DevOpsIreland-Nagios.pdf

Transcript

1. Nagios, Icinga and other horrors Thursday, 14 November 13
2. Hands up if you've used Nagios? (or Icinga, we’ll not fork hairs) Thursday, 14 November 13
3. Hands up if you hate Nagios? Thursday, 14 November 13
4. Hands up if you've seen the light and use something else? Thursday, 14 November 13
5. What? Thursday, 14 November 13
6. Still using Nagios but using tools to manage the pain? Thursday, 14 November 13
7. Pain Management • GroundWorks (Monarch) • NagiosQL • Custom Tools? • Generate config from some central source? Thursday, 14 November 13
8. Nagios Config from your CMDB • Lots of different ways to do this.. • completely custom generation? • What system knows what should be on every machine you care about? Thursday, 14 November 13
9. Puppet • Puppet has supported Nagios resources for a long time • I’ve never used it, so we’re not going to talk about that Thursday, 14 November 13
10. Chef and Nagios/Icinga • “Official Cookbook” - Marius Ducea Installs from source templates need extended to support custom roles https://github.com/mdxp/icinga-cookbook • “Alternative Method” - Patrick Galbraith Fork of Marius Ducea’s cookbook Package based install commands, checks, services & roles all definable in databags https://github.com/moniker-dns Thursday, 14 November 13
11. Icinga Patrick Style • icinga::client recipe takes care of NRPE setup • NRPE checks are easy to define • define any ‘custom’ objects (services, checks, commands) in a databag • you can vary databags for different setups (I’m trying not to use the work Thursday, 14 November 13
12. Chef Roles map to Hostgroups • Nodes with the same Chef role get grouped into Nagios hostgroups • Services get applied to hostgroups Thursday, 14 November 13
13. data bag example (server side check, client-side nrpe definition) Thursday, 14 November 13
14. "commands":  {    "check_mysql_cluster":  {        "command_name":  "check_mysql_cluster",        "command_line":  "$USER1$/check_nrpe  -­‐H  $HOSTADDRESS$  -­‐c  check_mysql_clus    }, ... "nrpe_checks":  {    "check_mysql_cluster":  {    "command":  "/usr/lib64/nagios/plugins/pmp-­‐check-­‐mysql-­‐status",    "args":  "-­‐x  wsrep_cluster_size  -­‐w  2  -­‐c  2  -­‐C  '<='"    }, }, ... "services":  {    "mysql_cluster":  {        "enabled":  1,        "service_description":  "mysql_cluster  size",        "use":  "default-­‐service",        "hostgroups":  ["percona"],        "check_command":  "check_mysql_cluster"    }, }, Thursday, 14 November 13
15. Why am I doing this again? • Chef knows about all the hosts you care about • When the Chef client on your Icinga node converges, it has rebuilt the Icinga config with all the nodes currently in Chef Thursday, 14 November 13
16. Flaws • We’re still using Icinga • We should be using Sensu (or something else agent based) Thursday, 14 November 13
17. Image Credits • laralinchenx • frankandmarysflickr Thursday, 14 November 13
18. Thank You • Simon McCartney Twitter: simonmcc • Patrick Galbraith Twitter: capttofu • Kiall Mac Innes Thursday, 14 November 13