Child pages
  • NMIS 9 Release Notes

Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

Table of Contents

NMIS 9.0.6c

The beta release was published on 12 September 2019.

Please contact us at beta@opmantek.com if you're interested in trying out NMIS 9 pre-releases.

Highlights

  • The dbcleanup function available in nmis-cli has been improved:
    • The process does not die if there is a failing query. The loop iteration will continue. 
    • New cleanup queries where added to cleanup inventory, latest_data, status and events collection where they have a node_uuid associated but the cluster_id does not exit. This will prevent a node appears as duplicate in opCharts, when a node is duplicate in different pollers and synchronised by opHA (Duplicate nodes in the poller is not supported yet). 
    • New option was added - use_performance_query - to use a non lookup query that could fail with big amount of data. This query is not used by default, as it could increase the nmis9d CPU usage.
    • Log messages added. 
    • Example of use: act=dbcleanup [simulate=t/f] [info=t/f] [use_performance_query=t/f]

NMIS 9.0.6b

The beta release was published on 3 September 2019.

Please contact us at beta@opmantek.com if you're interested in trying out NMIS 9 pre-releases.

Highlights

  • Bugfix in the rename node process failing when the new folder already contain the same rrd file names. Now the process is as follows: 
    • If there are some existing files in the destination folder, NMIS9 will try to rename the existing file to file-name.rrd.duplicate
    • If it can't rename the existing file, it is going to log a warning. The file will have to be handled manually. 
    • The process continues.
  • Now the collector adds the configuration.node in the latest_data collection and an index for this field and time. This allow certain queries to be more efficient, specially for opCharts 4.0.7B topN. 
  • Restore node now updates the last_update field. 

...

Please contact us at beta@opmantek.com if you're interested in trying out NMIS 9 pre-releases.

Highlights

  • Bugfix in the rename node process failing during the movement of certain rrd files. 
  • Now the collector adds the node_uuid in the latest_data collection. This allow certain queries to be more efficient, specially for opCharts 4.0.7A topN. 
  • Bugfix for new installations that prevent the process to complete. 

...

Please contact us at beta@opmantek.com if you're interested in trying out NMIS 9 pre-releases.

Highlights

  • NMIS 9 configuration can be now set up using the master node
  • Fixed bug when renaming a node on a specific situation, when the node had track of files that doesn't exist anymore. The message was "Failed to relocate inventory storage XXX: file "/nodes/namenode/interface/inloopback0.rrd" does not exist, cannot relocate!".
  • Fixed bug in the NMIS interface that, in some situations, shows the wrong color (green) for unreachable nodes. It happened when the node was unreachable but a node down event does not exist. 
  • Fixed cgi-bin/models to handle distinction between config nmis_default_models and nmis_models directories for listing, and to use nmisng::util::getmodelfile to read the appropriate model file (custom or default); any written out data already went to the custom model directory.

  • Replace legacy logging code.
  • Fixed "Network Tools >> SNMP Tool" that was opening when the node was not active. 

...

Please contact us at beta@opmantek.com if you're interested in trying out NMIS 9 pre-releases.

Highlights

  • Smarter non-interactive installation with Preseeding
  • Support to See poller nodes from the master.
  • Fixed visualisation problem with numeric node nodes. Nodes with only numbers on the name, like "12345" were causing problems on the system.
  • Show poller_event_log by default if the NMIS9 is a master, in case syslog is configured to get network events on master.
  • mojolicious 8.x is a new dependency that has a totally rotten default format for mojo::log. Fixed log format and set log_level to info which is no longer the default.
  • Fixed race condition in the poll process causing the catchall inventory data not being updating propertly when the ping operation runs at the same time. 
  • Fixed fpingd process on debian 10 was causing heaps of space padding to be added if $0 is changed.
  • Fixed NMIS9 runtime graph was not working. 
  • Change sysLocation field title to 'SNMP Location' to declash with manually configured location field, which is titled 'Location'.
  • Added hr* sections to net-snmp model so opCharts can know what charts to make. 
  • Fixed network_summary_group view had empty columns.
  • Fixed node response time graphs for nodes (shown NaN on previous):

...

Please contact us at beta@opmantek.com if you're interested in trying out NMIS 9 pre-releases.

Highlights

  • nmisng nodes now handle cluster_id on get_events_node: This is going to allow opHA to remove events from the poller on a peer deletion, so no data from the poller should remain. Also fix node_admin when dumping a node from the poller (Now will be able to get also event_data). 
  • Show message on delete node errors. 
  • Minor fixes ported from NMIS8:
    • Added support for fping inter packet gap for some firewalls. New configuration options added: fastping_interval and fastping_target_interval. 
    • IPrevent NMIS from failing when an escalation event element is c:\\
    • Adding the new opConfig "Node Configuration Change Detected event"
    • Fixed newlines being included in WMI error messages
    • Calling /proc/cpuinfo in the Support ZIP
    • run-reports.pl makes sure the reports directory exists
    • Optimisation to stop running services if no service polling required on a node
    • work around net-snmp snmpd which reports almost anything as hrfixeddisk
    • fixes for the Cisco Temperature model
  • loadServiceStatus: A new parameter is added that allows filter by all clusters. This will allow opCharts to show Monitored Services from the poller. 
  • services.nmis: Added Opmantek.pl
  • nmis-cli: On job type collect wasn't filter by local nodes.
  • purge_op_status: Changed to 7 days by default.
  • Display node name in opstatus.pl

...

Please contact us at beta@opmantek.com if you're interested in trying out NMIS 9 pre-releases.

Highlights

  • Avoid pre-fill the wmiusername and wmipassword on edit/add a Node.
  • Prompt a confirmation window before deleting a node.
  • Show only local nodes. Nodes from the pollers are not shown. 
  • Fix to being able to manage a node when having another node with the same name from the poller. 

...

This beta release was published on 28 May 2019.
Please contact us at beta@opmantek.com if you're interested in trying out NMIS 9 pre-releases.

Highlights

  • The main page load time was optimised by reducing the network_summary_view load time. 

...

This beta release was published on 17 May 2019.
Please contact us at beta@opmantek.com if you're interested in trying out NMIS 9 pre-releases.

Highlights

  • Lots of bug fixes, robustness and performance improvements.
  • It is no longer necessary to configure the list of groups that the NMIS GUI should show;
    by default members of all groups are visible now and only groups that are explicitely configured to be hidden (using the hide_groups  configuration item) are omitted from the display. Groups can be created and assigned freely when editing or creating nodes, both with node_admin as well as using the GUI.
  • Self-check faults are now logged as stateless events for node 'localhost' (if such a node exists).
  • NMIS 9 now supports authentication using the system's PAM authentication infrastructure.
  • NMIS 9 now properly and fully supports case-sensitivity in node names everywhere.
  • "-node" and "-info" files in var are no longer required or created.
  • Various information shown on the node dashboard is now updated immediately (e.g. "last ping" timestamp"), and no longer refreshed only during collect operations.
  • The node_admin tool now supports more complete snapshotting of nodes (with act=dump), which optionally includes the node's RRD files, events and other historic records.
    When importing a thusly dumped node with act=import it is now possible to have all identifiers localised to the current system (with localise_ids=true ); this causes the imported node to be  'adopted' by and become active on the current NMIS system immediately.
    This mechanism allows a node to be moved completely between NMIS systems, without losing any of the node's history.
  • Minor model improvements.
  • More efficient node configuration structures.
    Please note that it is necessary to run bin/nmis-cli act=noderefresh  once after upgrading to activate those changes; the installer will normally perform this operation for you.

...

This beta pre-release was published on 11 Apr 2019.
Please contact us at beta@opmantek.com if you're interested in trying out NMIS 9 pre-releases.

Highlights

  • The nmis9d (and the NMIS 9 installer) now interact cleanly with both systemd and sysv init systems.
  • Installing on Ubuntu 18 now works.
  • Orphaned worker processes left behind by a crashed nmis9d are now cleaned up more quickly and  reliably.
  • various feature extensions in admin/node_admin.pl, e.g. act=dump and act=restore can now also capture and restore a node's RRD files
  • Node deletion was  improved to ensure no scheduled collection jobs remain or interfere, and also includes historic/inactive events and operational status records.
  • Support tool no longer captures leftover legacy configuration files
  • Selftest is less likely to produce false positives
  • Node editing in the GUI now presets the configuration fields with correct defaults
  • The fping infrastructure now correctly handles the case of an admin modifying a node's IP address instead of caching stale data. 
  • Service tests whose monitoring scripts return unexpected exit codes are now  treated as 'failed/service down' and such occurrences are logged.
  • Fixed race condition in the configuration loading code, which could cause daemon crashes if the configuration is updated frequently (using the GUI or admin/patch_config.pl).
  • Various bug fixes and robustness improvements

...

This is a beta pre-release and was published on 21 Aug 2018.
Please contact us at beta@opmantek.com if you're interested in trying out NMIS 9 pre-releases.

Highlights

  • Feature Parity with NMIS 8.6.7G
    The improvements made in NMIS 8.6.7G have been incorporated into NMIS 9 where applicable.
  • Substantially improved GUI rendering speed and reduced resource usage
    The GUI is now usable on a system with only one CPU core and 2 GB of ram.
  • Now supports running on systems without systemd better: the installer now provides a classic init script for MongoDB
  • admin/node_admin.pl was extended to offer more flexible import and export options.
  • Various bug fixes and robustness improvements

...

This beta pre-release was published on 14 Jun 2018.
Please contact us at beta@opmantek.com if you're interested in trying out NMIS 9 pre-releases.

Highlights

  • Feature Parity with NMIS 8.6.6G
    NMIS 9.0.0c supports the new  Polling Failover mechanism, and all  recent improvements made in NMIS 8 are present in NMIS 9.0.0c (where applicable).
    This also includes the recent improvements for Single-Sign-On.
  • Can run in parallel with NMIS 8
    If your server specs are suitable (min. 4GB of RAM, 6-8GB recommended), then NMIS 8 and NMIS 9 can be installed on the same server without interference.
    NMIS 9 normally installs itself into /usr/local/nmis9, and its web entry point (http://localhost/nmis9/) doesn't interfere with NMIS 8.
    If the installer detects an NMIS 8 instance on your server, then it offers to import the NMIS 8 nodes' configuration: all nodes would then be polled in parallel by both NMIS 8 and NMIS 9.
  • Full installer support for platforms Debian 9, Ubuntu 16.04, CentOS 7 and 6.
  • Improved installer behaviour for both installations from scratch and upgrades from 9.0.0b.
    Upgrading from earlier releases of NMIS 9 (nightly or 9.0.0b) should now be supported seamlessly, ie. all required structural changes should be taken care of by the installer.
  • Automatic priming for monitoring of localhost, ie. the NMIS server itself
  • Node administration suite is now feature-complete
    admin/node_admin.pl can now perform all typical node operations, as well as export and import of a node's complete database information for diagnostic purposes.
  • Improved and more consistent logging
  • Fine-grained Operational Status information
    NMIS 9 now creates operational status overview records for every operation that is performed in the background.
    This operational status is accessible from the GUI (in the menu under System -> Host Diagnostics -> Ops Status).
  • Improved robustness and flexibility of the job scheduling logic
    Long-dead nodes are now demoted to fewer connection attempts after 14 days of inaccessibility.
    Job priorities can now be freely configured (see priority_schedule in conf-default/Config.nmis).
    The intial update operation for newly added nodes is now automatically prioritised above all other operations.
  • Improved self-test capability and support tool.
  • More flexibility for manually scheduled jobs
    bin/nmis-cli can now schedule any job with a specific (higher or lower than default) verbosity, which will affect just that one job.
    See the help text from bin/nmis-cli act=schedule for details.
    Furthermore, all log output for a particular job can be redirected to a separate file as well (using job.output=/some/path/nameprefix).
  • NMIS 9 CLI Improvements
    bin/nmis-cli can now show the live NMIS daemon and worker process status (with act=status).
    It is now possible to delete or abort scheduled jobs in bulk, using bin/nmis-cli act=delete-schedule id=ALL job.type=<something> job.uuid=<somenodeuuid>; note that "id=ALL" is required to indicate that all matching jobs are to be removed.
  • Various Minor GUI Improvements
    The Node dashboard widget now displays the last time  for ping, collect and update separately, and if there is a job in progress or pending for this node, then that is shown as well.

...

Version 9.0.0b is a late alpha/early beta pre-release, which was published on 6 Apr 2018.
Please contact us at beta@opmantek.com if you're interested in trying out NMIS 9 pre-releases.

Highlights

  • Full installer support for platforms Debian 9, Ubuntu 16.04, CentOS 7 and 6.
  • This version can coexist with NMIS 8 on the same machine.
  • New MongoDB backend which now holds almost all node and status information.
  • Better long-term maintainability (no more configuration file or default model copying necessary)
  • New NMIS polling engine which now uses an nmis9 daemon and a configurable number of worker processes,
    which results in more even (and somewhat reduced) server resource utilisation.
  • Daemons support reconfiguration without restart for verbosity/debugging changes with signals USR1 (more verbose) and USR2 (less verbose)

...