- discuss - lists.mariadb.org

[Maria-discuss] R: Re: R: Re: Switch from rsync to xtrabackup
by absolutely_free＠libero.it 18 Feb '15

18 Feb '15

Hi Nirbhay, thank you very much for your answer. This cluster is already in production, with few databases (for mail user authentication, radius authentication, and so on). My question is: what would will happen if I change wsrep_sst_method on one node? How the other node can be "synced" if wsrep_sst_method is different (even for a few seconds)? Thank you, regards

2 1

[Maria-discuss] R: Re: Switch from rsync to xtrabackup
by absolutely_free＠libero.it 18 Feb '15

18 Feb '15

Hi, thank you for your reply. What is the correct order for this step? For example: 1) shutdown node 1 2) change config to xtrabackup-v2 on node 1 3) restart node 1? 4) shutdown node 2 5) change config to xtrabackup-v2 on node 2 6) restart node 2 ..and what about garbd on the third node? :) Thank you very much ----Messaggio originale---- Da: nirbhay(a)mariadb.com Data: 17/02/2015 14.22 A: "absolutely_free(a)libero.it"<absolutely_free(a)libero.it> Cc: "MariaDB discuss"<maria-discuss(a)lists.launchpad.net> Ogg: Re: [Maria-discuss] Switch from rsync to xtrabackup Hi! On Tue, Feb 17, 2015 at 6:07 AM, absolutely_free(a)libero.it <absolutely_free(a)libero.it> wrote: Hi, I am using two nodes (Debian 7.8 64bit) with MariaDB (mariadb-galera-server 5.5.41) and garbd on a third node. Now I'm using rsync method, I would like to switch to xtrabackup to avoid "lock" issue. Is it possible? Yes, the lock is taken only for syncing system tables. Should I simply install percona package and change this configuration? Right, the package name is "percona-xtrabackup". wsrep_sst_method=rsync to wsrep_sst_method=xtrabackup ? Or rather go for "xtrabackup-v2".https://mariadb.com/kb/en/mariadb/galera-cluster-system-variables/#wsrep_sst_method SST methods other than "rsync" also require "wsrep_sst_auth" variables to be set.https://mariadb.com/kb/en/mariadb/galera-cluster-system-variables/#wsre… Best,Nirbhay Thank you very much _______________________________________________ Mailing list: https://launchpad.net/~maria-discuss Post to : maria-discuss(a)lists.launchpad.net Unsubscribe : https://launchpad.net/~maria-discuss More help : https://help.launchpad.net/ListHelp

2 1

[Maria-discuss] Network tuning
by absolutely_free＠libero.it 18 Feb '15

18 Feb '15

Hi all, according to this URL: http://www.fromdual.com/huge-amount-of-time-wait-connections is it recommended to change default behavior of tcp_tw_reuse and tcp_tw_recycle? I am using two MariaDB nodes with haproxy, keepalived and garbd on third node. Thank you very much

1 0

[Maria-discuss] INNODB_BUFFER_PAGE
by Rhys Campbell 17 Feb '15

17 Feb '15

Hi All, I've been playing with the innodb buffer pool save / load feature in MariaDB 10.0.14. I initially thought it wasn't working but I've since identified the cause of my confusion The buffer pool was being populated with the correct number of pages but no table/index name values were given in information_schema. INNODB_BUFFER_PAGE. These details are populated as soon as the database has been access. Steps to repeat. 1. Ensure buffer pool save/load is on. 2. Record the SPACE & PAGE_NUMBER details of a page in information_schema.INNODB_BUFFER_PAGE for one of your databases. 3. Stop MariaDB. 4. Start MariaDB 5. Allow time for buffer pool to load... 6. SELECT * FROM information_schema.INNODB_BUFFER_PAGE WHERE SPACE = 157 AND PAGE_NUMBER = 99212; <- NULL values for TABLE_NAME / INDEX_NAME 7. Execute USE <db_name> # Where db_name is the database containing the above page. 8. Repeat above select and TABLE_NAME and INDEX_NAME will be populated correctly. Now, on a busy database this wouldn't be an issue but it did throw me for a while on my dev box. I have found a MySQL bug report... http://lists.mysql.com/announce/847 "When running a query on INFORMATION_SCHEMA.INNODB_BUFFER_PAGE that requested table_name and index_name values, query results would include index pages without table_name or index_name values. (Bug #14529666)" But the bug has been deleted so I can't check the full details to see if this is resolved or still an issue. As an aside the logging for the buffer pool loading feature appears to use a different datetime format... 150217 18:03:37 [Note] InnoDB: Percona XtraDB (http://www.percona.com) 5.6.20-68.0 started; log sequence number 103721589973 2015-02-17 18:03:37 7f53285f8700 InnoDB: Loading buffer pool(s) from /log/file/path/... Rhys Campbell Database Administrator TradingScreen, Inc. 23 York House, 5th Floor London WC2B 6UJ Email: rhys.james.campbell(a)googlemail.com<mailto:rhys.james.campbell@googlemail.com> Follow TradingScreen on Twitter<http://twitter.com/#!/TradingScreen> , Facebook<http://www.facebook.com/pages/TradingScreen/214046251945650> and our blog Trading Smarter<tradingsmarter.tradingscreen.com> This message is intended only for the recipient(s) named above and may contain confidential information. If you are not an intended recipient, you should not review, distribute or copy this message. Please notify the sender immediately by e-mail if you have received this message in error and delete it from your system.

1 0

[Maria-discuss] Switch from rsync to xtrabackup
by absolutely_free＠libero.it 17 Feb '15

17 Feb '15

Hi, I am using two nodes (Debian 7.8 64bit) with MariaDB (mariadb-galera-server 5.5.41) and garbd on a third node. Now I'm using rsync method, I would like to switch to xtrabackup to avoid "lock" issue. Is it possible? Should I simply install percona package and change this configuration? wsrep_sst_method=rsync to wsrep_sst_method=xtrabackup ? Thank you very much

2 1

Re: [Maria-discuss] MariaDB 10 multisource replication from AWS MySQL 5.6 RDS and Percona Server 5.6
by Reindl Harald 16 Feb '15

16 Feb '15

Am 16.02.2015 um 22:53 schrieb Vadim Pestovnikov: > Thank you Reindl. > > So basically what that means is that mariadb is not 100% compatable with > percona and mysql? define 100% - replication is a *binary log* if it would be identical in every aspect it would exist only one product and even if all docs pretend 100% compatibility i would never come to the idea even run different minor versions in a replication setup just follow these instructions leads to the conclusion mix different products is a bad idea: http://dev.mysql.com/doc/refman/5.0/en/replication-upgrade.html P.S.: i don't appreciate off-list replies > On Feb 16, 2015 1:23 PM, "Reindl Harald" <h.reindl(a)thelounge.net > <mailto:h.reindl@thelounge.net>> wrote: > > > > Am 16.02.2015 um 22:19 schrieb Vadim Pestovnikov: > > Error in Log_event::read_log_event(): 'Found invalid event in binary > log', data_len: 44, event_type: 33 > > Error reading relay log event: slave SQL thread aborted because > of I/O error > > My slave is MariaDB 10 with data and logs partitions encrypted > by LUKS > and with SSL enabled. As far as RDS and Percona servers don't > use SSL I > am using non SSL user for the replication. > > Any recommendations or suggestions would be extremely appreciated > > > don't mix different server versions in replication setups > > there is a good reason why it is recommended always to update first > the slaves, run "mysql_upgrade" and then the masters to be 100% sure > that the maste rnever don't write anything into the binlog the slave > won't understand because differences in details

2 1

[Maria-discuss] MariaDB 10 multisource replication from AWS MySQL 5.6 RDS and Percona Server 5.6
by Vadim Pestovnikov 16 Feb '15

16 Feb '15

Hi all, I am trying to setup the multisource replication from AWS MySQL 5.6 RDS to MariaDB 10 using multisource replication. The installation was successful I can see from show slave 'connection' status\G that binary logs and their positions are changing, but If I add a new data on a master it is not replicated to the slave. Also tried to replicate from Percona Server 5.6 to MariaDB 10 using the multisource replication, but got errors like: Error in Log_event::read_log_event(): 'Found invalid event in binary log', data_len: 44, event_type: 33 Error reading relay log event: slave SQL thread aborted because of I/O error My slave is MariaDB 10 with data and logs partitions encrypted by LUKS and with SSL enabled. As far as RDS and Percona servers don't use SSL I am using non SSL user for the replication. Any recommendations or suggestions would be extremely appreciated. -- Vadim

2 1

[Maria-discuss] New Question: reading dbase records marked for deletion
by AskMonty KB 16 Feb '15

16 Feb '15

Hello, A new question has been asked in "Data Manipulation" by williamblocher. Please answer it at http://mariadb.com/kb/en/reading-dbase-records-marked-for-deletion/ as the person asking the question may not be subscribed to the mailing list. -------------------------------- Can you tell me if dbase records marked for deletion are treated as orinary records or ignored? Thank You Bill Blocher -------------------------------- To view or answer this question please visit: http://mariadb.com/kb/en/reading-dbase-records-marked-for-deletion/

1 0

[Maria-discuss] New Question: Maria 10.0.15 high CPU/disk usage without queries
by AskMonty KB 16 Feb '15

16 Feb '15

Hello, A new question has been asked in "MariaDB Community" by artemkuchin. Please answer it at http://mariadb.com/kb/en/maria-10015-high-cpudisk-usage-without-queries/ as the person asking the question may not be subscribed to the mailing list. -------------------------------- I just migrated the server from mysql 5.5 to mariadb 10.0.15 It is a web hosting server with about 100 sites on different platforms. So, there are about 100 databases. Most tables myisam, some innodb (mostly joomla 3) Before with mysql CPU load was less than 10%, disk uses by mysqld was also not very high. No, with top and top -mio i see pretty high numbers. mysqld takes from 20 to 30% cpu and 80-100% disk About configuration. It is xeon 4 cores+HT (8 cores seen by OS) with 32GB ram, 2 3TB SATA disks in software mirror. os is FreeBSD 10.1 stable I tried using thread of pools, one thread per connection - all the same The weird think what i show processlist i ass about 20 haning connection without any question. This is is normal, some sites leave connections for a while (mod_perl, php) the weirdness is that thare no queies while 20-30% load of CPU. What is mariddb doing? my.cnf [mysqld] #thread-handling= one-thread-per-connection thread_handling=pool-of-threads thread_pool_size=4 thread_cache_size = 45 init-connect='SET NAMES cp1251' default-storage-engine = MyISAM port= 3306 socket= /hoster/tmp/mysql.sock skip-external-locking skip-networking slow_query_log=1 slow_query_log_file=/hosts/MYSQLDB/slow.log # GLOBAL max_connections=800 max_allowed_packet = 50M key_buffer_size = 512M table_open_cache = 32768 query_cache_size = 256M query_cache_limit = 32M # PER THREAD read_buffer_size = 2M # PER REQUEST sort_buffer_size = 32M read_rnd_buffer_size = 1M myisam_sort_buffer_size = 128M join_buffer_size=2M server-id = 1 innodb_data_home_dir = /hosts/MYSQLDB innodb_data_file_path = ibdata1:2000M;ibdata2:10M:autoextend innodb_log_group_home_dir = /hosts/MYSQLDB innodb_buffer_pool_size = 384M innodb_log_file_size = 100M innodb_log_buffer_size = 8M innodb_flush_log_at_trx_commit = 3 innodb_lock_wait_timeout = 50 -------------------------------- To view or answer this question please visit: http://mariadb.com/kb/en/maria-10015-high-cpudisk-usage-without-queries/

1 0

Re: [Maria-discuss] Idea of monitoring tool of Mariadb
by Seung Hoon Yoo 12 Feb '15

12 Feb '15

Thanks reply Lefranc. I think 2 git monitoring solution u mention that would be using SQL execute and caching i think ( i guess, my short program knowledge) MY idea is not "SQL access" , but " GATHER info IN Memory Structure on mysqld proces" example select * from performance_schem => cache => GUI/HTTP ( SQL Method) => i think bad pfs structure (sql_parse.cc, pfs.cc) => memcopy... ( => I WANT METHOD) => i think good. 2 soultion u mentioned is use Direct Access Memory or SQL execute ?? Thanks!! and sorry My short English grammer. 2015-02-12 20:48 GMT+09:00 Guillaume Lefranc <guillaume(a)adishatz.net>: > Hello Seung Hoon Yoo, > > there are already some (experimental) tools which do that: > > https://github.com/svaroqui/http_monitor is a plugin for MariaDB 10 which > collects information based on information and performance schema > https://github.com/sjmudd/pstop is a performance monitor top-like you can > run as a command line UI > > Best regards, > > > 2015-02-12 12:36 GMT+01:00 Seung Hoon Yoo <sehyoo96(a)gmail.com>: > >> Hello, >> >> i think about monitoring gui tool ( sure, monyog Nagius support mariadb) >> Mysql Enterprise Monitor is cool, but commercial ver and that is oracle >> haha >> >> Anyway my idea first Performance schema info, >> that is store info in memory structure in source, >> so if during every 1 sec memcopy from performance schema to user-define >> structure >> and that information transfer out of mariadb by udp port etc.. >> >> So, until world in mysql / maria, their tool using proxy method or sql >> script >> thas will be some performance issue happen. >> >> so Direct Mariadb memory Access Method is useful.. >> >> >> So.. mariadb.com has any plan GUI Monitoring tool? >> and how about my idea about gathering performance schema ( + >> information_schema etc) >> info in mysqld thread info( memory structure) ?? >> >> >> Thanks for reading! >> >> p.s i am not used to english. ;) >> >> LG CNS post architecture Team >> >> _______________________________________________ >> Mailing list: https://launchpad.net/~maria-discuss >> Post to : maria-discuss(a)lists.launchpad.net >> Unsubscribe : https://launchpad.net/~maria-discuss >> More help : https://help.launchpad.net/ListHelp >> >> >

2 1