Changes in v1.4.2

Patch Release: Oct 30^th, 2017

• Build:
  • clustercontrol-controller-1.4.2-2189
• Controller:
  • MySQL based cluster: if the ‘mysql’ database was explicitly backed up, then it was restored in the wrong way causing permission denied and the restore to fail.
  • Galera: codership repository fixes.
  • Debian Jessie (Debian 9) support.

Patch Release: Oct 25^th, 2017

• Build:
  • clustercontrol-1.4.2-3958
  • clustercontrol-controller-1.4.2-2179
• Controller:
  • Only collect the relevant log files from each host.
  • Accounts daemon fix, to prevent doing any operations on accounts-daemon if running the environment as root or if it is not started.
  • Group-replication bugfixes.
  • Galera: Add replication slave: Properly detect if a replication slave is actually connected to the master.
  • Error-reporter: include node type(s) in the host directory names.
  • CmonDB ‘alarm’ table UTF-8 changes
  • HAProxy config check
• UI:
  • Resend alarm emails.
  • Removed banner from the Add Existing Slave, making it hard to understand what would happen.
  • Set default value as 1 by default for Compression Level for mysqldump.
  • Galera: Overview Page, Flow Control Paused now shows floating points value.
  • Host statistics graphs, the issue with multicore CPU graphing.
  • More verbosity when capturing LDAP logs.
  • Configuration Management: Applied the byte conversion mechanism for the mysql change parameter dialog.
  • Fixed the save settings for property ‘History’ and removed property SSH Options.
  • ProxySQL: Query Rules, added IN () format to match pattern generation.
  • Query Monitor: Adding query outliers explanation in the Overview page.
  • Query Monitor: Renamed Query Histogram to Query Outliers to match what it actually is.

Patch Release: Oct 3^rd, 2017

• Build:
  • clustercontrol-controller-1.4.2-2161
  • clustercontrol-notifications-1.4.2-62
  • clustercontrol-ssh-1.4.2-32
• Controller:
  • Backups: Always execute commands on the controller, only use the seen address (from node’s POV) for constructing the netcat sender command line.
  • s9s_error_reporter: Updates for better compatibility with all s9s cli version.
  • s9s_error_reporter: Prevent error reporting from being blocked by other jobs.
  • Deployment failure on MariaDB 10.2 and 10.1 for Galera Custer – mariadb-compat does not exist on Debian.
  • mysqldump: Handling the backup compression level (bugfix).
  • Galera (all vendors): mysql_upgrade must only run if monitored_mysql_root_password is set. The upgrade will failed if not possible to connect.
  • Galera: Fix advisor to handle wsrep_cluster_address arguments.
• UI:
  • System V Init – Prevent/disable the ‘cmon-events’ process to start (by cron or manually) when <webroot>/clustercontrol/bootstrap.php has set define('CMON_EVENTS_ENABLED', false);.
  • System V Init – Prevent/disable the ‘cmon-ssh’ process to start (by cron or manually) when <webroot>/clustercontrol/bootstrap.php has set define('SSH_ENABLED', false);.

Patch Release: Sept 11^th, 2017

• Build:
  • clustercontrol-1.4.2-3699
  • clustercontrol-controller-1.4.2-2091
• UI:
  • Non-default cluster-specific SSH port support for host validation when adding a new or an existing node.
  • Show all valid nodes for Rebuild Replication Slave and Change Replication Master. All node with binary logging enabled is a valid option.
  • Minor filtering fixes to Manage → Schemas and Users.
  • Removed controller host from PostgreSQL’s query monitor.
  • Minor performance optimization. Removed redundant repeated timezone call.
• Controller:
  • Use cluster-specific SSH settings for host validation when adding a new or an existing node.
  • New error report tarball naming convention – error-report-TIMESTAMP-clusterCID.tar.gz.
  • Include backup records and backup schedules in the error reports.
  • Minor fix to backup scheduling when using advanced cron format.

Patch Release: August 25^th, 2017

• Build:
  • clustercontrol-controller-1.4.2-2063
• Controller:
  • HAProxy: A problem with hidden properties made it impossible to view HAProxy details in the UI unless the stats admin user and password was not admin/admin.
  • Alarms: Possibility to disable the SwapV2 alarms (set swap_inout_period=0 in cmon_X.cnf).

Patch Release: August 24^th, 2017

• Build:
  • clustercontrol-1.4.2-3629
• UI:
  • Configuration Management: Correctly exclude non DB nodes from dropdowns.

Patch Release: August 22^nd, 2017

• Build:
  • clustercontrol-1.4.2-3607
  • custercontrol-controller-2058
• UI:
  • Group Replication: SUDO password not set in the job.
  • MySQL (all variants): Password validation updated to support more characters.
  • MySQL (all variants): Import existing MySQL cluster fails if the specified user is other than root.
• Controller:
  • PostgreSQL: A problem restoring a backup on the specified node (by job: server_address, UI sends master/writable) is fixed.
  • Error reporting: Important error reporter fix to be more tolerant of empty/invalid filenames.
  • Replication: Cluster state was not set if node/cluster recovery was disabled.

Patch Release: August 14^th, 2017

• Build
  • clustercontrol-1.4.2-3574
  • clustercontrol-controller-1.4.2-2045
• UI:
  • Group Replication: Create Cluster job did not submit the sudo password if set.
  • Galera: Restore backup host dropdown was empty unless the Galera node had log_bin enabled.
  • Postgres: small UI fix to remove empty columns.
• Controller:
  • MySQL(all variants)/PostgreSQL: use socat for streaming when it is available.
  • MySQL (all variants): Super read-only causing create a database to fail during restore.
  • MySQL (all variants): Backup, failed to read included config files from my.cnf (!includedir), if the included config dir was empty.
  • Error reporter: drop -W option from netstat (not supported by rhel/centos 6.x).
  • Error reporter: Add missing dependencies for error-reporter (tar/gzip) for minimal distros (eg.: containers.
  • MongoDb: Backup creation fix (for case when ssh user is not allowed to ssh to the controller itself).
  • ProxySQL: Installing an improved galera checker script for new ProxySQL installations.
  • ProxySQL: A fix to auto-restart a failed ProxySQL node.
  • Docker: Small fix to support HAProxy with Docker.
  • Docker: Do not set ulimit inside a container (as this makes some operation failing inside docker).
  • Query Monitor: Doesn’t collect queries with mysql local override and PS=off.
  • Replication: do not recover a user shutdown node.

Patch Release: August 1^st, 2017

• Build:
  • clustercontrol-1.4.2-3538
• UI:
  • Fix password reset script for php v7.
  • Fix LDAP regression with Active Directory and samba account.

Patch Release: July 31^st, 2017

• Build:
  • clustercontrol-1.4.2-3531
• UI:
  • Fix host filtering for Query Monitor.
  • Fix LDAP login regression.
  • Fix to show all databases for Group Replication backups.

Patch Release: July 27^th, 2017

• Build:
  • clustercontrol-ssh-1.4.2-26
• UI:
  • Fix not fatal duplicated symlink error creation at post-installation.

Patch Release: July 24^th, 2017

• Build:
  • clustercontrol-1.4.2-3505
  • clustercontrol-notifications-1.4.2-57
  • clustercontrol-ssh-1.4.2-25
  • clustercontrol-controller-1.4.2-2013
• Controller:
  • ProxySQL log rotate: ProxySQL logs can grow big very fast.
  • PostgreSQL: Improved master failure handling to prevent an old master from being accidentally restarted.
  • Galera/Replication: Adding a node did not update the load balancer HAProxy correctly. Xinetd was not started.
  • Minor fixes to printouts in cmon log file.
• UI:
  • Add support to disable automatic node discovery at import time for the Galera cluster. Manually add IPs/hostnames.
  • Add support to filter by host for PostgreSQL’s Query Monitoring.
  • Fix a race condition for ProxySQL graphs that would eventually consume all memory and crash the browser.
  • Fix escapes in match patterns for ProxySQL.
  • Remove execution flag for systemd service files for cmon-events and cmon-ssh.

  Patch Release: July 11^th, 2017

• Build:
  • clustercontrol-3465
• UI:
  • Fix master selection dropdown for add node. No longer shows non-master nodes.
  • Fix transient node switching glitch in the nodes page.
  • Fix regression of minimum 2 SQL nodes at deployment (MySQL/NDB). No longer required.
  • Fix node selection dropdown when restoring a mysqldump. Only masters allowed.
  • Add standalone option when importing a MySQL Replication cluster.
  • Remove ProxySQL load balancer option with MySQL/NDB Cluster. Currently not supported.
  • Fix activity viewer next/prev causing the page to scroll.
  • Fix missing sudo password if it was set when verifying/checking a host with deployment/add nodes.

Patch Release: July 4^th, 2017

• Build:
  • clustercontrol-controller-1981
• Controller:
  • Fixed a cmon grant error (for root and cmon passwords like ”!password$$”).
  • Skip .sst from db_growth calculation.
  • Restore mysqldump bugfix (for strange passwords)
  • Properly escape cmon password.
  • Don’t do smartctl on /dev/mapper devices at all.
  • MySQL:
    • Deployment (MySQL5.7 templates): added ignore-db-dir=lost+found.
    • Backup: Improved password handling of backup user.
    • Backup: Add compression level for backups.
  • ProxySQL:
    • Can’t remove node when the node is unreachable.
  • PostgreSQL:
    • Fix a minor systemd override file access rights issue.
    • Put the slave to failed state when replication is known to be broken.
    • Fix a minor systemd override.conf file access rights issue.
    • An important bugfix for failover (the solution for the nodes stuck in ‘startup’ replication state).
  • MySQL Replication:
    • deeper external checks when there is a master failure. Try to connect from the slaves to the master using the mysql client to determine if the slave can see the master or get a 2003/2013 error.
  • Galera:
    • Rolling-restart could fail due to an old value of the node’s cluster size. Collect the wsrep variables before checking the cluster size and this is now done in a time-controlled loop.

Initial Release: June 21^st, 2017

• Build:
  • clustercontrol-1.4.2-3421
  • clustercontrol-controller-1969
  • clustercontrol-cmonapi-279
  • clustercontrol-notifications-14.2-53
  • clustercontrol-ssh-1.4.2-21
• ProxySQL:
  • Copy, Export and Import ProxySQL configurations to/from other instances to make them in sync.
  • Add Existing standalone ProxySQL instance.
  • Add Existing Keepalived in active/passive setups with ProxySQL.
  • Support for 3 ProxySQL instances with a Keepalived active/passive setup.
  • Simplified Query Cache creation.
  • Query hits column
• Backup:
  • Verify/Restore a mysqldump on standalone host that is not part of your clusters.
  • Verify/Restore an xtrabackup on standalone host that is not part of your clusters.
  • Customize your backup schedule by using the cron format.
• Notifications (clustercontrol-notifications):
  • Send Alarms and Events to PagerDuty, VictorOps, OpsGenie, Slack, Telegram or user registered Webhooks.
• Web SSH Console (clustercontrol-ssh):
  • Open a terminal window to any cluster nodes.
  • Only supported with Apache 2.4+.
• PostgreSQL:
  • New Master-Slave(s) cluster deployment wizard (streaming replication).
  • Automated failover and slave to master promotion.
  • Rebuild slave.
• Misc:
  • Fixed TLS connection issues for e-mail sending (SMTP).
  • Improved configuration handling of include/includeDir directives.
  • Database user management RPC API for the s9s command-line client.
  • Continuous improvements to UX/UI.
  • New cmon-events process to handle notifications to 3rd party services.
  • New cmon-ssh process to handle Web SSH console access.
  • Improved error reporting for troubleshooting/support.
  • Use a custom mysql port when adding a MySQL Asynchronous slave (MySQL Galera).