cea-hpc / milkcheck

Highly parallel and flexible service manager.
Other
23 stars 6 forks source link

Add error nodeset per service in summary output #25

Closed wilfriedroset closed 7 years ago

wilfriedroset commented 9 years ago

Currently the output of milkcheck (with the summary flag) looks like this:

[root@nikka1 conf]# milkcheck -c . status --summary
status ping ran in 3.05 s
 > localhost: nikka2
 > localhost: nikka3
 > localhost: nikka5
 > localhost: nikka7
 > localhost: nikka8
 > localhost: nikka11
 > localhost: nikka12
 > localhost: nikka14
 > localhost: nikka15
 > localhost: nikka16
 > localhost exited with 1
ping - Ping all needed servers                                                                             [  ERROR  ]                                                                                          

 SUMMARY - 1 action (1 failed)
 + ping.status - Ping all needed servers                                                         

This output doesn't provide easy to read failed targets (nodes) per service. IMHO, the output would be better like this:

[root@nikka1 conf]# milkcheck -c . status --summary
status ping ran in 3.05 s
 > localhost: nikka2
 > localhost: nikka3
 > localhost: nikka5
 > localhost: nikka7
 > localhost: nikka8
 > localhost: nikka11
 > localhost: nikka12
 > localhost: nikka14
 > localhost: nikka15
 > localhost: nikka16
 > localhost exited with 1
ping - Ping all needed servers                                                                             [  ERROR  ]                                                                                          

 SUMMARY - 1 action (1 failed)
 + ping.status - Ping all needed servers
       nikka[0-1,4,6,9-11,13]                                                                             [NOK]
degremont commented 9 years ago

Could you post the configuration you used for this test?

wilfriedroset commented 9 years ago

From conf/samples/example.yaml

[root@nikka1 conf]# cat milkcheck.yaml 
services:
    ping:
        desc: Ping all needed servers
        mode: delegate
        target: "nikka[0-16]"
        actions:
            status:
                cmd: nodeset -S "\n" -e %TARGET | fping -r1 -u
            power:
                cmd: ipmipower -h %TARGET --stat
degremont commented 9 years ago

In this case, Milkcheck has no clue to guess that some nodes failed and other ran fine. delegate mode run an external command and milkcheck could not know what is really done by it.

degremont commented 9 years ago

However this could be done for regular services.

degremont commented 7 years ago

Closed by e5f60cf781a1e2e9d7210086547248c095a2f940