NMIS 8.5.6G

Friday 13 February 2015

Highlights for the 8.5.6G General Release

8.5.6 is a major new release with many exciting new features and improvements.

Usability, Robustness

• There is now a Basic Setup wizard/panel that automatically pops up on every load until dismissed. It provides a guided and easy-to-use interface for the most essential basic configuration settings. The menu item Setup now covers this basic setup panel and the other most common configuration dialogs.
• NMIS now performs a selftest before every collect or update, which covers disk space, operating system status, stuck NMIS processes etc. The results are displayed prominently in the GUI: the Metrics panel is replaced by the selftest status if there were any selftest failures. The selftest results are also accesible via the System/Host Diagnostics menu.
• NMIS does not create any new RRD files if the disk size component of the selftest fails. The logging of both successful and rejected RRD file creation was improved.

•  Config editing within the GUI was improved. Groups are now configurable in a convenient and safe fashion, and a number of bugs related to special characters were fixed. The GUI model editing infrastructure was reworked and now supports limited editing of model structures: existing elements can be changed or deleted, but no new elements can be added.

• Invalid values in configuration elements are now handled better, especially for editing nodes. It is still highly recommended that you don't use spaces or other special characters in node names, but our testing has shown great resilience even with quite ridiculously bad node names.

• Nodes can now have notes and the node editing dialog offers the notes field for editing.

• fpingd.pl now reacts to changes to NMIS configuration, nodes or events configuration, and restarts automatically if such changes are detected. fpingd now also logs more usefully, and fatal erorrs don't just vanish anymore.

• The default ping timeouts were adjusted to 5s (up from 0.3s). Nodes with RTT figures above that value are considered down.

• The notification and logging behaviour for events can now be configured (and disabled!) much more conveniently, using the menu Setup/Event Configuration. Model editing is no longer required.

• Backwards-compatibility with NMIS 4 was made a configuration option, because automated guessing and falling back could cause race conditions. system/nmis4_compatibility is the configuration option in question.

• NMIS now has a configuration backup tool which by default keeps the last 30 days of NMIS configuration, cron settings and model data in /usr/local/nmis8/backups.

• The file cleanup tool that NMIS enables by default now also cleans up any corrupt RRD files, and works better in the very common situation of nmis8/database and nmis8/var not being in the same filesystem as nmis8 itself.

• If you want to collect service monitoring data more frequently than normal type=collect activities,  then this is now possible: nmis.pl type=services can be run at any desired frequency from cron.

• The NMIS GUI now supports running with a custom configuration file better; the conf=<configname> URL argument is passed through correctly now.

• NMIS can now be temporarily locked by creating the file conf/NMIS_IS_LOCKED. The old mechanism (setting system/global_collect to false) still remains, but the new mechanism has the advantage of not requiring config editing and being simpler.

• The NMIS support tool now offers to fix the most common setup problems.

• The administration tools in admin/ are now subdivided into samples, archived/outdated and active ones.

• NMIS now ships with a command line configuration patching tool, admin/patch_config.pl, which will be handy for scripting configuration changes in large environments.

• The alert for high memory usage for an individual process was removed as it was too unreliable.

• Logging was generally improved and fewer nuisance log messages are created.

• A rolling history of the most recent NMIS operations (and how long they took) is now kept in var/nmis_system/timestamps/.

Display

• NMIS now has three modes for node status computation and display: "coarse", "classic" and "fine-grained" and the Basic Setup panel lets you switch between them. The differences are explained in detail on this page.
• The display of Systemhealth sections can now optionally include links to other pages or external sites. This feature is currently used for linking to a virtual/guest's NMIS page on VMWare ESXi hosts, if the virtual/guest is also managed by the same NMIS instance.

• Many graphs were improved. Service monitoring with external programs was improved and more of the collected data is shown on the default graphs. The service graphs for cpu and memory were split into separate graphs because combining those very disparate values on a single graph didn't work well.

• NMIS graphs no longer contain spurious spikes if a node resets or reboots.

• The per-node pages now include a link back to the dashboard if opened in a separate tab.

• The Nodes view now shows more information about when NMIS ran the last successful update etc. For nodes whose update times are far behind, those times are shown in yellow.

• The event log and nodestatus displays were improved.

New Capabilities

• New and improved device support for Scientific-Atlanta devices, Cisco IOSXE, Juniper, Net-SNMP, Ericsson-PPX, ...

• Most models were reviewed and improved to some extent.

• All models that support interface collection now also collect the interfaces' physical addresses.

• Improved support for dealing with layer 2/LAN concepts; NMIS now  supports the collection of CDP information, MAC and Bridge tables.

• NMIS now can syslog to tcp-based syslog recipients.

• NMIS email sending was reworked, an option to use STARTTLS was added, and setting custom SMTP ports now work fine. SMTP authentication is also much more robust in 8.5.6, and the logging of email sending was improved as well.

• NMIS now offers a plugin infrastructure for complex modelling scenarios. A number of example plugins are shipped in install/plugins, which also includes a README file describing the plugin infrastructure and capabilities. By default the installer activates all plugins (via copying to conf/plugins) except the TestPlugin. Plugins can be disabled globally if so desired, using the configuration setting "system/plugins_enabled".

• The nmis_post_proc facility was removed as the plugin infrastructure now provides that feature more efficiently and safely. The old "run custom plugin" facility was marked deprecated for the same reasons; it will be removed from the next version of NMIS.

 Installer

• The installer was reworked extensively and now performs all required operations for both initial installations as well as upgrades, which includes the installation of distro-level and CPAN prerequisites, apache 2.4 integration, automated crontab generation and so on. The installer was tested extensively on Centos 6, Debian 7 and Ubuntu 12 and 14.
• We now recommend that you use a system-wide cron snippet in /etc/cron.d/nmis, instead of the per-user root crontab, because that's much more robust and easier to maintain over time. The installer offers to create a suitable default snippet for you, and nmis.pl type=crontab can create both types of cron schedule.

• NMIS 8.5.6 recommends a different, simplified rrd directory structure. Common-database now contains those settings, and the tool admin/migrate_rrd_locations.pl is provided to perform the migration. The installer offers to run that tool on your behalf. Further details are available on this page.

• A tool for convenient bulk comparison between models directories has been added, admin/compare_models.pl. The installer offers to run it on your behalf automatically. The process for upgrading your models is documented in detail on this page.

NMIS 8.5.4G

Monday 24 November 2014

...

• Added config options for threshold_falling_reset_dampening and threshold_rising_reset_dampening
• Changed how runEscalate uses interface info, not working well for 500000 interfaces 
• Added new method to Sys.pm for per node interface index by ifDescr on demand.
• Fixed bug in dashboard where SNMP polled nodes not showing node down properly.
• Fixed nodeConf which was using Lower Case node names and not matching.
• New device support: 
  • PaloAlto devices
  • ServersCheck devices
  • Emmerson Energy Systems
  • Alcatel ASAM's
  • Mikrotik
  • more
• Additional mibs added for traps
• Fixed problems with JavaScript Dashboard, including upgrade of JQuery
• Added interface links to email notifications.

...

• Added support for Cisco AP, currently 1200 and 1240, but should work for others
• Added collections for Spanning Tree Topology Changes on a switch and the Number of Routes in a router.
• Added ZyXEL Switch support for MGS-3712F and GS2200-24
• Added temperature support for ZyXEL and Cisco Nexus (under environment)
• Added temperature thresholding
• Fixed PKTS_HC graphs to include better colours (so they can be seen)
• Display the PKTS_HC graph if available.
• Added interface descriptions to Interface up/down events and emails
• General model enhancements
• Added support for Cisco ASR9000 (IOS-XR) and Cisco ASR1000 (IOS-XE)
• Added run-reports.pl which was missing from NMIS8
• Added server health calculations to reflect CPU, Memory and Disk usage
• Service polling for servers now polls CPU and Memory for those services.
• Added full support for Cisco Nexus Devices
• Improved Support for Sun Solaris 5.10
• Converted to HTML5
• NMIS AUTH working with LDAP including MS AD LDAP
• Added a configuration option for setting the protocol in the emails which go out, nmis_host_protocol
• Added ability to notify using syslog requires Net::Syslog (add to install guide)
• NMIS can use snmpEngineTIme in the model instead of sysUpTime and this prevents bad Node Reset events.
• added logOrder field so Logs.nmis can be edited from GUI
• Fixed NMIS Auth splash screen
• Fixed bug with handling of event clearing, in particular stateful proactive events were not getting UP notifications.
• default authorisation privileges with configuration options for group as well
• added faster escalation for Level0, and optional skip escalation for syslog notifications.
• Added Locations, BusinessServices and Status to NMIS Nodes table.
• Add JSON Logging for escalations.
• added default values for Monkey business
• syslog_use_escalation
• disable thresholding globally with global_threshold, and on poll cycle with threshold_poll_cycle
• added custom alerts for creating alerts from SNMP results on the fly.
• fixed problem with SNMP Services being checked when node was down.
• Global model over-rides, added global overrides for the models, so models don't have to be edited individually.
• created models for TCP information, including graphs for tcp-conn and tcp-segs, added to Windows2008 and net-snmp for now.
• Created a config audit, mainly for checking file permissions
• Table configuration moved from tables.pl to configuration files, it is now possible to add new tables to NMIS dynamically.
• Fixed up file permissions in setFileProt, so now non root users can set group permissions.
• Added an NMIS Audit type=audit function and improved type=config to fix all file permissions
• Fixed the ip graph to better represent local and forwarded packets.
• Improved the CBQoS graphs.
• Improved LOTS of graphs.
• Asymmetric (e.g. ADSL) interface support in Node Configuration and throughout NMIS, including thresholding.
• Added many modelling enhancements to the net-snmp model, to assist with NMIS performance monitoring
• Consolidated models with pkts and pkts_hc graphs to save on disk space, this included updates to use pkts_hc data for thresholding.
• Enhanced custom alerts to cover all objects in the modelled node.
• Finalised custom interface modelling for the manual interface table collection for Alcatel.
• added a default Model with HC counters and ifHighSpeed support, must be manually selected.
• improved the look of graphs and made graph fonts configurable.
• new multi-tenant setup
• Updated NMIS OID file with new products from Cisco Product MIB, now scripted for future updates.
• Improved support for Cisco Catalyst IOS Devices
• Changed Catalyst 4000/5000/6000 running CatOS to NOT collect when no interface description.
• Added support for Cisco CSR IOSXR, should support other IOSXR devices.
• Added support for Cisco 1240 Access Points, should support other device types.
• Added testing mode for model verification, debug lite (mode=true)
• Updated login splash screen, included configurable MOTD and Custom Logo
• Contact level event level filter, define the event criticality which a contact is interested in.

...

NMIS Auth, Single Sign On (SSO) Configuration

• If using SSO, you must be using FQDN's for servers.
• Each NMIS server must be configured with the same auth_sso_domain and auth_web_key, and the auth_sso_domain must contain at least two "." (periods), so the correct auth_sso_domain value for nmis8.opmantek.com would be ".opmantek.com"

...

• fixed Windows2008 model for mib2ip being wrong type, created rrd_tune_mib2ip.pl to fix any bad mib2ip RRD's
• added mib2ip to net-snmp
• Removed deprecated MIB object ipOutNoRoutes
• Fixed an Internet Explorer Problem with popup graphs.
• fixed large post edit's in nodeconf
• found wrong datatype on TopChanges (STP Changes), admin/rrd_tune_topo.pl run=true change=true

...