This page lists the major changes between releases.

The complete change list can be found at: https://secure.opsview.com/wsvn/wsvn/opsview/trunk/CHANGES.

• This is the final Opsview Community release. Opsview Community will be superseded by Opsview Core
• We recommend that you upgrade your Opsview Community system to 20120424 before upgrading to Opsview Enterprise 4.0 or Opsview Pro

• Added additional indexes for common status and events view queries in Runtime database
• Added logging to nightly database housekeeping to show timing of each step performed
• Web now shows Access Denied if a user tries to view parts of the host group tree where no host groups are allowed based on roles
• Improvement to opsview_rest for linefeeds and help page with –url-prefix
• Display if module is installed and clarify the meaning of enabled (displayed in menu)
• Opsview update plugin now returns WARNING if there is a local system issue reaching the downloads site
• Consider dormant SNMP interfaces as UP (thanks to Robert von Bismarck)
• Re-synchronised Runtime database schema by removing unnecessary indexes and fixing column definitions
• Added more data in /rest/status/host, to keep consistent with /rest/status/service
• Log opsview_cronjobs output to file, to avoid unnecessary emails
• Renamed RANCID module to Netaudit in modules table
• clone_host tool has new –delete option, to specify if it should delete hosts as well
• Housekeeping script can now reads parameters file if specified. Combined 2 other housekeeping scripts.

• Fixed error where perl compatibility links on slave systems were removed after a send2slave -p (plugins) or send2slaves -s (SNMP MIBs)
• Fixed Solaris dependency due to package rename to opsview-perl
• Fixed snmptrap rule changes via web interface so now it lights up the “configuration changes” indicator
• Fixed debian packaging so it correctly calls the common preremove scripts
• Removed unnecessary calls to /etc/rc.d/init.d/functions
• Fixed Nagvis error messages that may get displayed. Removed Nagvis favicon functionality
• Fixed import_slaveresultsd not starting up correctly on slaves
• Fixed possible script injections (thanks to J Oquendo@e-fensive.net)
• Fixed SNMP cascade checks error with interfaces with duplicated ifDescr
• Fixed double acknowledgements being sent out when using “Mass Ack's” in the UI
• Fixed JDBC plugins, -p was specified for port and pool name
• Fixed slaves logging passive checks when not configured to

• Version numbering for Opsview Community is now based on the release date
• Opsview Data Warehouse Graphing feature has been removed as it has been superceded
• check_http_authentication option change: -T now specifies text to search for as it previously clashed with -t for timeout
• Due to perl library location changes, any custom perl modules added into /usr/local/nagios/perl will be deleted

• New SNMP interface poller check now uses a single active check to poll all interfaces at once, reducing CPU and memory usage by 70%. Gathers information about unicast, broadcast and multicast packets. Notes:
  • Errors and Discard services no longer go into a failure state if interfaces are not enabled and gathering data
  • Absolute throughput performance data now returned in 'bps' instead of 'b'. This may affect ODW reporting if you filter by this unit of measurement
• Layout improvements and button changes to improve overall user interface design

• Viewport details now include contextual menus if authenticated and keyword has it defined
• Nagios modified to process check results in time order
• Stopping Nagios during processing of check results now exits cleanly
• Nagios now writes active check results and passive checks directly to spool directory, reducing one I/O move operation for each result
• Allow the checkresults directory to be moved more easily
• Allow updating of default interface thresholds via opsview_web.yml file
• Mass rechecks only lists services that are active types
• Allow objects.cache to be moved more easily
• New cascade option for passive service checks, to support single plugin returning many results
• Passive service checks allowed to define number of failures before alerting
• Allow logging of all REST requests for playback
• Added swap thresholds on check_memory_solaris (thanks to Michael Caplin)
• Reinstated 'Record Output Changes' options (stalking) as servicecheck advanced option
• Perl libraries now expected in /opt/opsview/perl. The old location, /usr/local/nagios/perl, will be removed prior to installation. Slave systems may have issues upgrading if you have updated perl modules in /usr/local/nagios/perl
• Small performance improvement to avoid writing Nagios configuration service templates for services with no relevant host attributes
• Start Opsview Agent when sending Opsview to slave
• Rewritten check_dnsbl to use Nagios::Plugins rather than the deprecated utils.pm
• REST API has a new global URL parameter, alwaysReturn200, which will return a structure of {“rest”:DATA,”status”:ACTUALSTATUSCODE}. This is useful if you are using JSONP which does not give the response information if the status is not 200
• REST API addition of /rest/user to get user information. Also /rest/login?include_user_data=1 will return same data
• New accesses for REPORTUSER and REPORTADMIN
• Authentication cookie now includes the list of accesses this user has
• Added tool to clone hosts via REST API

• Fixed ODW's seconds_not_ok_scheduled when overlapping downtimes occur (was only calculating first one)
• Fixed mrtg-rrd.cgi when inputting data by form (thanks to Jo Greene)
• Fixed nlcat dieng on line 121
• Fixed cosmetic issue on Safari when entering downtime values
• Fixed create new notification profile page
• Amend check_snmp_linkstatus MySQL table type from MyISAM to InnoDB for improved performance
• Fixed system preference access for NetAudit module email address on slaves
• Fixed event filtering by date where incorrectly searched on UTC time
• Fixed opsview database backups to include the full db schema to allow restores to work correctly
• Fixed Nagios debug logging
• Fixed master server running host checks when it wasn't meant to be
• Removed requirement on a MIB having to have an enterprise type before being promoted
• Fixed Host Interfaces edit page where max SNMP message size and tidy ifDescr settings were lost when saving on host edit page
• Fixed Host Interfaces status page where different max SNMP message size and tidy ifDescr level used
• Fixed access to info/internal wiki URL's using names, e.g. /info/host/name/<hostname>
• Fixed possible Nagios segfault when deleting downtimes
• Fixed breadcrumbs from graph page to ignore service parameter
• Fixed not being able to specify the timeout correctly on check_http_authentication
• Fixed SNMPv3 privprotocol not being understood correctly by cfmaker in mrtgconfgen.pl
• Fixed error on slave startup script on Debian where mysql was required
• Fixed event view listing events twice when filtering by multiple keywords with duplicated services
• Fixed event REST API if filtered by hostgroup that did not exist
• Fixed opsview-agent startup message
• Fixed sync_cluster_node_status when nothing to sync
• Stop dying when nagios user's profile not set correctly - show a warning instead. Force profile to be sourced in opsview-web
• Fixed gaps in new graphs for checks set with 60 minute interval (gaps occur when actual interval is more than 60 minutes; set interval to 55 minutes to avoid gaps for existing checks and graph data)
• Fixed issue with clustered slaves not taking over checks properly
• Fixed possible problem with Nagios where freshness events are created unnecessarily due to check_options. Workaround is to remove this value from being retained
• Fixed 'percent' UOM graph perfdata to use 'raw' Y units by default instead of using 'm' for sub-second data points
• Fixed calculation of “service unhandled” so now only includes services where host is in a UP state (a host state of DOWN or UNREACHABLE considers the service as handled). “host unhandled” now only applies for hosts in DOWN state
• Fixed error where ordering service states by last state change gave an exception

• Support for Solaris 10 using OpenCSW packages. See http://docs.opsview.com/doku.php?id=opsview-community:solaris-installation

• Allow nlcat to search within x logs y days ago (-n y -e x)
• Increate default Nagvis URL limit from 1900 chars to 3000
• Email notifications now display data about a host's host group hierarchy
• Allow ability to change email template
• check_opsview_slave_cluster now propagate multi line error messages back to master
• Allow service checks with blank check periods to inherit from host
• Added in check_nagios_check_spread for watching how many service checks run in last minute on an opsview server
• Added ODW diagnostic tool for dumping portion of database
• Added ODW tool to fix issues with services with incorrect acknowledged states
• Improved help text on check_aws_ec2 and check_aws_ec2_instances
• Applied patch on check_snmp_linkstatus to reduce the number of 'database locked' messages at busy times (thanks to Luca)
• check_snmp_linkstatus can use a MySQL backend database rather than SQLite - see plugin help for details

• Fixed interactive graphs displaying times in UTC instead of server timezone
• Fixed check_snmp_linkstatus for better handling of devices that do not have high speed interface speed OIDs
• Fixed IE7's host group menu options in the host group hierarchy configuration page. It still sometimes interprets a click on the menu as a drag event, but cannot seem to overcome. This works fine in IE8 and IE9.
• Improved drag of host group hierarchy configuration page where an ajax call is not made if dropped onto the same parent. There is still a flicker occasionally in IE7
• Correct SLES slave RPM prereqs
• Fixed string formatting bug in utils/parse_retention
• Fixed install error seen on RPM packages for opsview-web upgrades
• Fixed host template edit page where all referenced hosts removed
• Fixes for double proxying without forcing X-Forwarded-Hosts changes at Apache level
• Fixes for check_snmp_memutil to use warning and critical limits correctly; swap limits split off from real mmemory limits
• Fixed opsview-agent running as nagios user
• Fixed IE9 graphs not displaying correctly
• Fixed for perl arch-specific library directory not being found correctly
• Fixed contextual menu for hosts where incorrect “Access denied” is shown when using only keywords for access control. WARNING: The URLs for /info/host/{id} are now based on the id of the runtime host object, not the opsview host object
• Fixed security issue where wiki comments were changeable by any authenticated user
• Fixed possible problem with mysql indexes not finding correct objects in runtime database (thanks to levfpt and luca)
• Fixed timeout of 11 seconds (increase to 240) when using 'invert plugin results' on a service check
• Fixed typo in check_opsview_slave
• Fixed Nagios not deleting old check result files

• The BURT based reporting tool has now been removed. This feature was replaced by the Reports Module and has been deprecated since Opsview Community 3.7.0.
• REST API responses now default to presenting time values in unix epoch format. You can get older behaviour by setting a URL parameter of format_datetime=1, which will return back time based on system timezone. This affects /rest/downtime, /rest/status/service
• Commands for notification methods will automatically remove invalid shell characters if used as parameters because they can break notifications
• Important announcement for users running Opsview on Solaris:
  • SPARC architecture is no longer a supported platform for Opsview on the Solaris operating system.
  • Packages for Solaris 10 x86 (32 / 64 bit) will not be available for 3.13.1. This is due to the extensive refactoring work required for continuing support of this platform. We expect to reintroduce support before then end of the 3.13 release cycle.
  • Monitoring of Solaris servers by Opsview in unaffected by these changes.

• REST API extended to include:
  • Performance data for graphing
  • Event data
  • Runtime objects such as host and services and performance metrics
  • Added host and service state type to /status/service REST calls
  • Added host state_type and duration to /status/host REST calls and provided filtering of these states
  • Added server_timezone to /rest/info REST calls
  • Added includeextradetails to /status/host REST calls
• Added wiki support for services, including ability to redirect to third-party wiki. This capability already exists for Hosts and Host Groups.
• Performance improvements to reload and post-reload time
• Microsoft Internet Explorer 9 is officially supported

• Halved the time of reloads on systems with a large number of contacts (> 50)
• Over 75% improvement in time taken for post-reload job in ndoutils_configdumpend
• Updated check_snmp_apcups to include more metrics in performance data
• Updated check_snmp_linkstatus and query_host to support high speed interfaces over 1Gb/s
• Include nagios_acknowledgements in db_runtime dump_status
• 100% performance improvement of object creation in Opsview Web (thanks to C.H.Kang)
• Added more debugging for SNMP communication from Host edit pages
• Amend slicehost check to output WARN not CRIT when 'Service temporarily disabled'
• Removed call to redhat-lsb in opsview-agent startup script as not required
• Allow status.dat to be moved more easily (OPS-1677)
• Added 'CONNECT' method to check_http for checking web proxies (patch by 'Frank4DD')
• Added in a nagios/etc/local.d directory for local Nagios® Core customisations (macro, timeperiods, etc)
• Allow MRTG forks and refresh rate to be specified within opsview.conf (OPS-310)
• Added host and service state type data and host state_duration data to old status XML calls
• Added new metrics to check_nagios_stats plugin (active and passive service checks last 60 mins, MAX and AVG active service check execution time)
• Improved error message for import_runtime when a reload wipes the nagios_programstatus table
• Included validation for notification methods to remove invalid characters
• Added support for avg_execution_time, max_execution_time, num_active_service_results_hour, and num_passive_service_results_hour to /rest/serverinfo REST calls
• Improved slave checking for NSCA failure by running diagnostics automatically for troubleshooting. Warning: This could raise a critical alert the first time, but this should resolve itself for the next run

• Fixed host group hierarchy configuration screen
• Fixed hover over colour when dragging host group onto an empty leaf host group
• Fixed MRTG list page to not refresh
• Fixed SNMP communication when device returned a blank sysDescr value on host edit and host interfaces pages
• Fixed validation of duplicated host attribute values
• Fixed maintenance page file name used by Apache error handler (/usr/local/nagios/etc/maintenance)
• Fixed occasional host attributes ARRAY(0xZZZZZ) error (OPS-1663)
• Fixed '[[: not found' error in set_profile on install/upgrade
• Fixed /var/run/opsview permissions not being reset correctly
• Fixed debian postinst output when checking database access
• Fixed cases where the opsview database name is changed from the default opsview in the DBIx::Class models
• Fixed 'check_opsview_sessions' to count active Opsview sessions correctly again
• Fixed 'notify_by_email' hanging on some systems when run by nagios user
• Updated check_snmp_apcups to fix bad performance data output for battery runtime
• Reduced amount of “Couldn't process packet” messages from NRD in opsviewd.log
• Fixed bug in acknowledgements where sticky and notifications were not respected correctly in UI
• Fixed check_snmp_perfstats to run with SNMP v5.4.3 and newer (thanks to Alan Wijntje)
• Fixed bug in nagvis config generation when it resides in a different filesystem. Added tests to capture nagvis changes
• Fixed security of wiki comments which were not being checked against access controls
• Fixed bug in check_snmp_loadavg plugin which were not accepting load avg parameters
• Fixed bug where a user without CONFIGURESAVE was able to list all roles in the user preference page
• Fixed bug in check_snmp_linkstatus for 'noSuchObject' when checking for high speed interfaces
• Fixed opsview-slave packages (rpm and deb) creating required initial directories
• Fixed generated slave crontab for rsync task from introducing cron scheduling error
• Fixed NDOutils bug with ndo2db where multiple HELLO connections cause memory leak

• Important announcement for users running Opsview on Solaris:
  • SPARC architecture is no longer a supported platform for Opsview on the Solaris operating system.
  • Packages for Solaris 10 x86 (32 / 64 bit) will not be available for 3.13.0. This is due to the extensive refactoring work required for continuing support of this platform.
  • Monitoring of Solaris servers by Opsview in unaffected by these changes.
• The old XML status API no longer has the “highest” state at the host group level nor in the services list summary - this is replaced with “computed_state”
• The downtime REST API no longer uses the hostgroupid, hostid or serviceid URL parameters as a shorthand for hg.hostgroupid, hst.hostid, svc.serviceid when submitting downtime - you must use the full parameter instead
• Support has been introduced for Red Hat Enterprise Linux 6.0 and Debian 6.0 Squeeze

• REST API extended to include:
  • Status information summarised by host group
  • Status information summarised by host
  • Status information by service
  • Status information summarised by viewport
  • Acknowledgements
  • Mass rechecks
• Added check_opsview_services plugin to monitor unhandled services
• Added check_selenium plugin to run and monitor perl test cases generated by the Selenium IDE
• Added inheritance of arguments to host attributes from the default attribute configuration (contributed by Capside)
• Also note that Opsview Mobile for Android is now available

• Configuration generation speed improvement of 60% when using lots of keywords
• Host “resolve service checks” page now includes multiple service checks information (contributed by Capside)
• Changed behaviour of 'check_snmp_interfaces' and 'check_snmp_linkstatus' plugins. SNMP or network errors now result in an UNKNOWN status rather than WARNING or CRITICAL
• Host interface screen now shows throughput in bps (a reload is required for the new data to be graphed)
• Turned down the font size of handled/unhandled labels and fixed size of table
• Added host attributes expansion to host check commands and timed exceptions
• Performance viewport defaults to 1 month
• Allow for nmis rsync between slaves within a cluster to be run at different times
• Allow commas to be used for full name of contacts
• Improve check_snmp_interfaces (Interface Status) to work across more devices
• Hosts added to runtime.opsview_viewports table and columns renamed
• Events view filtering by keyword includes host state information now
• Audit log messages now show where a user has logged in via (web, REST API, RSS, old API)
• Opsview configuration DB backup now excludes temporary data, including current sessions
• REST API allows switching of content type via a URL parameter, so now supports JSONP via cross domain <script> tags
• REST API allows login via authtkt
• /rest/serverinfo includes each monitoringserver's activated value
• Host unreachable is now considered to be handled
• Host group summary and viewport summary now has a computed state which represents the highest host or service state associated
• opsviewd will restart ssh tunnels if the ports cannot be forwarded on the initial connection (Note: will not be used with ssh versions less than 4.6p1) (OPS-1620)
• Amend slave node plugin to check forwarded ports on slaves are open (OPS-1620)
• When upgrade lockfile exists, say so when import_runtime -v used
• Allow create_monitor_pack to parse argument syntax and add any referenced attributes
• Set syslog “Successfully connected to database” messages to respect the ndo2db_debug_level parameter, to reduce amount of syslog messages
• Display helpful information about roles with all keywords, when defining related roles in the keyword edit page
• New switch on nlcat to remove common entries from nagios.log file
• Added in new util: 'copy_host_attributes'
• Allow setting of host check_command to NULL in import excel spreadsheet
• Updated RANCID module name to Netaudit for new installs
• Added easyXDM 2.4.15.118
• Improve how old NMIS reports and rrd files are removed, and reduce time limit from 730 days to 365
• Initial installs now contain multiple service checks for Unix filesystems and Windows drives

• Fixed notification profiles losing a keyword selection when role has all_keywords but role is not specified in keyword's object list
• Fixed bug where saving a host template would reorder the host templates on the host
• Fixed listings for downtimes when more than 50 rows returned
• Downtimes at HH are prefixed by host group name where the request was made, not each individual host group name
• Fixed check_snmp returning graphing data for timeticks type values
• Fixed Test SNMP connection on host edit page
• Fixed performance viewport always returning no items
• Increase rsync_nmis_database bandwidth usage and also use incremental rynsc transfers
• Make use of ionice and renice (if available) when running rsync_nmis_database to reduce load on server
• Fixed events view captions and information box for hosts in timeline
• Fixed viewports so now displays public keywords for logged in users with VIEWSOME access
• Fixed viewports to not show keywords without access in drop down list
• Fixed SNMP polling configuration generation for check_snmp, when doing a string comparison
• Fixed slave checking when NRD not configured on upgrades
• Fixed Nagios Core bug re: retention of display_name and host alias, rather than reading from configuration
• Fixed installation of monitor pack when new plugin included
• Fixed possible bug with epoch time when importing into ODW
• Fixed cluster checks erroring on ssh banner between nodes
• Fixed check_snmp_linkstatus for when two checks run and complete within 30 seconds
• Fixed opsview-agent use on OpenVZ systems where wrong agent process was used
• Fixed excessive growth of opsview.session table by trunating session data over 7 days old
• Fixed display of audit logs and downtimes in local timezone rather than UTC
• Fixed viewports so that public keywords are listed correctly. Used to fail to be seen if other roles had this keyword in their access list
• Fixed case insensitivity of username when connecting via REST API
• Fixed bug where deletion of host group with wiki comments fails due to foreign key constraint
• Changed RPM packaging to use Requires instead of PreReq
• Fixed downtimes being recorded multiple times in Runtime database when Nagios Core restarted during a downtime window
• Fixed German translation for “Resolve service checks” (Tobias Stolz)
• Fixed Nagios Core CGI XSS vulnerability in config.cgi and statusmap.cgi
• Fixed XSS vulnerability in status pages
• Fixed not being able to disable Modules menu items
• Fixed setting timezone on database connection, caused by changes to DBIx::Class/Catalyst connection process
• Fixed host group graph page to use DBIx::Class model, instead of Class::DBI
• Fixed mass rechecks link when filtering by service handled
• Fixed Alerts - Hierarchy view when filtering by states
• Fixed SIGPIPE (error 141 out of bounds) in check_smtp
• Fixed dpkg upgrades where no MIB files reside in /usr/local/nagios/snmp/all
• Fixed opsview-agent rpm from starting nrpe when chkconfig is disabled on an upgrade
• Fixed Solaris finding SFW logrotate command by adding /usr/local/sbin to PATH (Emilio Scalise)
• Fixed check_smtp and check_tcp from displaying duplicate messages when checking certificates
• Fixed mrtg_genstats.sh to use bash not bourne shell
• Fixed viewport summary and viewport page so that the browser back button works properly after visiting those pages
• Correct port use on URLs by check_http_authentication
• Fixed sporadic error where viewports would show time in failure offset by wrong amount based on timezone

• Viewports enhanced for multi-tenancy environments. Individual viewports can either be:
  1. Publicly accessible
  2. Controlled via keyword access in a user's role
• Keyword list page now displays related objects and includes a link to status page, to see affected services
• Amazon AWS plugins now allow you to specify in which region your resources are located

• Fixed SNMP test for a host used as a slave
• Fixed startup script for opsview-web for a start() when already running (OPS-1551)
• Fixed send2slaves on Solaris platform
• Fixed viewport summary filtering of keywords when logged in with specific keywords
• Fixed viewport summary listing keywords with no services
• Fixed notification via keywords when all_keyword was selected at the notification profile
• Fixed formatting of object information where the display page doesn't show the same as the edit page.

• New alerts view “Unhandled - Hierarchy” based on host group hierarchy. Renamed other alert views
• Status pages now display filtering options

• The number of notification profiles are now listed under the Notifications tab for contact edit and user preference pages
• Notification profile information is only displayed in /user/preference if user has NOTIFYSOME access
• Help text added to provide guidance when no notification profiles set for a contact

• New DOWNTIMESOME and DOWNTIMEALL access points added to Roles. DOWNTIMEALL users can now create and cancel downtime at host group hierarchy
• You can now selectively delete downtimes, rather than all downtimes per object
• Nagios Core changes made to allow distributable downtime deletion commands
• Downtime scheduling page now allows Jira style time notation, such as ”+2h” and dynamically validates entered times for simpler entry
• On a reload, expired downtimes are automatically removed from the runtime database

• Now includes listing, setting and deletion of downtimes, to allow automated changes
• API call failures are now logged for debugging purposes
• Host group parent information includes matpath to uniquely identify parent host group

• Added in configurable 'SNMP max message size' option to query host and interface checks
• Added in a device scan lock on check_snmp_linkstatus plugin to prevent simultaneous multiple scans
• Remove some specified words from device interface ifDescr information to reduce instances of duplicate interface ids being used

• check_opsview_keyword plugin extended to calculate percent availability of all services in a keyword - useful for clustered checks
• New check_opsview_cluster plugin to calculate the state of clustered hosts to give an overall health of your host clusters

• Remove demo maps from NagVis as the opsview map acts as a working demo

• Set a bandwidth limit of 50Kbps on NMIS rsync tasks between nodes within a cluster to prevent network saturation
• Background cluster node takeover process to avoid Nagios Core time limit of 60 seconds

• Will now remove metric from popup menu when too many metrics are displayed on a page in order to simplify usage

• Extended System Survey to include more system performance data
• Updated System Preferences to include more information about the System Survey capability
• 'New Opsview Release' service check renamed to 'Opsview Updates' on new systems

• Added 'overrides' from opsview.conf to opsview.sh output
• New framework for updating configuration with sets of host templates, service groups, service checks and plugins
• New helper tables for service check and service group information in Runtime db

• Pass through non-existing host attributes unchanged on service checks
• Fixed host edit page for rancid options
• Fixed check_snmp_linkstatus raising 'UNKNOWN - The requested table is empty or does not exist' on devices that do not support ifAlias. Thanks to Christian Leonte
• Fixed URL links to be relative to web apps
• Fixed NMIS stats collection on Ellacoya devices without an alias set on an interface Thanks to Chris Coates
• Include 'notes' for services in slave configuration
• Fixed opsviewd not starting up properly if the database was down
• Fixed pid matching for nrpe process in init.d/opsview-agent for Solaris
• Fixed CONFIGUREKEYWORDS access not allowing new keyword creation
• Stop creation of Nagios Core contact definitions if no relevant access granted
• Made rsync_nmis_database more resilient if process died unexpectedly
• Reordered rsync_nmis_database invocations for slaves on an Opsview reload, to start earlier and avoid possible clashes between nodes
• Fixed cloning of service checks with disable_name_change set
• Fixed /rest/serverinfo which was giving wrong slave node status
• Fixed restarting snmptrapd on slaves when promoting a MIB
• Edit host page did not restrain options based on master monitoring servers
• Fixed layout issues of parents choose box on edit host page
• /rest/config/hostgroup?order=dependency traverses the tree, rather than ordering by level
• Allow IE8 to display Opsview web pages within an iframe on a different domain
• Fixed bug with service check “SNMP Trap - Link State” where exceptions were raised for other traps
• Added 'net-snmp' as a build requirement on SLES to create check_snmp binaries
• Fixed reloading image for Opsview Community edition

• There have been URL parameter changes for /status/host. If you use filter=unhandled or state=1, these now filter based on the services, not the host. Use host_filter and host_state instead. See documentation

• Host templates edit screen can now update list of hosts it refers to
• Smart host template removal to delete host specific monitors with host template removal. An upgrade will set this option to off to retain the existing behaviour
• Changed flapping values to default Nagios Core, to reduce possibility of flapping start and a smaller window before flapping stops
• Performance improvement for NDO inserts of host check results, now done in a single insert (thanks to Michael Friedrich @ Icinga)
• send2slaves now runs in parallel to all slave systems, reducing upgrade times for multi-slave distributed systems
• send2slaves now confirms that nagios binary runs on slave after sending new executable
• Added CSS class for allhandled items in a viewport summary
• Changed default Network Map Layout to Circular
• Reduced length of time files are kept in nagios/var/log from 20 weeks to 5
• Amend order of Solaris linking flags to reduced number of unnecessary dependencies in binaries

• Fixed memory leak in NRD::Daemon which caused import_slaveresultsd to increase its virtual memory size
• Force single instance of Solaris packages rather than allow in-place upgrades
• Fixed database install and upgrade scripts for Mysql 5.5
• Fixed discrepency between metric name from service's performance data and the RRD stored name
• Fixed performance viewport for metric which had all values as null
• Fixed IE7 errors on performance viewport and status viewports and interactive graphs
• Fixed help output on check_snmp_loadavg
• Fixed math calculations in check_snmp_memutil for warning and cricital limits
• Minor updates to debian control file
• Cater for zones on Solaris in opsview-agent init.d script
• Fixed copyright years

• Support for SUSE Linux Enterprise 11
• An improved mechanism for sending check results from Slaves to Master, based on NRD:
  • ability to send results with multi-line plugin output
  • 16K limit on amount of data sent, so long plugin output is not truncated
  • performance improvement of up to 50% when sending a large number of results
  • queuing of check results on the slave. results are now cached for up to 60 seconds before being discarded
  • results written to 'checkresults' directory, cutting out a Nagios Core processing loop
  • results are now transactional so if a failure occurs the whole transaction is aborted
  • automatic scaling of NRD processes as more slaves are created
  • timestamp now based on the client send time, not the server receive time
  • flexible packet size, so only the necessary amount of data is transferred
  • ability to switch back to the NSCA communication mechanism if desired
  • Plugin 'check_opsview_slave_node' returns CRITICAL if NRD is not working. Returns error messages and performance statistics
• New passive slave functionality for easier migration from existing Nagios Core distributed systems

• Status screens now display multi-line output
• Change to Status View by Service: Make the host link a “smart link” if not all services listed
• Viewport summary list / cells display a handled icon when all hosts and services within the viewport are in a handled state.
• Allow page title to be amended on viewport summary pages, to have custom viewport summary pages
• You can remove the top header by adding include_header=0 to URL
• Added breadcrumb navigation to include host object for easier navigation. Also added to interactive graph pages

• Access to monitoring objects now controlled by roles, rather than by contacts. This makes systems with lots of contacts easier to manage
• Keyword edit page now allows filtering of service checks based on hosts selected, to simplify choosing of related service checks
• You can now choose “All keywords” as list of objects for notification profiles and roles, for simpler configuration
• Created a role edit page with tabbed interface
• Added 'Date Format' system preference for Nagios Core CGIs
• The monitoring server page now displays the list of roles associated with it
• Converted audit logs to use DBIx::Class for faster responses in large systems
• Reduced size of roles list page if there are lots of contacts in system
• User preference page no longer shows all the different roles available, only the user's
• Introduced a new keyword for the from parameter, so that edit object can return to default list view

• API version number is now based on Opsview's version number
• Added monitoring servers to list of object types. These are read-only.
• Added a 'serverinfo' object to return back general information including list of monitoring servers, nodes and status/latency values
• Added easyXDM library for cross-domain REST communication
• Added icon path to image in host serialisation
• Added some response data when deleting an object via REST API
• Complex search available via json_filter
• Unexpected errors in delete operations are now caught, for better diagnostics
• Errors returned via JSON are now forced to be a string
• Mass object changes are now transactional
• API has a synchronise parameter, so you can delete all other objects not in your original list
• You can now request a one time login token for logging into web browser
• All JSON values returned are forced to be string values
• A URL parameter of 'order=dependency' will return list of objects with the most dependencies at the bottom. This is available for host groups and service checks

• All Opsview keywords are created as contact groups within Nagios Core

• Changed display scale for performance data with 'uom=bytes' to base 1024
• Automatically change y axis unit options list if 'uom' (based on the first performance data point) is bytes
• Hover over plot point now displays the automatic value and the raw value
• Empty fields in Add Series will trigger a search - hit delete to get a list
• We now allow certain default values to be changed in opsview.conf
• Add 'multiplier' and 'base' to RRD graph options

• Scrolling requests now result in only 1 JSON update. Timeline is not updated if hidden, in order to reduce JSON calls
• Any change of settings now resets refresh timer. New URL parameter: show_timeline to hide timeline if required
• Default for time-line controlled via system preferences
• Added support for markdown filtering from plugins

• Import speed improved by another 20%. Achieved by reduced unnecessary queries and performing multi-valued inserts (changes contributed in part by CAPSiDE)

• Reverse SSH tunnels are restarted as part of the 'send2slaves' routine, to ensure is running with latest port redirects
• It is now possible to change some NSCA options in opsview.conf

• Added Webmin support to virtual machine
• Add (commented) rewrite rule to the shipped apache conf file for easier access to webmin console

• Converted web ACLs to use Catalyst Action Roles
• Added extra ACL: CONFIGUREKEYWORDS
• Set Class::DBI database connections to automatically reconnect
• Updated all thirdy party perl modules

• Keyword name restrictions are now enforced. The upgrade script will convert any invalid characters to underscores. See upgrade documentation for more details

• Fixed servicegroup popup list for service check new page. Will migrate service checks without a service group to the first group
• Fixed systempreferences changes setting configuration status light
• Fixed showing welcome page when first logged into Opsview for admin user
• Fixed default value for auditlog's username column
• Fixed ambiguous term for hostgroup status query
• Fixed cron import_runtime errors from not having profile run
• Fixed SEGV when removing downtime by IP address
• SNMP polling of devices that return strings for numeric data are not converted to numeric information correctly (OPS-1429)
• Fixed header bar's curved edges when opsview not anchored to root
• Fixed performance viewport when opsview-web not anchored to root
• Fixed menu font size when looking at Network Map
• Fixed security issue where an authenticated user can view other user's notification profiles
• Fixed viewport page where links from viewport summary was not correctly filtering on failures
• Fixed check_reqs requesting libexpat when alredy installed on debian
• Increased Template Toolkit's maximum iterations limit
• Fixed bug where changing the plugin name did not update the plugin help text (introduced by XSS fixes)
• Show nagconfgen errors to stdout if a module dies
• Fixed list of hostgroups in the Events options panel for users that have all_hostgroups set
• Disallow spaces in host configuration page Hostname/IP field
• Fixed opsview_sync_ldap for keywords
• Fixed service check edit screen not storing when disabling all notification options
• Fixed size of version information on login screen
• Fixed webkit display of y axis lines not lining up correctly in graph
• Fixed 500 web server error if reload error messages were not linked to a monitoring server
• Fixed bug where realms were being displayed on user preference page. Can now change fullname field on contacts page
• Fixed link from /status/host page for services in table header
• Force plugins to be ordered by name in service check edit page
• Fixed excessive logging for import_perfdatarrd when errors found in perfdata parsing
• Fixed Nagios Core import of contacts
• Fixed large amount of data failing to get into runtime database (OPS-751)
• Fixed error in import_runtime if an object is created, downtime is then set and host is renamed within the same hour
• Fixed Nagios Core migration tool where host emails named differently
• Fixed REST API returning id for snmptraprules related to a service check
• Removed setting max_msg_size for check_snmp_linkstatus as some devices were not responding correctly with this set
• Fixed unnecessary extra json call for events data on initial page load
• Events view: Fixed refresh button, so no longer refreshes when auto refresh is off. Fixed respecting of auto_refresh=1 parameter
• Events view: Fixed new event where date time cell not coloured correctly
• Fixed mass-rechecks and mass-acknowledgements links when viewed from a host group page to filter correctly
• Fixed assumption of host id=1 was Opsview master for Parent Tree page
• Fixed Nagios Core bug where a host failure would cause lots of alerts to be raised for services on that host (OPS-1473)
• Fixed REST API where authentication into /rest/reload was using cookie rather than headers
• Fixed opsview_diag for coping with a schema change
• Fixed check_memory_solaris to cope with systems with no swap devices assigned
• Fixed possible error if nagiosgraph.log doesn't exist at startup
• Cleanup of debian packaging of init script for opsview-web
• Fixed logic for upgrade process so that new files are sent to slaves first, then a new configuration generation is done, followed by a software restart on slaves with no restarts on the master (packages will restart appropriately)
• Fixed possible bug where check_mysql_performance could have a bad path to save state data (thanks to Olegario Craig)
• Fixed Nagios Core bug where hosts with multi-line output were not parsed correctly
• Fixed bug whereby modules could not be cloned correctly
• Fixed bug in check_procs where regex input with a pipe symbol was being output causing problems for Nagios' parsing of perf data
• Fixed configuration for Alfresco Share Connection service check by setting Accept-Language (thanks to Robin Bramley)
• Fixed redirects on Nagios Core CGI host page so that Schedule downtime for all services on this host redirects to right Opsview page
• Made import_runtime more resilient in situations where a bad hostname or servicename were found in Runtime database

• UI updates to page headers (to increase amount of visible area - contributed by Capside)
• Improved readability of menus
• Performance data updates now run asynchronously, reducing gaps in graphs
  • You will need to manually create a service check on the Opsview master using the check_opsview_async_imports plugin with an argument of “perfdatarrd” to monitor the usage of the directory. See https://docs.opsview.com/doku.php?id=opsview-community:performancedataimports for details of how this works
• ODW now defaults to import standard statistics for all new systems
• Performance data in ODW is now summarised on an hourly basis, to allow fast queries of performance data for reporting
• Integration with SolarWinds SNMP trap forwarding, so you can forward traps for processing on Opsview master or slaves
• Performance improvements in post-reload tasks:
  • Status updates happen concurrently so views are up to date
  • The duration of the post-reload task is reduced with more efficient queries
• Performance improvements when querying Events view

• Added nsca_server_address option to override in opsview.conf
• Allow freshness for stale settings to still work if no notifications enabled
• Allow searching on plugin name in service check list page
• Update message about no interfaces, to suggest clicking Query Host to start the scan
• Graph values are now displayed to a base unit (bytes, seconds) for easier understanding. For example, performance data with a y-axis of MB and a line value of 1.3K will now be displayed with a y-axis of bytes and a value of 1.3G
• Viewports now show if a service is in a downtime state, to allow the user to assess the state of a service easier
• Better error message when import_runtime blocks due to upgrade flag (Tony Hunter)
• Upgraded Nagios Core to 3.2.3
• Added port option into SNMP tab on host edit page, for increased flexibility on a host basis
• Ensure hostgroup commands page downtime is redirected via Opsview web page, so all validation done appropriately (OPS-1396)
• The mass acknowledgements page now gives an option for sticky acknowledgements, to reduce the number of notifications for advanced users. The Auto Remove Comment has been taken out as this should always be the case
• When acknowledging a host or service from the Nagios Core CGIs, Sticky is now disabled by default. This is to prevent acknowledgements lasting longer than desired. More information about acknowledgements at http://docs.opsview.com/doku.php?id=opsview-community:acknowledgements
• When re-checking a host or service from the Nagios Core CGIs, Force Check is now disabled by default as this can cause additional checks to run in a distributed environment
• When a cluster node recovers, the status information is synchronised from the secondary node to the primary, to ensure that the status information is consistent
• check_snmp_cisco_loadavg now monitors Cisco ASA devices, to simplify configuration of monitoring Cisco devices
• Allow synchronous reloads from REST API, so you can have automated restarts of Opsview
• Removed Nagios Core restart messages from Nagios Core CGI trends view, to reduce clutter in trends graph (contributed by Capside)
• Can now enable partial ODW importing (statistics) and enable full ODW importing separately, reducing data storage requirements
• Automatically remove trailing spaces in service check names and in host attribute values, to avoid potential issues in Nagios Core configuration
• Added ability to get data from REST API using JSONP
• Added links to Resolve service checks from host edit page and back again, for easier navigation
• Removed AQL notification method from default install
• ODW now stores summary statistics of number of hosts/services/serviceresults/perfdata per hour
• ODW now automatically cleans up failed imports
• Added feature to RESTAPI for JSONP to return windowname method for cross-site REST calls
• Add icon to Modules menu for enterprise modules

• There are some manual post install tasks required. See http://docs.opsview.com/doku.php?id=opsview-community:upgrading#opsview_391
• As the old Opsview API did not clone the enable_snmp parameter correctly and since Enable SNMP overrides Use MRTG and Use NMIS, you will need to confirm that this flag is set appropriately for each host. This only affects hosts cloned via the API
• The REST API now requires setting the content-type and accept headers to get the expected data

• Fixed new creation of SNMP trap rule type service check redirecting to rules page on create/clone
• Fixed cloning of SNMP trap rules when cloning a service check
• Fixed error from check_snmp_linkstatus where pipe symbols in ifAlias were conflicting with performance data
• Fixed security exposure from NMIS admin script (thanks to C.H.Kang for reporting)
• Fixed incorrect servicegroup modelling, which caused deletion errors
• Fixed Cisco ASA devices using check_snmp_interfaces (Henry78)
• Removed incorrect debug logging of opsview_notificationprofiles
• Fixed permissions on error_pages/handle_error.cgi
• Fixed possible XSS vulnerabilities in web output on admin list pages
• Fix for status.cgi problem when no parameters passed
• Fixed host interfaces screen where used short interface names used
• Fixed return code from opsview-web init script
• Only show Acknowledge link if host/service requires it in popup menu
• Fixed host and host group wiki bulb lights
• Re-engineered check_snmp_linkstatus, due to changes in 3.9.0, to reduce the amount of traffic when rescanning a device
• Fixed “Cannot reload” messages which were missing error details
• Fixed NMIS error output on initial slave install as this can be ignored
• Fixed slave node resync, where it was not getting the correct list of hosts to send the latest Nagios Core status to
• Fixed bug with submit button when all notification options switched off in notification profile
• Fixed searching for host by name via API
• Fixed Opsview Web 500 error if badly structured data sent in to REST API
• Fixed Nagios Core trends.cgi from stretching to large sizes (contributed by Capside)
• Fixed bug where Server updated is not updated on Host status page
• Fixed pages=all link on configuration edit pages when using other languages (patch by Capside)
• Fixed central service check event handler configuration being pulled in to a host specific monitors list. The upgrade script will remove redundant information at host monitors, so a change to the central service check will be used as expected
• MRTG and NMIS configuration is not invoked now, if enable SNMP is off on a per host basis
• Cloning of Enable SNMP now works on a host, using the old Opsview API. However, hosts previously cloned will not have this parameter cloned, so will need to be set correctly
• Fixed UTF8 encoding from Nagios Core CGIs returned through Opsview
• Fixed visual glitch of background image on forms with a long amount of data
• Fixed 2 small Nagios Core CGI bugs
• Fixed italics and bold text formatting in host comments
• Fixed bug in REST API when searching by id

• New configuration API. Provides REST interface plus command line tools for data import and export. See our documentation for more details
• Added new performance viewport with sparkline graphs
• You can now dynamically update viewport style
• Graphs now have an export to CSV option
• Added Amazon Elastic Cloud (EC2) monitoring, Amazon Simple Storage Services (S3) monitoring and Slicehost server monitoring

• Updated Italian translation file (thanks to Emilio Scalise)
• Updated tagcloud viewport summary style to display highest failure state
• Allow specification of bind address for Opsview Web
• Amended Apache error page to allow for custom maintenance messages from /usr/local/nagios/etc/maintenance (OPS-1075). Note: requires Apache proxy configuration file update
• Added new 'Opsview Application Processes' check to 'Application - Opsview Master' template
• Added new 'check_opsview_mysql' and 'check_opsview_mysql_performance' for use with 'MySQL DB Performance', 'MySQL DB Connections' and 'MySQL Status' on master server (OPS-1158)
• Updated Nagios Plugins to v1.4.15 and Nagios Core to v3.2.2
• Updated label for notification methods and documented
• Reordered checkbox_groups so that it takes a newspaper column approach (down then next col), rather than text approach (left to right)
• Added new check_plugin_output to help with diagnosing plugin errors
• Enhanced check_snmp_linkstatus to catch database exceptions and create tables if required
• Allow up to four arguments for each host attribute
• Allow a default attribute to be set with arguments - this allows a global attribute that can be overridden per host
• Updated import_excel tool:
  • can set 'Enable SNMP' and 'Use MRTG'
  • can set NONE to have no keywords, host templates, service checks, etc
  • added warning about importing service checks, which will lose existing exceptions information
• Reload now complains if it finds trailing spaces in service check names
• Added unique constraint on servicegroups.name
• Separated hosts and host templates columns in service check list page
• Added unique constraints on SNMP trap rules and host template management URLs. If there are duplicate names during the upgrade, the ID will be suffixed to make it unique
• Added page numbers to top of configuration list tables
• Allow time period edit page to use cut and paste keystrokes. Added client side validation of timeperiod fields
• The number of changes since last reload will now be displayed on the reload screen
• Automatically convert attribute names to uppercase, otherwise it fails validation
• Contacts database backend converted so name > fullname and username > name. This is so that there is consistency with the name field always being the uniquely constrained column
• Added a message for contacts that do not have any notification profiles, to remind to create one for notifications
• Added /usr/local/nagios/libexec to PERL5LIB path for plugins that expect that location
• Added UDP port configuration option to SNMP plugins
• Added index to improve query for viewport summary pages
• Speedup to ndoutils_configdumpend when creating helper tables

• Fixed query_host when going to slave over a reverse ssh tunnel (Phil Rich)
• Fixed dependencies when using host templates without host defining that service check
• Fixed dependencies when multiple service checks depend on another service check
• Fixed dependencies when multiple service checks used where number of attributes = 0
• Removed listing of multiple service checks from options for dependencies
• Fixed temporary problem where downtime is removed from host/service statuses on a reload
• Do not clear runtime's scheduled downtime table, so state retained over a reload
• Fixed missing command_line field in runtime's nagios_servicechecks table
• Fixed graph load when selecting from more than 10 metrics
• Fixed graph refresh invoking 3 rrdfetches, when it should only be 2
• Stop auto-populated snmp polling label going over 40 chars
• Fixed error creating database when default charset=utf8
• Fixed error messages when running gathering plugin help during upgrade
• Include missed ODW table from nightly housekeeping (service_outages)
• Added help links for RSS fields in contact page
• Ignore nagios/var/spool directory in nightly opsview backups
• Stopped 'Successfully disconnected from MySQL database' messages to allow for syslog 'Last message repeated n times' on connections (OPS-341). Note: rsyslog needs to have '$RepeatedMsgReduction on' set to enable repeated message reduction
• Fixed possible vulnerability with invalid cookie values
• Fixed XSS vulnerability on login page
• Fixed bad contact notifications tab when only one notification method is active (OPS-1255)
• Fixed viewport summary keyword list so that only keywords with a service on are listed
• Fixed webexception page so that it doesn't try to do refreshes automatically
• Fixed incorrect eventhandler for cluster node failovers
• Fixed problem where FLAPPINGSTOP messages were not being notified correctly
• Fixed host information/comments page links on contextual menus
• Fixed Nagios Core CGI displaying no comments if there were any comments with persistent=0
• Added dependency for postgres libraries for opsview-base on Debian for check_pgsql
• Update slave sending results to use 127.0.0.1 rather than localhost to remove name resolution dependency
• Fixed DB upgrades which failed due to useragents primary key column being too large when charset is UTF8. Only affects upgrades from Opsview 3.4 upwards
• Fixed web interface where host's Flap Detection and Enable SNMP cannot be de-selected
• Fixed problem where apt-get install would hang for opsview-agent
• Fixed uncommitted flag when deleting a host
• Fixed problem for some SNMP interfaces where interface names contained trailing spaces
• Fixed issue on check_snmp_linkstatus where some SNMP devices returning msg too big errors - now uses same technique as query_host so if query_host works, check_snmp_linkstatus should too
• Fixed problem where MRTG links were broken for slave monitored hosts
• Fixed internationalisation for MRTG page
• Fixed on delete cascade for service check dependencies
• Fixed the uniqueness check for notification profiles names per contact
• Fixed missing titles on Nagios Core cmd.cgi
• Fixed check_snmp_cisco_ifstatus for interface ids that are not sequential (Claus Guttesen)
• Stop generating MRTG configuration for SNMPv3 hosts when MRTG < 2.15.0; put warning message on reload page
• Fixed the missing nagios/var directory on RPM agent installations