RE: Request for help in testing new replication code in 4.0.2
I have finished a script that takes the bit keeper sources and makes MySQL-4.0.2 RPMS for Redhat 6.x and apparently 7.X as well. I can put these up on our site for download if you want. They might get others trying the code out. I use the spec file included with the source code. I'll put it up if it's something you might find useful. -Original Message- From: Steven Roussey [mailto:[EMAIL PROTECTED]] Sent: Tuesday, February 12, 2002 12:41 PM To: 'Sasha Pachev' Cc: 'Mysql' Subject: Re: Request for help in testing new replication code in 4.0.2 * If you want to do it this very moment, read the instructions at http://www.mysql.com/doc/I/n/Installing_source_tree.html and install a pre-release 4.0.2 MySQL on your test server. You may also wait until 4.0.2 is released, but in that case, replication in 4.0.2 may have a bug you could have caught on your system and it will not be fixed until 4.0.3. Is that possible to provide a binary snapshot of the pre-release? This can be done with some simple shell scripting, I think. I always found the one I build myself is about 30-50% slower then the official binary. I think that's related to the debian patched glibc-2.95 I am using. Hi, I have to agree with this. I don't want to have to factor in RedHat's 7.2 compiler oddities (which one to use?). Some software projects have a machine to do regular builds on a schedule. This would be ideal. You would find more people testing it. You might just do it in special cases like this one. Testing the new code is relatively easy, and does not require risking the stability of your application by running alpha code. Yes and no. Taking down a server and copying gigs of data can take down a service for a long time (moving gigs of data just is never fast, and the machine I'm going to use has lots of power but no RAID). I don't have time to download and install gcc, bitkeeper, autoconf 2.13, automake 1.4, libtool, m4, and mysql today, but I'll try and get to it this week. Taking down the server will have to wait until maintenance windows over the weekend anyhow. Our test plan for MySQL 4 is as follows: 1. Serialized writes (i.e. replication slave) for two weeks (or about 1 billion insert/update/delete type queries) with no crashes. 2. Everything. We will have the app send duplicates of all queries to the mysql 4 server, but will automatically halt when mysql 4 crashes. We will keep trying this until mysql doesn't crash for a week. 3. Check that the two servers hold the same information and return the same results from queries. With the type and volume of queries that we have, mysql alpha versions either fail within minutes/hours/days or they don't fail at all. We certainly are not the biggest user of mysql, but historically, we have been able to push it hard enough to crash. Not the stable version though. Thankfully, that takes a licking and keeps on ticking. :) Eventually (spring/summer), we will move the main server to v4 and keep the slave around as a live backup. I've been wanting such a slave for a long time. Stopping the slave to make a backup would be great. Not to mention faster fulltext and order by DESC. 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 - 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: Request for help in testing new replication code in 4.0.2
That would be great. Thanks! Sincerely, Steven Roussey http://Network54.com/?pp=e -Original Message- From: Brian P. Austin [mailto:[EMAIL PROTECTED]] Sent: Tuesday, February 12, 2002 10:45 pm Subject: RE: Request for help in testing new replication code in 4.0.2 I have finished a script that takes the bit keeper sources and makes MySQL-4.0.2 RPMS for Redhat 6.x and apparently 7.X as well. I can put these up on our site for download if you want. They might get others trying the code out. I use the spec file included with the source code. I'll put it up if it's something you might find useful. -Original Message- From: Steven Roussey [mailto:[EMAIL PROTECTED]] Sent: Tuesday, February 12, 2002 12:41 PM To: 'Sasha Pachev' Cc: 'Mysql' Subject: Re: Request for help in testing new replication code in 4.0.2 * If you want to do it this very moment, read the instructions at http://www.mysql.com/doc/I/n/Installing_source_tree.html and install a pre-release 4.0.2 MySQL on your test server. You may also wait until 4.0.2 is released, but in that case, replication in 4.0.2 may have a bug you could have caught on your system and it will not be fixed until 4.0.3. Is that possible to provide a binary snapshot of the pre-release? This can be done with some simple shell scripting, I think. I always found the one I build myself is about 30-50% slower then the official binary. I think that's related to the debian patched glibc-2.95 I am using. Hi, I have to agree with this. I don't want to have to factor in RedHat's 7.2 compiler oddities (which one to use?). Some software projects have a machine to do regular builds on a schedule. This would be ideal. You would find more people testing it. You might just do it in special cases like this one. Testing the new code is relatively easy, and does not require risking the stability of your application by running alpha code. Yes and no. Taking down a server and copying gigs of data can take down a service for a long time (moving gigs of data just is never fast, and the machine I'm going to use has lots of power but no RAID). I don't have time to download and install gcc, bitkeeper, autoconf 2.13, automake 1.4, libtool, m4, and mysql today, but I'll try and get to it this week. Taking down the server will have to wait until maintenance windows over the weekend anyhow. Our test plan for MySQL 4 is as follows: 1. Serialized writes (i.e. replication slave) for two weeks (or about 1 billion insert/update/delete type queries) with no crashes. 2. Everything. We will have the app send duplicates of all queries to the mysql 4 server, but will automatically halt when mysql 4 crashes. We will keep trying this until mysql doesn't crash for a week. 3. Check that the two servers hold the same information and return the same results from queries. With the type and volume of queries that we have, mysql alpha versions either fail within minutes/hours/days or they don't fail at all. We certainly are not the biggest user of mysql, but historically, we have been able to push it hard enough to crash. Not the stable version though. Thankfully, that takes a licking and keeps on ticking. :) Eventually (spring/summer), we will move the main server to v4 and keep the slave around as a live backup. I've been wanting such a slave for a long time. Stopping the slave to make a backup would be great. Not to mention faster fulltext and order by DESC. 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 - 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: Request for help in testing new replication code in 4.0.2
* If you want to do it this very moment, read the instructions at http://www.mysql.com/doc/I/n/Installing_source_tree.html and install a pre-release 4.0.2 MySQL on your test server. You may also wait until 4.0.2 is released, but in that case, replication in 4.0.2 may have a bug you could have caught on your system and it will not be fixed until 4.0.3. Is that possible to provide a binary snapshot of the pre-release? This can be done with some simple shell scripting, I think. I always found the one I build myself is about 30-50% slower then the offical binary. I think that's related to the debian patched glibc-2.95 I am using. - 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 - 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: Request for help in testing new replication code in 4.0.2
On Sat, Feb 09, 2002 at 09:41:25PM -0700, Sasha Pachev wrote: Dear MySQL users, I have just pushed my latest changes in the replication code in our 4.0 development tree, which change the slave to use two threads - I/O thread that gets the data from the master and logs it, and SQL thread which processes the logged data. I have tested it extensively, and it has passed all of my tests. However, as we all know, field testing tends to find nasty bugs that have not been caught in regression tests and visual code examination. Excellent. I'm ready to test... Testing the new code is relatively easy, and does not require risking the stability of your application by running alpha code. Here is what you need to do: Find a machine that will act as a test server. With the availablity of Linux and FreeBSD, this can be done cost-effectively by sacrificing some useless dust-gathering PC in your office. In some case, you may even try it with your own desktop. Got it. If you want to do it this very moment, read the instructions at http://www.mysql.com/doc/I/n/Installing_source_tree.html and install a pre-release 4.0.2 MySQL on your test server. You may also wait until 4.0.2 is released, but in that case, replication in 4.0.2 may have a bug you could have caught on your system and it will not be fixed until 4.0.3. I just updated my bitkeeper tree and have been having a hell of a time getting it to build. I've checked that I have automake 1.4 and autoconf 2.13. I run `BUILD/compile-pentium --prefix=/home/mysql` and it fails after several minutes. The output is large, so I've posted it here: http://public.yahoo.com/~jzawodn/mysql-build.log (the good stuff is at the end, of course...) I believe I've followed the directions properly, but let me know if it looks like I've missed anything. Thanks, 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 3 days, processed 112,608,902 queries (396/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 - 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
Request for help in testing new replication code in 4.0.2
Dear MySQL users, I have just pushed my latest changes in the replication code in our 4.0 development tree, which change the slave to use two threads - I/O thread that gets the data from the master and logs it, and SQL thread which processes the logged data. I have tested it extensively, and it has passed all of my tests. However, as we all know, field testing tends to find nasty bugs that have not been caught in regression tests and visual code examination. So I need your help with field testing of my code. For those of you who are wondering why you should - this will help us stabilize 4.0 a lot of faster, and not only replication, but also the general SQL features. So if you depend on MySQL for your mission-critical applications, helping us test our alpha is directly in your best interest, and is worth the hassle. Testing the new code is relatively easy, and does not require risking the stability of your application by running alpha code. Here is what you need to do: * Find a machine that will act as a test server. With the availablity of Linux and FreeBSD, this can be done cost-effectively by sacrificing some useless dust-gathering PC in your office. In some case, you may even try it with your own desktop. * If you want to do it this very moment, read the instructions at http://www.mysql.com/doc/I/n/Installing_source_tree.html and install a pre-release 4.0.2 MySQL on your test server. You may also wait until 4.0.2 is released, but in that case, replication in 4.0.2 may have a bug you could have caught on your system and it will not be fixed until 4.0.3. * Follow instructions at http://www.mysql.com/doc/R/e/Replication_HOWTO.html and set up the test server as a slave of one of your production machines. Add log-slave-updates and log-bin to your slave configuration for easier troublshooting. The master can be either 3.23.33+ or 4.0.2 * Monitor your slave to make sure it does not crash ( watch error log for stack trace messages), slave keeps running ( check with SHOW SLAVE STATUS), and data is consistent. * If there are problems, I will need the following: - error logs on the master and on the slave - binary logs on the master, binary logs on the slave, and relay logs on the slave ( by default, placed in datadir and called `hostname`-relay-bin.* ) - output of SHOW SLAVE STATUS - version of the master Please send all follow-up mails to [EMAIL PROTECTED] and do not forget to CC to one of the lists, depending on the nature of the message. A bug report with precise instructions on how we can repeat it should be CC'ed to [EMAIL PROTECTED], patches and comments on the source code should be CC'ed to [EMAIL PROTECTED], and everything else should be CC'ed to [EMAIL PROTECTED] Please do not send personal mails without CC'ing them to a list - three reasons: a) I may not be the best person to answer, b) somebody else can get around to answering it sooner, and c) others may want to participate in the discussion. Overall, it is generally a good idea to avoid personal-only mails to MySQL employees unless you have a strictly personal question, of course. If you are talking about MySQL, you may send it to the person you think would be the best to answer your mail, but always make sure to CC to some general alias so that others can follow the discussion. -- MySQL Development Team For technical support contracts, visit https://order.mysql.com/ __ ___ ___ __ / |/ /_ __/ __/ __ \/ / Sasha Pachev [EMAIL PROTECTED] / /|_/ / // /\ \/ /_/ / /__ MySQL AB, http://www.mysql.com/ /_/ /_/\_, /___/\___\_\___/ Provo, Utah, USA ___/ - 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 - 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: Request for help in testing new replication code in 4.0.2
Hi, In myisamchk.c replace the following line : if (argument *argument == '0') DBUG_POP(); else DBUG_PUSH(argument ? argument : d:t:o,/tmp/myisamchk.trace); by if (argument *argument == '0') { DBUG_POP(); } else { DBUG_PUSH(argument ? argument : d:t:o,/tmp/myisamchk.trace); } Like this it works perfectly for me :) Regards, Jocelyn Fournier Presence-PC - Original Message - From: Jeremy Zawodny [EMAIL PROTECTED] To: Sasha Pachev [EMAIL PROTECTED] Cc: [EMAIL PROTECTED]; [EMAIL PROTECTED] Sent: Sunday, February 10, 2002 10:20 PM Subject: Re: Request for help in testing new replication code in 4.0.2 On Sat, Feb 09, 2002 at 09:41:25PM -0700, Sasha Pachev wrote: Dear MySQL users, I have just pushed my latest changes in the replication code in our 4.0 development tree, which change the slave to use two threads - I/O thread that gets the data from the master and logs it, and SQL thread which processes the logged data. I have tested it extensively, and it has passed all of my tests. However, as we all know, field testing tends to find nasty bugs that have not been caught in regression tests and visual code examination. Excellent. I'm ready to test... Testing the new code is relatively easy, and does not require risking the stability of your application by running alpha code. Here is what you need to do: Find a machine that will act as a test server. With the availablity of Linux and FreeBSD, this can be done cost-effectively by sacrificing some useless dust-gathering PC in your office. In some case, you may even try it with your own desktop. Got it. If you want to do it this very moment, read the instructions at http://www.mysql.com/doc/I/n/Installing_source_tree.html and install a pre-release 4.0.2 MySQL on your test server. You may also wait until 4.0.2 is released, but in that case, replication in 4.0.2 may have a bug you could have caught on your system and it will not be fixed until 4.0.3. I just updated my bitkeeper tree and have been having a hell of a time getting it to build. I've checked that I have automake 1.4 and autoconf 2.13. I run `BUILD/compile-pentium --prefix=/home/mysql` and it fails after several minutes. The output is large, so I've posted it here: http://public.yahoo.com/~jzawodn/mysql-build.log (the good stuff is at the end, of course...) I believe I've followed the directions properly, but let me know if it looks like I've missed anything. Thanks, 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 3 days, processed 112,608,902 queries (396/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] - 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 - 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: Request for help in testing new replication code in 4.0.2
Hi, In myisamchk.c replace the following line : if (argument *argument == '0') DBUG_POP(); else DBUG_PUSH(argument ? argument : d:t:o,/tmp/myisamchk.trace); by if (argument *argument == '0') { DBUG_POP(); } else { DBUG_PUSH(argument ? argument : d:t:o,/tmp/myisamchk.trace); } Like this it works perfectly for me :) Regards, Jocelyn Fournier Presence-PC - Original Message - From: Jeremy Zawodny [EMAIL PROTECTED] To: Sasha Pachev [EMAIL PROTECTED] Cc: [EMAIL PROTECTED]; [EMAIL PROTECTED] Sent: Sunday, February 10, 2002 10:20 PM Subject: Re: Request for help in testing new replication code in 4.0.2 On Sat, Feb 09, 2002 at 09:41:25PM -0700, Sasha Pachev wrote: Dear MySQL users, I have just pushed my latest changes in the replication code in our 4.0 development tree, which change the slave to use two threads - I/O thread that gets the data from the master and logs it, and SQL thread which processes the logged data. I have tested it extensively, and it has passed all of my tests. However, as we all know, field testing tends to find nasty bugs that have not been caught in regression tests and visual code examination. Excellent. I'm ready to test... Testing the new code is relatively easy, and does not require risking the stability of your application by running alpha code. Here is what you need to do: Find a machine that will act as a test server. With the availablity of Linux and FreeBSD, this can be done cost-effectively by sacrificing some useless dust-gathering PC in your office. In some case, you may even try it with your own desktop. Got it. If you want to do it this very moment, read the instructions at http://www.mysql.com/doc/I/n/Installing_source_tree.html and install a pre-release 4.0.2 MySQL on your test server. You may also wait until 4.0.2 is released, but in that case, replication in 4.0.2 may have a bug you could have caught on your system and it will not be fixed until 4.0.3. I just updated my bitkeeper tree and have been having a hell of a time getting it to build. I've checked that I have automake 1.4 and autoconf 2.13. I run `BUILD/compile-pentium --prefix=/home/mysql` and it fails after several minutes. The output is large, so I've posted it here: http://public.yahoo.com/~jzawodn/mysql-build.log (the good stuff is at the end, of course...) I believe I've followed the directions properly, but let me know if it looks like I've missed anything. Thanks, 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 3 days, processed 112,608,902 queries (396/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] - 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 - 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: Request for help in testing new replication code in 4.0.2
I don't actually see the error message in the output you sent. but I found a similar problem yesterday and have the following work around: = myisamchk.c 1.69 vs edited = *** /tmp/myisamchk.c-1.69-20535 Thu Feb 7 17:21:33 2002 --- edited/myisamchk.c Sat Feb 9 22:48:57 2002 *** *** 574,582 check_param.testflag|= T_UPDATE_STATE; break; case '#': ! if (argument *argument == '0') DBUG_POP(); ! else DBUG_PUSH(argument ? argument : d:t:o,/tmp/myisamchk.trace); break; case 'V': --- 574,582 check_param.testflag|= T_UPDATE_STATE; break; case '#': ! if (argument *argument == '0') { DBUG_POP(); ! } else DBUG_PUSH(argument ? argument : d:t:o,/tmp/myisamchk.trace); break; case 'V': Of course after applying this I ran into a different problem: 020210 00:07:21 mysqld started 020210 0:07:21 Could not find first log during relay log initialization 020210 0:07:21 Warning: Can't create threads to handle slave /usr/local/libexec/mysqld: ready for connections Which is apparently because it isn't asking the master for the correct file name: mysql show slave status; +-+-+-+---+-+-++---+---+--+---+-+-+++--+-+ | Master_Host | Master_User | Master_Port | Connect_retry | Master_Log_File | Read_Master_Log_Pos | Relay_Log_File | Relay_Log_Pos | Relay_Master_Log_File | Slave_IO_Running | Slave_SQL_Running | Replicate_do_db | Replicate_ignore_db | Last_errno | Last_error | Skip_counter | Exec_master_log_pos | +-+-+-+---+-+-++---+---+--+---+-+-+++--+-+ | | | 0 | 0 | | 0 | noritake-relay-bin.001 | 4 | | No | No| | | 0 || 0| 0 | +-+-+-+---+-+-++---+---+--+---+-+-+++--+-+ 1 row in set (0.00 sec) which should be spode-bin.001 My guess is that I need an extra entry of some sort in /etc/my.cnf on the slave: [mysqld] replicate-ignore-table=curioscape.avstemp replicate-ignore-table=tias.avtemp replicate-ignore-table=inventory.availableQuantity replicate-ignore-table=inventory.thisItem replicate-ignore-table=inventory.auctionStatus replicate-ignore-table=inventory.itemTable master-host=db1.tias.com master-user=slave master-password=password log=db.log # master-port= server-id=2 But I'm nto sure what it should be. /var/lib/mysql/master.info looks OK: spode-bin.001 251797980 db1.tias.com slave password 3306 60 0 075316828 Jeremy Zawodny wrote: On Sat, Feb 09, 2002 at 09:41:25PM -0700, Sasha Pachev wrote: Dear MySQL users, I have just pushed my latest changes in the replication code in our 4.0 development tree, which change the slave to use two threads - I/O thread that gets the data from the master and logs it, and SQL thread which processes the logged data. I have tested it extensively, and it has passed all of my tests. However, as we all know, field testing tends to find nasty bugs that have not been caught in regression tests and visual code examination. Excellent. I'm ready to test... Testing the new code is relatively easy, and does not require risking the stability of your application by running alpha code. Here is what you need to do: Find a machine that will act as a test server. With the availablity of Linux and FreeBSD, this can be done cost-effectively by sacrificing some useless dust-gathering PC in your office. In some case, you may even try it with your own desktop. Got it. If you want to do it this very moment, read the instructions at http://www.mysql.com/doc/I/n/Installing_source_tree.html and install a pre-release 4.0.2 MySQL on your test server. You may also wait until 4.0.2 is released, but in that case, replication in 4.0.2 may have a bug you could have caught on your system and it will not be fixed until 4.0.3. I just updated my bitkeeper tree and have been having a hell of a time getting it to build. I've checked that I have automake 1.4 and autoconf 2.13. I run `BUILD/compile-pentium --prefix=/home/mysql` and it fails after several
Re: Request for help in testing new replication code in 4.0.2
On Sat, Feb 09, 2002 at 09:41:25PM -0700, Sasha Pachev wrote: So I need your help with field testing of my code. For those of you who are wondering why you should - this will help us stabilize 4.0 a lot of faster, and not only replication, but also the general SQL features. So if you depend on MySQL for your mission-critical applications, helping us test our alpha is directly in your best interest, and is worth the hassle. I upgraded one of our slaves from 3.23.47 to 4.0.2-alpha from bitkeeper. It's happily replicating off its master. 200,000 queries and going. It'll be a lot busier come Monday morning. One question: Can you explain the relay log a bit. Does it shrink eventually? I see it growing and growing, so I don't know if the space is recycled, or if I need to do something to periodically flush the executed queries from the log. Okay, a second question: How can I control where the relay log goes? I always have my binlog go to a special directory on a different disk from my active tables. I'd like to do something similar for the relay log. Thanks, 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 3 days, processed 119,729,462 queries (401/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 - 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: Request for help in testing new replication code in 4.0.2
* If you want to do it this very moment, read the instructions at http://www.mysql.com/doc/I/n/Installing_source_tree.html and install a pre-release 4.0.2 MySQL on your test server. You may also wait until 4.0.2 is released, but in that case, replication in 4.0.2 may have a bug you could have caught on your system and it will not be fixed until 4.0.3. Is that possible to provide a binary snapshot of the pre-release? This can be done with some simple shell scripting, I think. I always found the one I build myself is about 30-50% slower then the official binary. I think that's related to the debian patched glibc-2.95 I am using. Hi, I have to agree with this. I don't want to have to factor in RedHat's 7.2 compiler oddities (which one to use?). Some software projects have a machine to do regular builds on a schedule. This would be ideal. You would find more people testing it. You might just do it in special cases like this one. Testing the new code is relatively easy, and does not require risking the stability of your application by running alpha code. Yes and no. Taking down a server and copying gigs of data can take down a service for a long time (moving gigs of data just is never fast, and the machine I'm going to use has lots of power but no RAID). I don't have time to download and install gcc, bitkeeper, autoconf 2.13, automake 1.4, libtool, m4, and mysql today, but I'll try and get to it this week. Taking down the server will have to wait until maintenance windows over the weekend anyhow. Our test plan for MySQL 4 is as follows: 1. Serialized writes (i.e. replication slave) for two weeks (or about 1 billion insert/update/delete type queries) with no crashes. 2. Everything. We will have the app send duplicates of all queries to the mysql 4 server, but will automatically halt when mysql 4 crashes. We will keep trying this until mysql doesn't crash for a week. 3. Check that the two servers hold the same information and return the same results from queries. With the type and volume of queries that we have, mysql alpha versions either fail within minutes/hours/days or they don't fail at all. We certainly are not the biggest user of mysql, but historically, we have been able to push it hard enough to crash. Not the stable version though. Thankfully, that takes a licking and keeps on ticking. :) Eventually (spring/summer), we will move the main server to v4 and keep the slave around as a live backup. I've been wanting such a slave for a long time. Stopping the slave to make a backup would be great. Not to mention faster fulltext and order by DESC. 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: Request for help in testing new replication code in 4.0.2
It could also have been fixed by changing the: DBUG_POP(); to: DBUG_POP() Notice the ; on the end effectively terminating the if statement. Happy coding. Chris -Original Message- From: Jeremy Zawodny [EMAIL PROTECTED] [mailto:X] Sent: Sunday, February 10, 2002 3:28 PM To: Subject: Re: Request for help in testing new replication code in 4.0.2 On Sun, Feb 10, 2002 at 10:24:30PM +0100, Fournier Jocelyn [Presence-PC] wrote: Hi, In myisamchk.c replace the following line : if (argument *argument == '0') DBUG_POP(); else DBUG_PUSH(argument ? argument : d:t:o,/tmp/myisamchk.trace); by if (argument *argument == '0') { DBUG_POP(); } else { DBUG_PUSH(argument ? argument : d:t:o,/tmp/myisamchk.trace); } Like this it works perfectly for me :) Yeah, that works. I had just commented out that bit and moved on. Worked fine after that. Since I didn't know if it was just a typo or the result of an incomplete commit, I didn't want to try and fix it. But I'll probably use you solution until a fix is pushed. Thanks, 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 3 days, processed 113,132,847 queries (396/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 - 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 - 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: Request for help in testing new replication code in 4.0.2
On Sunday 10 February 2002 01:17 am, you wrote: I can't get it to start replicating. Note, the master (db1.tias.com) is a 3.23 server. The slave is 4.0.2 (noritake.tias.com). On the master I get: mysql show master status; +---++--+--+ | File ? ? ? ? ?| Position ? | Binlog_do_db | Binlog_ignore_db | +---++--+--+ | spode-bin.020 | 1058168617 | ? ? ? ? ? ? ?| ? ? ? ? ? ? ? ? ?| +---++--+--+ 1 row in set (0.00 sec) mysql show slave status; +-+-+-+---+-+-++---+---+--+---+-+-+++--+-+ | Master_Host | Master_User | Master_Port | Connect_retry | Master_Log_File | Read_Master_Log_Pos | Relay_Log_File ? ? ? ? | Relay_Log_Pos | Relay_Master_Log_File | Slave_IO_Running | Slave_SQL_Running | Replicate_do_db | Replicate_ignore_db | Last_errno | Last_error | Skip_counter | Exec_master_log_pos | +-+-+-+---+-+-++---+---+--+---+-+-+++--+-+ | ? ? ? ? ? ? | ? ? ? ? ? ? | 0 ? ? ? ? ? | 0 ? ? ? ? ? ? | ? ? ? ? ? ? ? ? | 0 ? ? ? ? ? ? ? ? ? ?| noritake-relay-bin.001 | 4 ? ? ? ? ? ? | ? ? ?| No ? ? ? ? ? ? ? | No ? ? ? ? ? ? ? ?| ? ? ? ? ? ? ? ? | ? ? ?| 0 ? ? ? ? ?| ? ? ? ? ? ?| 0 ? ? ? ? ? ?| 0 ? ? ? ? ? ? ? ? ? | +-+-+-+---+-+-++---+---+--+---+-+-+++--+-+ 1 row in set (0.01 sec) and in noritake.tias.com.err: 020210 00:07:21 ?mysqld started 020210 ?0:07:21 ?Could not find first log during relay log initialization 020210 ?0:07:21 ?Warning: Can't create threads to handle slave /usr/local/libexec/mysqld: ready for connections and in /etc/my.cnf on the slave: [mysqld] replicate-ignore-table=curioscape.avstemp replicate-ignore-table=tias.avtemp replicate-ignore-table=inventory.availableQuantity replicate-ignore-table=inventory.thisItem replicate-ignore-table=inventory.auctionStatus replicate-ignore-table=inventory.itemTable master-host=db1.tias.com master-user=slave master-password=password log=db.log # master-port= server-id=2 and on the server: [mysqld] set-variable = tmp_table_size=8M set-variable = key_buffer=384M set-variable = max_allowed_packet=2M set-variable = max_connections=1024 set-variable = table_cache=8192 set-variable = thread_cache_size=32 set-variable = sort_buffer=4M set-variable = record_buffer=2M log-slow-queries log-bin server-id=1 It appears the problem is that the new replication code is using a different file name convention? This set up did work before. Until the 4.0.1 MySQL starting crashing because of a fulltext index bug that I reported earlier. Mike: Thanks for reporting the bug. To troubleshoot it, lets do the following: in slave.cc around line 186, you find the following code: if (rli-relay_log.find_first_log(rli-linfo,log)) { *errmsg=Could not find first log during relay log initialization; goto err; } make it: if (rli-relay_log.find_first_log(rli-linfo,log)) { sql_print_error(Could not find first log during relay initialization); abort(); *errmsg=Could not find first log during relay log initialization; goto err; } and then run BUILD/compile-pentium-debug and install the new binary. It will coredump when it hits the error. You will find a core file ( core.pid_of_mysql) in your datadir. FTP the core file to support.mysql.com/pub/mysql/secret along with your binary and a readme file all in a tar archive. Thanks again for your efforts to help us make MySQL 4.0 more stable. -- MySQL Development Team For technical support contracts, visit https://order.mysql.com/ __ ___ ___ __ / |/ /_ __/ __/ __ \/ / Sasha Pachev [EMAIL PROTECTED] / /|_/ / // /\ \/ /_/ / /__ MySQL AB, http://www.mysql.com/ /_/ /_/\_, /___/\___\_\___/ Provo, Utah, USA ___/ - 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:
Re: Request for help in testing new replication code in 4.0.2
On Sunday 10 February 2002 06:06 pm, Jeremy Zawodny wrote: One question: ? Can you explain the relay log a bit. ?Does it shrink eventually? ?I ? see it growing and growing, so I don't know if the space is ? recycled, or if I need to do something to periodically flush the ? executed queries from the log. Relay log rotates once a written event pushes it over max_binlog_size in the I/O thread, and the SQL thread deletes a processed relay log when switching to the next one. So if it works without bugs, you can just leave it unattended and it should take care of itself. Okay, a second question: ? How can I control where the relay log goes? ?I always have my binlog ? go to a special directory on a different disk from my active tables. ? I'd like to do something similar for the relay log. relay-log=/path/to/relay-log similar to log-bin= Relay log in essence is just another binary log, and most of the code is reused from the log-bin. -- MySQL Development Team For technical support contracts, visit https://order.mysql.com/ __ ___ ___ __ / |/ /_ __/ __/ __ \/ / Sasha Pachev [EMAIL PROTECTED] / /|_/ / // /\ \/ /_/ / /__ MySQL AB, http://www.mysql.com/ /_/ /_/\_, /___/\___\_\___/ Provo, Utah, USA ___/ - 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: Request for help in testing new replication code in 4.0.2
On Sunday 10 February 2002 01:17 am, you wrote: I can't get it to start replicating. Note, the master (db1.tias.com) is a 3.23 server. The slave is 4.0.2 (noritake.tias.com). On the master I get: mysql show master status; +---++--+--+ | File ? ? ? ? ?| Position ? | Binlog_do_db | Binlog_ignore_db | +---++--+--+ | spode-bin.020 | 1058168617 | ? ? ? ? ? ? ?| ? ? ? ? ? ? ? ? ?| +---++--+--+ 1 row in set (0.00 sec) mysql show slave status; +-+-+-+---+-+-++---+---+--+---+-+-+++--+-+ | Master_Host | Master_User | Master_Port | Connect_retry | Master_Log_File | Read_Master_Log_Pos | Relay_Log_File ? ? ? ? | Relay_Log_Pos | Relay_Master_Log_File | Slave_IO_Running | Slave_SQL_Running | Replicate_do_db | Replicate_ignore_db | Last_errno | Last_error | Skip_counter | Exec_master_log_pos | +-+-+-+---+-+-++---+---+--+---+-+-+++--+-+ | ? ? ? ? ? ? | ? ? ? ? ? ? | 0 ? ? ? ? ? | 0 ? ? ? ? ? ? | ? ? ? ? ? ? ? ? | 0 ? ? ? ? ? ? ? ? ? ?| noritake-relay-bin.001 | 4 ? ? ? ? ? ? | ? ? ?| No ? ? ? ? ? ? ? | No ? ? ? ? ? ? ? ?| ? ? ? ? ? ? ? ? | ? ? ?| 0 ? ? ? ? ?| ? ? ? ? ? ?| 0 ? ? ? ? ? ?| 0 ? ? ? ? ? ? ? ? ? | +-+-+-+---+-+-++---+---+--+---+-+-+++--+-+ 1 row in set (0.01 sec) and in noritake.tias.com.err: 020210 00:07:21 ?mysqld started 020210 ?0:07:21 ?Could not find first log during relay log initialization 020210 ?0:07:21 ?Warning: Can't create threads to handle slave /usr/local/libexec/mysqld: ready for connections and in /etc/my.cnf on the slave: [mysqld] replicate-ignore-table=curioscape.avstemp replicate-ignore-table=tias.avtemp replicate-ignore-table=inventory.availableQuantity replicate-ignore-table=inventory.thisItem replicate-ignore-table=inventory.auctionStatus replicate-ignore-table=inventory.itemTable master-host=db1.tias.com master-user=slave master-password=password log=db.log # master-port= server-id=2 and on the server: [mysqld] set-variable = tmp_table_size=8M set-variable = key_buffer=384M set-variable = max_allowed_packet=2M set-variable = max_connections=1024 set-variable = table_cache=8192 set-variable = thread_cache_size=32 set-variable = sort_buffer=4M set-variable = record_buffer=2M log-slow-queries log-bin server-id=1 It appears the problem is that the new replication code is using a different file name convention? This set up did work before. Until the 4.0.1 MySQL starting crashing because of a fulltext index bug that I reported earlier. Mike: Thanks for reporting the bug. To troubleshoot it, lets do the following: in slave.cc around line 186, you find the following code: if (rli-relay_log.find_first_log(rli-linfo,log)) { *errmsg=Could not find first log during relay log initialization; goto err; } make it: if (rli-relay_log.find_first_log(rli-linfo,log)) { sql_print_error(Could not find first log during relay initialization); abort(); *errmsg=Could not find first log during relay log initialization; goto err; } and then run BUILD/compile-pentium-debug and install the new binary. It will coredump when it hits the error. You will find a core file ( core.pid_of_mysql) in your datadir. FTP the core file to support.mysql.com/pub/mysql/secret along with your binary and a readme file all in a tar archive. Thanks again for your efforts to help us make MySQL 4.0 more stable. -- MySQL Development Team For technical support contracts, visit https://order.mysql.com/ __ ___ ___ __ / |/ /_ __/ __/ __ \/ / Sasha Pachev [EMAIL PROTECTED] / /|_/ / // /\ \/ /_/ / /__ MySQL AB, http://www.mysql.com/ /_/ /_/\_, /___/\___\_\___/ Provo, Utah, USA ___/ - 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:
Re: Request for help in testing new replication code in 4.0.2
On Sunday 10 February 2002 02:20 pm, Jeremy Zawodny wrote: I run `BUILD/compile-pentium --prefix=/home/mysql` and it fails after several minutes. ?The output is large, so I've posted it here: ? http://public.yahoo.com/~jzawodn/mysql-build.log (the good stuff is at the end, of course...) This was a bug myisamchk.c that Monty has fixed but had not pushed by the time I thought he had. He has now pushed it. I realize you had to edit myisamchk.c to get MySQL to compile without debug, so you would have to do bk unedit myisamchk.c before the pull will work. -- MySQL Development Team For technical support contracts, visit https://order.mysql.com/ __ ___ ___ __ / |/ /_ __/ __/ __ \/ / Sasha Pachev [EMAIL PROTECTED] / /|_/ / // /\ \/ /_/ / /__ MySQL AB, http://www.mysql.com/ /_/ /_/\_, /___/\___\_\___/ Provo, Utah, USA ___/ - 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: Request for help in testing new replication code in 4.0.2
* If you want to do it this very moment, read the instructions at http://www.mysql.com/doc/I/n/Installing_source_tree.html and install a pre-release 4.0.2 MySQL on your test server. You may also wait until 4.0.2 is released, but in that case, replication in 4.0.2 may have a bug you could have caught on your system and it will not be fixed until 4.0.3. Is that possible to provide a binary snapshot of the pre-release? This can be done with some simple shell scripting, I think. I always found the one I build myself is about 30-50% slower then the offical binary. I think that's related to the debian patched glibc-2.95 I am using. - 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: Request for help in testing new replication code in 4.0.2
On Sat, Feb 09, 2002 at 09:41:25PM -0700, Sasha Pachev wrote: Dear MySQL users, I have just pushed my latest changes in the replication code in our 4.0 development tree, which change the slave to use two threads - I/O thread that gets the data from the master and logs it, and SQL thread which processes the logged data. I have tested it extensively, and it has passed all of my tests. However, as we all know, field testing tends to find nasty bugs that have not been caught in regression tests and visual code examination. Excellent. I'm ready to test... Testing the new code is relatively easy, and does not require risking the stability of your application by running alpha code. Here is what you need to do: Find a machine that will act as a test server. With the availablity of Linux and FreeBSD, this can be done cost-effectively by sacrificing some useless dust-gathering PC in your office. In some case, you may even try it with your own desktop. Got it. If you want to do it this very moment, read the instructions at http://www.mysql.com/doc/I/n/Installing_source_tree.html and install a pre-release 4.0.2 MySQL on your test server. You may also wait until 4.0.2 is released, but in that case, replication in 4.0.2 may have a bug you could have caught on your system and it will not be fixed until 4.0.3. I just updated my bitkeeper tree and have been having a hell of a time getting it to build. I've checked that I have automake 1.4 and autoconf 2.13. I run `BUILD/compile-pentium --prefix=/home/mysql` and it fails after several minutes. The output is large, so I've posted it here: http://public.yahoo.com/~jzawodn/mysql-build.log (the good stuff is at the end, of course...) I believe I've followed the directions properly, but let me know if it looks like I've missed anything. Thanks, 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 3 days, processed 112,608,902 queries (396/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: Request for help in testing new replication code in 4.0.2
Hi, In myisamchk.c replace the following line : if (argument *argument == '0') DBUG_POP(); else DBUG_PUSH(argument ? argument : d:t:o,/tmp/myisamchk.trace); by if (argument *argument == '0') { DBUG_POP(); } else { DBUG_PUSH(argument ? argument : d:t:o,/tmp/myisamchk.trace); } Like this it works perfectly for me :) Regards, Jocelyn Fournier Presence-PC - Original Message - From: Jeremy Zawodny [EMAIL PROTECTED] To: Sasha Pachev [EMAIL PROTECTED] Cc: [EMAIL PROTECTED]; [EMAIL PROTECTED] Sent: Sunday, February 10, 2002 10:20 PM Subject: Re: Request for help in testing new replication code in 4.0.2 On Sat, Feb 09, 2002 at 09:41:25PM -0700, Sasha Pachev wrote: Dear MySQL users, I have just pushed my latest changes in the replication code in our 4.0 development tree, which change the slave to use two threads - I/O thread that gets the data from the master and logs it, and SQL thread which processes the logged data. I have tested it extensively, and it has passed all of my tests. However, as we all know, field testing tends to find nasty bugs that have not been caught in regression tests and visual code examination. Excellent. I'm ready to test... Testing the new code is relatively easy, and does not require risking the stability of your application by running alpha code. Here is what you need to do: Find a machine that will act as a test server. With the availablity of Linux and FreeBSD, this can be done cost-effectively by sacrificing some useless dust-gathering PC in your office. In some case, you may even try it with your own desktop. Got it. If you want to do it this very moment, read the instructions at http://www.mysql.com/doc/I/n/Installing_source_tree.html and install a pre-release 4.0.2 MySQL on your test server. You may also wait until 4.0.2 is released, but in that case, replication in 4.0.2 may have a bug you could have caught on your system and it will not be fixed until 4.0.3. I just updated my bitkeeper tree and have been having a hell of a time getting it to build. I've checked that I have automake 1.4 and autoconf 2.13. I run `BUILD/compile-pentium --prefix=/home/mysql` and it fails after several minutes. The output is large, so I've posted it here: http://public.yahoo.com/~jzawodn/mysql-build.log (the good stuff is at the end, of course...) I believe I've followed the directions properly, but let me know if it looks like I've missed anything. Thanks, 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 3 days, processed 112,608,902 queries (396/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] - 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: Request for help in testing new replication code in 4.0.2
I don't actually see the error message in the output you sent. but I found a similar problem yesterday and have the following work around: = myisamchk.c 1.69 vs edited = *** /tmp/myisamchk.c-1.69-20535 Thu Feb 7 17:21:33 2002 --- edited/myisamchk.c Sat Feb 9 22:48:57 2002 *** *** 574,582 check_param.testflag|= T_UPDATE_STATE; break; case '#': ! if (argument *argument == '0') DBUG_POP(); ! else DBUG_PUSH(argument ? argument : d:t:o,/tmp/myisamchk.trace); break; case 'V': --- 574,582 check_param.testflag|= T_UPDATE_STATE; break; case '#': ! if (argument *argument == '0') { DBUG_POP(); ! } else DBUG_PUSH(argument ? argument : d:t:o,/tmp/myisamchk.trace); break; case 'V': Of course after applying this I ran into a different problem: 020210 00:07:21 mysqld started 020210 0:07:21 Could not find first log during relay log initialization 020210 0:07:21 Warning: Can't create threads to handle slave /usr/local/libexec/mysqld: ready for connections Which is apparently because it isn't asking the master for the correct file name: mysql show slave status; +-+-+-+---+-+-++---+---+--+---+-+-+++--+-+ | Master_Host | Master_User | Master_Port | Connect_retry | Master_Log_File | Read_Master_Log_Pos | Relay_Log_File | Relay_Log_Pos | Relay_Master_Log_File | Slave_IO_Running | Slave_SQL_Running | Replicate_do_db | Replicate_ignore_db | Last_errno | Last_error | Skip_counter | Exec_master_log_pos | +-+-+-+---+-+-++---+---+--+---+-+-+++--+-+ | | | 0 | 0 | | 0 | noritake-relay-bin.001 | 4 | | No | No| | | 0 || 0| 0 | +-+-+-+---+-+-++---+---+--+---+-+-+++--+-+ 1 row in set (0.00 sec) which should be spode-bin.001 My guess is that I need an extra entry of some sort in /etc/my.cnf on the slave: [mysqld] replicate-ignore-table=curioscape.avstemp replicate-ignore-table=tias.avtemp replicate-ignore-table=inventory.availableQuantity replicate-ignore-table=inventory.thisItem replicate-ignore-table=inventory.auctionStatus replicate-ignore-table=inventory.itemTable master-host=db1.tias.com master-user=slave master-password=password log=db.log # master-port= server-id=2 But I'm nto sure what it should be. /var/lib/mysql/master.info looks OK: spode-bin.001 251797980 db1.tias.com slave password 3306 60 0 075316828 Jeremy Zawodny wrote: On Sat, Feb 09, 2002 at 09:41:25PM -0700, Sasha Pachev wrote: Dear MySQL users, I have just pushed my latest changes in the replication code in our 4.0 development tree, which change the slave to use two threads - I/O thread that gets the data from the master and logs it, and SQL thread which processes the logged data. I have tested it extensively, and it has passed all of my tests. However, as we all know, field testing tends to find nasty bugs that have not been caught in regression tests and visual code examination. Excellent. I'm ready to test... Testing the new code is relatively easy, and does not require risking the stability of your application by running alpha code. Here is what you need to do: Find a machine that will act as a test server. With the availablity of Linux and FreeBSD, this can be done cost-effectively by sacrificing some useless dust-gathering PC in your office. In some case, you may even try it with your own desktop. Got it. If you want to do it this very moment, read the instructions at http://www.mysql.com/doc/I/n/Installing_source_tree.html and install a pre-release 4.0.2 MySQL on your test server. You may also wait until 4.0.2 is released, but in that case, replication in 4.0.2 may have a bug you could have caught on your system and it will not be fixed until 4.0.3. I just updated my bitkeeper tree and have been having a hell of a time getting it to build. I've checked that I have automake 1.4 and autoconf 2.13. I run `BUILD/compile-pentium --prefix=/home/mysql` and it fails after several
Re: Request for help in testing new replication code in 4.0.2
On Sat, Feb 09, 2002 at 09:41:25PM -0700, Sasha Pachev wrote: So I need your help with field testing of my code. For those of you who are wondering why you should - this will help us stabilize 4.0 a lot of faster, and not only replication, but also the general SQL features. So if you depend on MySQL for your mission-critical applications, helping us test our alpha is directly in your best interest, and is worth the hassle. I upgraded one of our slaves from 3.23.47 to 4.0.2-alpha from bitkeeper. It's happily replicating off its master. 200,000 queries and going. It'll be a lot busier come Monday morning. One question: Can you explain the relay log a bit. Does it shrink eventually? I see it growing and growing, so I don't know if the space is recycled, or if I need to do something to periodically flush the executed queries from the log. Okay, a second question: How can I control where the relay log goes? I always have my binlog go to a special directory on a different disk from my active tables. I'd like to do something similar for the relay log. Thanks, 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 3 days, processed 119,729,462 queries (401/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
Request for help in testing new replication code in 4.0.2
Dear MySQL users, I have just pushed my latest changes in the replication code in our 4.0 development tree, which change the slave to use two threads - I/O thread that gets the data from the master and logs it, and SQL thread which processes the logged data. I have tested it extensively, and it has passed all of my tests. However, as we all know, field testing tends to find nasty bugs that have not been caught in regression tests and visual code examination. So I need your help with field testing of my code. For those of you who are wondering why you should - this will help us stabilize 4.0 a lot of faster, and not only replication, but also the general SQL features. So if you depend on MySQL for your mission-critical applications, helping us test our alpha is directly in your best interest, and is worth the hassle. Testing the new code is relatively easy, and does not require risking the stability of your application by running alpha code. Here is what you need to do: * Find a machine that will act as a test server. With the availablity of Linux and FreeBSD, this can be done cost-effectively by sacrificing some useless dust-gathering PC in your office. In some case, you may even try it with your own desktop. * If you want to do it this very moment, read the instructions at http://www.mysql.com/doc/I/n/Installing_source_tree.html and install a pre-release 4.0.2 MySQL on your test server. You may also wait until 4.0.2 is released, but in that case, replication in 4.0.2 may have a bug you could have caught on your system and it will not be fixed until 4.0.3. * Follow instructions at http://www.mysql.com/doc/R/e/Replication_HOWTO.html and set up the test server as a slave of one of your production machines. Add log-slave-updates and log-bin to your slave configuration for easier troublshooting. The master can be either 3.23.33+ or 4.0.2 * Monitor your slave to make sure it does not crash ( watch error log for stack trace messages), slave keeps running ( check with SHOW SLAVE STATUS), and data is consistent. * If there are problems, I will need the following: - error logs on the master and on the slave - binary logs on the master, binary logs on the slave, and relay logs on the slave ( by default, placed in datadir and called `hostname`-relay-bin.* ) - output of SHOW SLAVE STATUS - version of the master Please send all follow-up mails to [EMAIL PROTECTED] and do not forget to CC to one of the lists, depending on the nature of the message. A bug report with precise instructions on how we can repeat it should be CC'ed to [EMAIL PROTECTED], patches and comments on the source code should be CC'ed to [EMAIL PROTECTED], and everything else should be CC'ed to [EMAIL PROTECTED] Please do not send personal mails without CC'ing them to a list - three reasons: a) I may not be the best person to answer, b) somebody else can get around to answering it sooner, and c) others may want to participate in the discussion. Overall, it is generally a good idea to avoid personal-only mails to MySQL employees unless you have a strictly personal question, of course. If you are talking about MySQL, you may send it to the person you think would be the best to answer your mail, but always make sure to CC to some general alias so that others can follow the discussion. -- MySQL Development Team For technical support contracts, visit https://order.mysql.com/ __ ___ ___ __ / |/ /_ __/ __/ __ \/ / Sasha Pachev [EMAIL PROTECTED] / /|_/ / // /\ \/ /_/ / /__ MySQL AB, http://www.mysql.com/ /_/ /_/\_, /___/\___\_\___/ Provo, Utah, USA ___/ - 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