact commands he gave what exact answer and I try to reproduce it on my own computer. I cannot fix it before I can reproduce it. Environment: Linux 2.2.19 MySQL 3.23.38 mysql's data directory is on a NFS partition below the create statement I used to create my table: create table urlword(url_id integer auto_increment primary key, site_id integer not null, deleted tinyint DEFAULT 0 NOT NULL, url varchar(128) not null, next_index_time INT NOT NULL, status int(11) DEFAULT '0' NOT NULL, crc char(32) DEFAULT '' NOT NULL, last_modified varchar(32) DEFAULT '' NOT NULL, etag varchar(48) DEFAULT '' NOT NULL, last_index_time INT NOT NULL, referrer int(11) DEFAULT '0' NOT NULL, tag int(11) DEFAULT '0' NOT NULL, hops int(11) DEFAULT '0' NOT NULL, redir integer, origin integer, unique index(url), index(next_index_time), index(hops,next_index_time), index crc (site_id,crc(8))) DELAY_KEY_WRITE=1 RAID_TYPE=STRIPED RAID_CHU NKS=8 RAID_CHUNKSIZE=153600 PACK_KEYS=1 TYPE=MyISAM; the mysql configuration file is my-huge.cnf taken from the mysql's source-distribution. BTW, why did you change the chunksize? By default it is 256000 and I tested it for speed with many different values. 256000 seemed to be best value. If you have any experience showing that I was wrong, please share it with me. -- For technical support contracts, goto https://order.mysql.com/ __ ___ ___ __ / |/ /_ __/ __/ __ \/ /Mr. Tonu Samuel [EMAIL PROTECTED] / /|_/ / // /\ \/ /_/ / /__ MySQL AB, Security Administrator /_/ /_/\_, /___/\___\_\___/ Hong Kong, China ___/ www.mysql.com - 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 [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Re: raid tables
On Mon, 18 Jun 2001, Jeremy Zawodny wrote: I'd use a combination of my.cnf entries and SQL. The my.cnf entry (or entries) woudl list the available directories, like others have suggested. Then, using SQL, you could: * Mark a table to be spread out at creation time vis some attribute. * Possibly do it after the fact using ALTER TABLE. Maybe you are right. I am thinking about security and this is a not the best idea if we introduce a way to write data anywhere on the disks on user request. This can make MySQL as most used cracking tool on the web ;). Meanwhile any change in cnf file requires restart of MySQL which is not always possible :(. -- For technical support contracts, goto https://order.mysql.com/ __ ___ ___ __ / |/ /_ __/ __/ __ \/ /Mr. Tonu Samuel [EMAIL PROTECTED] / /|_/ / // /\ \/ /_/ / /__ MySQL AB, Security Administrator /_/ /_/\_, /___/\___\_\___/ Hong Kong, China ___/ www.mysql.com - 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 [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
configuring MySQL for raid tables
MySQL's raid function says that it can use multiple drives/directories to put its database in to get over the 2GB limit. How do I configure MySQL to make sure that it knows about the drives when its instructed to create a table as RAID_TYPE = STRIPPED ? Would I put in my.cnf: [mysqld] datadir=/dbdata1:/dbdata2:/dbdata3 Or would I need to create the database and then move the files and use symlinks? Thanks, Jamie Krasnoo [EMAIL PROTECTED] - 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 [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Re: raid tables doesnt seem to split files.
Hi Firdaus! On Thu, 12 Oct 2000, Firdaus Mah wrote: Hi Jerome, Heh... I'm already doomed. I'ved just inserted 32 million records and its stopped at 2GB even using on reiserfs. I'ved not checked the detail or experience in Postgress development. Does it do stripping as well ie breaking up tables and index file. It is not reiserfs which is limiting you to 2GB (reiserfs is unsigned and should limit you to 4GB), however the VFS layer is only signed so it will not communicate the extra bit to reiserfs so that it will access 2GB. I suspect that if you were to apply the LFS patches it will start working, but I've never done it on a reiserfs partition. My conclusion is that until this moment, MySQL is suitable for small to medium databases. Not true, I have MySQL running on a Alpha and have none of these limitations =) -Myron firdaus Jerome Abela wrote: On Wed, Oct 11, 2000 at 12:51:50PM +0800, Firdaus Mah wrote: this is how i'ved created create table test (field char(70) not null primary key) RAID_TYPE=STRIPED RAID_CHUNKS=1024 RAID_CHUNKSIZE=1024; This only works if mysqld was compiled with the --with-raid option. You can check this by grepping 'CONFIGURE_LINE=' from mysqlbug script. Otherwise, the RAID option to CREATE is silently ignored. Note that only the data file (MYD) is stripped. If your index becomes too big, you are doomed. Jerome. -- - Please check "http://www.mysql.com/documentation/manual.php" before posting. To request this thread, e-mail [EMAIL PROTECTED] To unsubscribe, send a message to: [EMAIL PROTECTED] If you have a broken mail client that cannot send a message to the above address (Microsoft Outlook), you can use: http://lists.mysql.com/php/unsubscribe.php Emanuel.exe - 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 [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
RE: raid tables doesnt seem to split files.
RAID tables works fine for me. I don't know which OS you are running, i got it to work with a linux box like this: 1. Upgrade to the 2.4 linux kernel(Or patch an older kernel with the LFS patch) 2. Recompile mysql 3.23 with the --with-raid configure option (the precompiled binaries doesn't seem to have this feature compiled in). 3. Create a RAID table like the manual describes. Sincerely, Christian. btw: I am using ext2, but it should work the same with reiserfs. -Original Message- From: funky gao [mailto:[EMAIL PROTECTED]] Sent: Tuesday, January 16, 2001 12:41 PM To: Firdaus Mah Cc: Jerome Abela; [EMAIL PROTECTED] Subject: Re: raid tables doesnt seem to split files. Hi Firdaus! On Thu, 12 Oct 2000, Firdaus Mah wrote: Hi Jerome, Heh... I'm already doomed. I'ved just inserted 32 million records and its stopped at 2GB even using on reiserfs. I'ved not checked the detail or experience in Postgress development. Does it do stripping as well ie breaking up tables and index file. It is not reiserfs which is limiting you to 2GB (reiserfs is unsigned and should limit you to 4GB), however the VFS layer is only signed so it will not communicate the extra bit to reiserfs so that it will access 2GB. I suspect that if you were to apply the LFS patches it will start working, but I've never done it on a reiserfs partition. My conclusion is that until this moment, MySQL is suitable for small to medium databases. Not true, I have MySQL running on a Alpha and have none of these limitations =) -Myron firdaus Jerome Abela wrote: On Wed, Oct 11, 2000 at 12:51:50PM +0800, Firdaus Mah wrote: this is how i'ved created create table test (field char(70) not null primary key) RAID_TYPE=STRIPED RAID_CHUNKS=1024 RAID_CHUNKSIZE=1024; This only works if mysqld was compiled with the --with-raid option. You can check this by grepping 'CONFIGURE_LINE=' from mysqlbug script. Otherwise, the RAID option to CREATE is silently ignored. Note that only the data file (MYD) is stripped. If your index becomes too big, you are doomed. Jerome. -- - Please check "http://www.mysql.com/documentation/manual.php" before posting. To request this thread, e-mail [EMAIL PROTECTED] To unsubscribe, send a message to: [EMAIL PROTECTED] If you have a broken mail client that cannot send a message to the above address (Microsoft Outlook), you can use: http://lists.mysql.com/php/unsubscribe.php VIGILANTe.com NOTICE - AUTOMATICALLY INSERTED The information transmitted is intended only for the person or entity to which it is addressed and may contain confidential and/or privileged material. Any review, retransmission, dissemination or other use of, or taking of any action in reliance upon, this information by persons or entities other than the intended recipient is prohibited. Any opinions expressed in this email are those of the individual and not necessarily the Company. If you receive this transmission in error, please email to [EMAIL PROTECTED], including a copy of this message. Please then delete this email and destroy any copies of it. DISCLAIMER END - 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 [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
