From: Andrew Morgan Date: January 7 2013 2:56pm Subject: MySQL Cluster 7.2.10 has been released List-Archive: http://lists.mysql.com/announce/816 Message-Id: <82ce9232-d82e-4a82-b90a-56ca31f8ff36@default> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: quoted-printable Dear MySQL users, MySQL Cluster 7.2.10, has been released and can be downloaded from http://w= ww.mysql.com/downloads/cluster/ - where you will also find "Quick Start" gu= ides to help you get your first MySQL Cluster database up and running. To c= onfigure more sophisticated Clusters, try out the prototype, browser-based = MySQL Cluster Auto-Installer from http://labs.mysql.com/=20 The release notes are available from http://dev.mysql.com/doc/refman/5.5/en= /mysql-cluster-news-5-5-29-ndb-7-2-10.html=20 MySQL Cluster enables users to meet the database challenges of next generat= ion web, cloud, and communications services with uncompromising scalability= , uptime and agility. An introductory demo video can be viewed at http://ww= w.oracle.com/pls/ebn/swf_viewer.load?p_shows_id=3D11464419 and more details= can be found at http://www.mysql.com/products/cluster/=20 Best Regards, Andrew Morgan. MySQL High Availability Principal Product Manager andrew.morgan@stripped @andrewmorgan www.clusterdb.com =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D Functionality Added or Changed ------------------------------ Added several new columns to the transporters table and counters for the co= unters table of the ndbinfo information database. The information provided = may help in troubleshooting of transport overloads and problems with send b= uffer memory allocation. For more information, see the descriptions of thes= e tables. (Bug #15935206) To provide information which can help in assessing the current state of arb= itration in a MySQL Cluster as well as in diagnosing and correcting arbitra= tion problems, 3 new tables-membership, arbitrator_validity_detail, and arb= itrator_validity_summary-have been added to the ndbinfo information databas= e. (Bug #13336549) Bugs Fixed ---------- The multithreaded job scheduler could be suspended prematurely when there w= ere insufficient free job buffers to allow the threads to continue. The gen= eral rule in the job thread is that any queued messages should be sent befo= re the thread is allowed to suspend itself, which guarantees that no other = threads or API clients are kept waiting for operations which have already c= ompleted. However, the number of messages in the queue was specified incorr= ectly, leading to increased latency in delivering signals, sluggish respons= e, or otherwise suboptimal performance. (Bug #15908684) The management client command ALL REPORT BackupStatus failed with an error = when used with data nodes having multiple LQH worker threads (ndbmtd data n= odes). The issue did not effect the node_id REPORT BackupStatus form of thi= s command. (Bug #15908907) The setting for the DefaultOperationRedoProblemAction API node configuratio= n parameter was ignored, and the default value used instead. (Bug #15855588= ) Node failure during the dropping of a table could lead to the node hanging = when attempting to restart. (Bug #14787522) The recently added LCP fragment scan watchdog occasionally reported problem= s with LCP fragment scans having very high table id, fragment id, and row c= ount values. This was due to the watchdog not accounting for the time spent draining the= backup buffer used to buffer rows before writing to the fragment checkpoin= t file. Now, in the final stage of an LCP fragment scan, the watchdog switches from= monitoring rows scanned to monitoring the buffer size in bytes. The buffer= size should decrease as data is written to the file, after which the file = should be promptly closed. (Bug #14680057) Job buffers act as the internal queues for work requests (signals) between = block threads in ndbmtd and could be exhausted if too many signals are sent= to a block thread. Performing pushed joins in the DBSPJ kernel block can execute multiple bran= ches of the query tree in parallel, which means that the number of signals = being sent can increase as more branches are executed. If DBSPJ execution c= annot be completed before the job buffers are filled, the data node can fai= l. This problem could be identified by by multiple instances of the message sl= eeploop 10!! in the cluster out log, possibly followed by job buffer full. = If the job buffers overflowed more gradually, there could also be failures = due to error 1205 (Lock wait timeout exceeded), shutdowns initiated by the = watchdog timer, or other timeout related errors. These were due to the slow= down caused by the 'sleeploop'. Normally up to a 1:4 fanout ratio between consumed and produced signals is = permitted. However, since there can be a potentially unlimited number of ro= ws returned from the scan (and multiple scans of this type executing in par= allel), any ratio greater 1:1 in such cases makes it possible to overflow t= he job buffers. The fix for this issue defers any lookup child which otherwise would have b= een executed in parallel with another is deferred, to resume when its paral= lel child completes one of its own requests. This restricts the fanout rati= o for bushy scan-lookup joins to 1:1. (Bug #14709490) References: See also Bug #14648712. Under certain rare circumstances, MySQL Cluster data nodes could crash in c= onjunction with a configuration change on the data nodes from a single-thre= aded to a multi-threaded transaction coordinator (using the ThreadConfig co= nfiguration parameter for ndbmtd). The problem occurred when a mysqld that = had been started prior to the change was shut down following the rolling re= start of the data nodes required to effect the configuration change. (Bug #= 14609774) Cluster Replication: Setting slave_allow_batching had no effect. (Bug #1595= 3730) =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D