Shinken v1.0 Release Notes
Release Date: 2012-02-28 // about 12 years ago-
CORE ENHANCEMENTS * Add: shinken.sh script and skonf.py with a LOT of features for setup and configuration management. * Add: better logging for warnings for host, service and npcd configuration * Add: poller daemon will use all available CPUs by default * Add: poller daemon can now load standard external modules like namedPipe to get external commands from it * Add: brok queue watchdog for satellites, activated for Broker by default at 100000 broks to kill/restart the module.. * Add: if the run/var directories are missing, try to create them in the init.d script. * Add: service templates can now be applied on a host template * Add: reactionner can run under android * Add: android SMS module * Add: imlemented a check for dead queue threads. When detected restarts the thread * Add: mail python script * Add: better error text when the pickle load gets an error * Add: manage connexion timeouts in the broker and Pyro4 * Add: $$ are now interpreted as $ in macro solving * Add: (reported by: sduchesneau) check_shinken.py now has a timeout. It is set by default at 10s * Add: business rules now support the NOT operand * Add: send broks in external modules as a biglist instead of n broks. It's more efficient (X3 perf for this, can be huge time for huge conf) * Add: templates for monitoring common devices and services are now included * Add: manage \; to be changed in ; without cutting the rest of the line to be removed as a comment * Add: (reported by: foobar1111) arbiter only matched hostname, and not fqdn. Now it will try fqdn and then the hostname. * Add: retention is now enabled by default at installation time (Pickle) * Add: (From Michael Leinartas) man pages updates * Add: '+' in discovery rule management, so you can 'add' something, like a template * Add: using an undefined template in the configuration is no longer a critical error, just a warning. So the user can just 'tag' hosts first and then create the template * Add: Sebastien Coavoux conducted a large code review! :) * Add: xkilian reviewed and corrected many many pages of the wiki documentation and a few english fixes in the distribution :) * Add: skonf daemon preview. It is not yet meant for production, but is included to get feedback * Add: mongodb insert capabilities for the discovery * Add: TIMEPERIOD TRANSITION logs from the Arbiter to get timeperiods transition output * Add: is_admin property is now available for contacts * Add: make the nmap discovery do tcp and udp scan * Add: (Victor Igumnov) installation for Solaris CORE FIXES * Fix: use StringIO to read configuration files into a string (much faster with lots of config files) * Fix: (reported by: lminoza) bug in inheritance if templates give only a + elements, should continue to loop and not stop here. * Fix: keep multiple spaces in config files. (check_command!"args " were stripped) * Fix: pyro 4.10 management * Fix: (reported by: MINOZA, Landy) multi-layer management for service template on host template. So if you apply a service on layer 3 and your host inherit from layer1->layer2 * Fix: (reported by: Thibaut Notteboom) catch ConnectionClosedError on poller/reactionner connections * Fix: A better safe_print * Fix: lot of encoding stuff! * FIX: proxy support and python-setuptools installation on RHEL/CENTOS 6 * Fix: (reproted by: darkweaver87) somethimes we got a malformed external command. If so, bailout and warn in debug mode. * Fix: when there is a dispatching problem, the arbiter send and send again the same configuration to satellites. * Fix: crash bug when timezone set in config file * Fix: make all log pass the utf8 management * Fix: use auto-generated, absolute path for README (Daniel Widerin) * Fix: last line of email notification must finish by a \n otherwise it will not be sent (Laurent Ollagnier) * Fix: (from rootix2) catch case where service do not have imported_from from modules. * Fix: (reported by: Venelin Petkov) print in hostgroup names was a problem. * Fix: if a scheduler restarted, we got problem because the arbiter did not resend it the configuration * Fix: "host_notification_period" and "service_notification_period" defined twice in the contact definition (Fournet Matthieu) * Fix: (reported by: Httqm) default notification_interval was one minute, 1hour is better. * Fix: (reported by: sprudhomme) longoutput parsing and perfdata did not follow nagios way. * Fix: change sourceforgetrac with github in tracebacks. * Fix: (reported by: lminoza) too many notifications for a contact with multiple notification methods, all where sent when only some would pass filter. * Fix: (reported by: puisea) kill all sub process action tree on unix * Fix: flaw on multiple date and multiple timerange * Fix: if missing alias in objects, put the name if available * Fix: (reported by: denetariko) manage spaces before and after type name in define line * Fix: (reported by: Steve Kieu) missing get_name() method for config * Fix: (reported by: Mihai Efrim) flapping notifications MODULE ENHANCEMENTS * Add: rewrite the LiveStatus module, performance boost and simpler. Thanks Gerhard for this huge undertaking! * Add: two new modules for storing livestatus logs. sqlite and mongodb * Add: graphite module for the Broker to export performance data to a Graphite time series database. * Add: module to deal with flat file dependencies * Add: mongodb module for the arbiter, to load hosts object from it * Add: mongodb retention module * Add: the Service-perfdata module now open/close the perfdata file each second. So it's compatible with tools like Centreon that move it * Add: macros supported in skonf. Lots of bug fixes * Add: merlindb now inserts host_contactgroups * Add: NDO Nagios/Shinken mix in database (Sebastien Coavoux) * Add: port option for the mysql connexion with NDO * Add: ip_tag module that can change configuration properties based on the host IP address or IP address range. * Add: simulation mode for the nmap disco wrapper, so we can ask for xml output from users, and then simulate it easily. * Add: new process for nmap discovery output. There is no longer a big mapping pass. It will output what it can, and it's for the discovery_rule to do this job now, it's far more a BIG thanks to all folks that send me sample xml output :) * Add: enhanced capabilities for the mysql import module (OlivierHa) MODULE FIXES * Fix: useless and dangerous str in redis module. * FIX: htpasswd.users file was not correctly updated at installation * Fix: (reported by: DGuenault) hack_poller_tag_by_macros was not applying poller_tag to commands, only root objects * Fix: memcache and redis retention modules. They were loading notifications->contacts->commands but this class got slots but no __getstate__ * Fix: (reported by: sprudhomme): long_output in ndo for 1.4b9 version * Fix: NSCA, a loop when the client initiate the socket closing UI * Add: now page rendering can be launched in parallel, and so page can now call 'long' queries without breaking all. But during a long query, we still cannot eat broks. * Fix: the WebUI now exits when the brok thread encounters a problem * Add: UI PNP module * Add: if contact is not is_admin, then the /problems view only shows its related elements (if he is a contact of the elemetn, of an impact or a source problems one). * Add: add graph time selection in the detail page * Add: Graphite graph backend * Add: /mobile part * Add: External Authentification for WebUI (SSO) (olivierHa) * Add: tips on the eltgraph view * Fix: acknowledge typo * Fix: Change the scale of the impact view divs * Fix: UI page navigation should not propose page too high if not need * Add: gesture canvas as smaller and visible * Add: address to the eltdetail.tpl * Fix: utf8 names in the UI * Add: allow_html_output parameter for the WebUI * Add: generic perfometer manager for the UI * Fix: search button (Thanx to pydubreucq) * Add: make all nodes apears in the dep graph, but hidden one in very smalls * Add: perfometer image hovering with the graph module like PNP * Add: password iphone like in the login screen * Add: an orange outline to see where the focus is on input