The changelog for key MemSQL Ops releases is listed here. To verify and download the latest GA version of Ops, see http://versions.singlestore.com/memsql-ops/latest-v6.

2020-09-01 Version 7.0.7

• Fixes a bug where updating to MemSQL 7 would not work if there were stopped DR databases on the cluster

2020-04-20 Version 7.0.6

• Fixes a bug where an internal upgrade flag would prevent the upgrade process from skipping pre-upgrade checks when instructed

2020-04-20 Version 7.0.5

• Fixes a bug where Ops would incorrectly estimate the required disk space before upgrading from MemSQL 6 to 7
• Increases the timeout for stopping the cluster when upgrading

2020-04-15 Version 7.0.4

• Fixes a bug where Ops would crash when adding a memory capacity license

2020-04-02 Version 7.0.3

• Fixes a bug where getting the value for transaction_buffer on the Master Aggregator would fail if the Master Aggregator and the Ops Primary Agent were not on the same host

2020-03-06 Version 7.0.2

• Fixes a bug where getting the MemSQL version failed when upgrading a single node
• Fixes a bug where licenses would be incorrectly detected as non-enterprise

2020-01-17 Version 7.0.1

• Adds an additional pre-upgrade safety check for sufficient disk space
• Adds automatic snapshotting post-upgrade
• Fixes upgrade checks failing if a user or application is dropping tables while the upgrade checks are running
• Fixes mismatching license format between MemSQL Ops and MemSQL

2019-12-03 Version 7.0.0

• Supports upgrading to MemSQL 7.0 clusters

2019-10-08 Version 6.8.0

• Adds Ops to SingleStore Tools migration tool
• Fixes a bug where Ops would write sync variables to memsql.cnf

2019-08-16 Version 6.7.8

• Fixes a bug where Ops would not be able to deploy MemSQL 6.0
• Fixes a bug that applies to MemSQL versions after 6.5 where Ops would wait for leaves to finish recovering before starting aggregators. After 6.5, Ops no longer needs to wait for the leaves to finish recovering.

2019-08-05 Version 6.7.7

• Adds a more descriptive error message that agent-uninstall returns when it cannot uninstall the agent because there are still SingleStore nodes managed by the agent. Now, the error message includes the agent ID and the MemSQL IDs of the nodes that are blocking the uninstall.
• Fixes a bug where MemSQL Ops would read the wrong license key variable for new license formats
• Fixes a bug with unit-based license capacity formatting

2019-07-09 Version 6.7.6

• Add support for new license format
• Add SingleStore Forums link to the sidebar

2019-05-13 Version 6.7.5

• Fix a bug where generating a cluster report would crash against MemSQL 6.7.11 or higher

2019-03-04 Version 6.7.4

• Fix memory usage bug due to recent psutil upgrade

2019-02-26 Version 6.7.3

• Correctly setup the temporary directory to use as configured in settings.conf
• Upgrade psutil to 5.5.1 to fix issues on newer Linux distributions

2019-01-24 Version 6.7.2

• Deprecated memsql-top from MemSQL Ops
• Now, the output of EXPLAIN CLEAR ORPHAN DATABASES appears in the cluster report
• Fixed a bug where the sql_mode engine variable, when used globally, could break some queries that MemSQL Ops runs

2018-11-16 Version 6.7.1

• Fixed a bug in update-config that could prevent a SingleStore node from being auto-restarted

2018-11-06 Version 6.7.0

• Initial release of MemSQL Ops 6.7

2018-09-13 Version 6.5.11

• Setup system limits during _upgrade
• Fixed bug that could prevent a primary agent from stopping cleanly

2018-09-11 Version 6.5.10

• Fixed bug that could result in half-connected follower
• Fixed bug that could result in a double-start during upgrade
• Fail to start Ops if port is taken
• Improved backtick support in cluster reports

2018-08-15 Version 6.5.9

• Run sqlite in a read-optimized mode
• Numerous performance improvements through reducing the number of queries executed by agents
• Better support for disabling analytics entirely in large clusters
• Slow down the update interval for node transitions
• Upgrade can skip starting the cluster after upgrading the filesystem
• Cluster page in the UI only shows a list view if there are more than 60 hosts in the cluster
• Fixed double-start bug in Jobs engine
• Numerous small bug fixes

2018-07-30 Version 6.5.8

• Wait for all leaves to fully recover before proceeding with starting aggregators

2018-07-17 Version 6.5.7

• Adds functionality to support MemSQL 6.5
  • memsql-ops memsql-update-root-password no longer needs to remove/add leaf
• Added more information to cluster report
  • Output of SHOW VARIABLES LIKE '%'
  • Output of SHOW RESOURCE POOLS
  • Output of SHOW WORKLOAD MANAGEMENT STATUS
  • Output of SELECT * FROM information_schema.MV_EVENTS
• Fixed bug in the report command where a bad connection on a SQL query would affect all subsequent queries
• Fixed bug in the report command where failed agent reports in individual agents would cause invalid cluster reports

2018-02-14 Version 6.0.10

• Fixed bug in the add aggregator code that can prevent upgrade from finishing

2018-02-05 Version 6.0.9

• Improved portability of MemSQL Ops on certain systems

2018-01-17 Version 6.0.8

• Ensure that connections from Ops to MemSQL use OpenSSL rather than YaSSL

2017-11-09 Version 6.0.7

• Change disk usage check for the memsql-upgrade command to be on the MemSQL Ops install directory, rather than on the SingleStore DB data mounts
• Move disk usage check to an earlier stage of memsql-upgrade
• Update pipelines API to support 64 bit ints
• Disable sync variable check on memsql-config-update for MemSQL 6.0.5 and older
• Add a tmpdir setting to Ops config

2017-10-18 Version 6.0.6

• Support for MemSQL 6
• Remove support for non-colocated nodes
• Improvements to the job-list command
• Added sysctl -a and Transparent Huge Pages information to Cluster Report
• Upgraded Paramiko version to support hmac-sha2-256
• Renamed MemSQL installation directories to include the install’s ID
• Removed Spark Streamliner, Data Import and Speed Test

2018-02-14 Version 5.8.4

• The agent-upgrade command will check for the latest version of MemSQL Ops over the Internet if the user supplies a version or tarball path

2018-01-19 Version 5.8.3

• Ensure that connections from Ops to MemSQL use OpenSSL rather than YaSSL

2017-08-15 Version 5.8.2

• Bumped bundled MySQL client from version 5.5 to 5.7
• The memsql-upgrade command supports a flag to set a timeout for stopping the cluster
• The agent-upgrade command supports a flag to set a timeout for upgrading each agent
• Allow memsql-upgrade –memsql-id flag to accept prefixes of MemSQL IDs
• Add memsql.cnf to agent report
• Use new _BT in agent report when available
• Fix bug in agent report swapping database name and memsql id in queries
• Stop pipelines engine from running pipelines status queries if there are no pipelines running
• Change job system to reduce the volume of each agent’s API requests
• Allow user to disable analytics

2017-06-07 Version 5.8.1

• A timeout argument has been added to the MEMSQL-STOP command.

2017-04-16 Version 5.8.0

• Ops now supports custom tokens for inter-agent authentication. See the following topics for more information:
• API-TOKEN
• AGENT-UPDATE-API-TOKEN
• FOLLOW

2017-01-26 Version 5.7.1

• Fixed issues that could occur when creating a columnstore-based pipeline.
• Metrics on the Status page are now presented in International System of Units (SI) format.

2017-01-24 Version 5.7.0

• If Ops doesn’t have internet connectivity, it will not check to see if upgrades are available
• Adds support for upgrading individual SingleStore nodes using the memsql-upgrade command
• No longer runs _BT during cluster reports by default
• Adds support for all SKIP ... ERRORS modes in the Pipelines interface. See LOAD DATA for more information.
• Ops will not take a network lock when running a remote API call
• Now sets TCP Keepalive on sockets to allow the primary to handle disconnections sooner
• Adds the output of mount to cluster reports
• Ops will not attempt to fetch more pipeline errors than its limit
• Enforces selecting an error handling mode when a user creates a pipeline
• Improved error messages in jobs and intentions

2016-11-15 Version 5.5.3

• Fixes a bug where the cluster engine was incorrectly queueing start jobs when in the Unreachable state
• Ops is now more careful about starting and stopping orphaned Spark components
• The Agent Report now handles DNS resolution failures more cleanly
• Fixes a bug where Pipelines’ In Progress batches were being incorrectly cached by Ops, which means they would not disappear from the interface
• Ops now prioritizes High Availability notifications over Speed Test notifications
• The Agent Report now also includes a stack dump of all the threads running in Ops
• Fixes a bug where analytics data was being counted twice
• Fixes a bug where Ops could be started during an agent upgrade
• Fixes a bug where Pipelines metrics were disappearing when navigating around Ops pages
• The status page now features spinners while loading data and warns the user if an online cluster cannot be found
• Fixes a bug where clicking the back button in the Deploy MemSQL modal would go two steps back instead of one
• Fixes a bug where Ops would slow down when too many jobs had been queued (including finished jobs)
• Ops now closes open MemSQL connections during memsql-ops stop
• Fixes a bug where the Schema page did not expect certain values to be NULL

2016-10-19 Version 5.5.2

• Ops will now ensure all the nodes in a cluster are fully offline before upgrading
• Adds a new MemSQL State called UNREACHABLE which indicates that Ops cannot connect to the SingleStore DB instance
• Adds a new arg of --limit to job-list to limit the number of jobs listed
• When attempting to upgrade an offline cluster, Ops will prevent you from upgrading unless --skip-snapshot is enabled
• Adds information_schema.processlist to cluster report
• Fixes help strings in memsql-unmonitor to make them clearer
• Fixes error strings in memsql-upgrade to make them more explanatory
• Fixes a bug where port numbers were being incorrectly suggested when adding new nodes
• Fixes a bug where SSH credentials were being stored with the wrong user
• Fixes a bug where the errors table was incorrectly using the wrong time format
• Fixes a bug where byte usage for a table was being incorrectly displayed when High Availability is enabled

2016-10-03 Version 5.5.1

• Improved debuggability of Agent upgrade
• Fixed bug preventing the deprecated cluster-upgrade command from working
• Warns the user if we detect a maximum_memory imbalance between leaf nodes in the cluster
• Fixed bug with setting up license via support modal

2016-09-26 Version 5.5.0

• Adds UI Support for new Pipelines feature
• Changes to our storage/sqlite layer to improve performance and allow Ops to scale to larger clusters
• memsql-upgrade no longer restarts on rollback by default
• Changed “Swap” bar to “Paging”
• agent-ssh uses a UNIX socket to interact with the secure credentials store
• Warns the user if we detect a maximum_memory imbalance between leaf nodes in the cluster
• Ensures that HA failover doesn’t needlessly happen during cluster stop/start/restart
• Treats the local disk being out of inodes as an out of disk error
• Schema explorer suggests running ANALYZE TABLE if it detects the table has changed more than 30% since the last time the table was analyzed.
• Warns the user if we detect multiple versions of SingleStore DB running in the cluster
• Added –force option to memsql-update-root-password to allow the user to inform Ops of a password change that happened outside of Ops control
• memsql-optimize now tells the user what it is going to do before doing it
• Adds a profiler to the Ops runtime to assist with debugging of production systems without a custom Ops build
• The user can now provide passwords via ENVIRONMENT variables in order to prevent them from showing up process lists
• Increases default report size to 150MB
• Enabled –hard-stop when stopping the entire cluster
• Security: updated version of OpenSSL
• Adds a NO_SPARK variable to the Ops settings, if this is set on an Agent, Ops will not deploy any Spark processes to that agent
• Ops queries will no longer show up in the MemSQL query log
• When deleting MemSQL leaves ensure that REMOVE LEAF is always run
• Fixed Memory usage bug relating to memory values being displayed incorrectly

2016-06-14 Version 5.1.0

• New iops monitoring bar in Ops cards
• Better HA support
• Bug and stability fixes
• syslog support in Ops

2016-04-29 Version 5.0.3

• Bug: Fixed issues with checkpointing database in Streamliner

2016-03-30 Version 5.0.2

• Feature: The new Schema Explorer allows users to explore the hierarchy of databases and tables in SingleStore DB. This tool can be used to help capacity plan or provide insight on data distribution.
• Feature: The new Query Explorer provides access to under-performing queries along with aggregated metrics over the historical record.
• Feature: Ops now includes a visual way to import your data into MemSQL from S3, HDFS and MySQL.
• Feature: Ops now includes a feedback form to enable issues to be easily reported.
• Stability: MemSQL Ops 5.0.0 (BETA) includes numerous performance, stability and bug fixes.