[dev.icinga.com #2697] host/service selection for commands, leave out passive only checks on check reschedule in cmd.cgi #1011
Comments
Updated by mjbrooks on 2012-07-25 23:14:26 +00:00 I agree that a host/service that is fully passive shouldn't be allowed to be re-scheduled by default. The passive hosts/service could maybe be listed separately on the confirmation page with an additional "re-schedule these passive hosts/services too" check box. |
Updated by mfriedrich on 2012-07-26 06:56:21 +00:00 cool idea! +1 from me :) |
Updated by mfriedrich on 2012-07-31 19:00:59 +00:00
|
Updated by mfriedrich on 2012-08-01 01:07:03 +00:00 we could modify print_object_list, and add a second param, indicating the cmd_type. then there is a further problem - the passed host/service strings are not checked for integrity (so no find_host|service() calls) so we actually need to do that in order to filter out "passive only" hosts/services - this might decrease performance on the cmd.cgi form. |
Updated by ricardo on 2012-08-17 23:36:42 +00:00
in current 'dev/cgis' please Test and tell me what you think.
|
Updated by tgelf on 2012-08-18 09:33:35 +00:00 I like this feature, but I'd also opt for special care when implementing it. I'm aware of environments where people really want to reschedule passive-only checks having freshness-checking disabled. One example that comes to my mind is a specialized check daemons providing aggregating thousands of SNMP Traps and sending aggregated passive check results to different Icinga services. The "reschedule" is (miss?)used to tell the deamon to reset it's knowledge at a specific aggregation level. Others have managed it to set up working reschedule mechanisms in distributed environments. Listing passive checks separately as suggested by mjbrooks is much better. From a usability point of view I think someone choosing a passive only check for reschedule shall not be forced to declare "Yes, I want to do this with this passive only check". Usually in that situation he knows what he is doing. Therefore I'd suggest the following: if someone selects only passive checks when submitting a command, he could be shown a warning message telling him that he did so. However he should still be able to reschedule them without additional clicks. Cheers, |
Updated by ricardo on 2012-08-18 09:46:30 +00:00 Have you tested it? |
Updated by tgelf on 2012-08-19 09:20:36 +00:00 Hi Ricardo! No, I didn't test it yet - shall I? I stumbled over Michis "what do others think?", and as I know more than one environment where they are regularly rescheduling passive-only service checks I wanted to add my thoughts. Cheers, |
Updated by ricardo on 2012-08-19 13:29:55 +00:00 I would appreciate if vould test the patch. I could also add a select all checkbox. With this it should fullfill every need. |
Updated by mfriedrich on 2012-08-19 14:19:03 +00:00
i will test that after hunting #2993 |
Updated by mfriedrich on 2012-08-19 18:03:53 +00:00 either way - this works with the other checkboxes as well, and only the passive checks will be excluded.
aren't the checkboxes themselves a new feature at all? if so, please add a new issue for that, marking it a feature. as well as add a CHANGES entry to Changelog then. |
Updated by tgelf on 2012-08-20 19:46:14 +00:00 Looks great. How about showing icons with tooltips instead of expecting that people want to hover each single checkbox? There is plenty of space available... Regards, |
Updated by ricardo on 2012-08-20 20:25:36 +00:00 @tgelf: Go for it! |
Updated by mfriedrich on 2012-08-31 12:51:23 +00:00
i've done a little work on that.
imho those changes will allow users to identify the command table with the select boxes even better, as the style goes hand in hand with the status table. thoughts? |
Updated by mfriedrich on 2012-09-02 07:47:46 +00:00 what comes to mind - since we already read the status file on cmd.cgi now, we do have all the states within the objects as well. we could use the status.cgi color scheme for the background as well, indicating the current status of the service again on $whateveraction. what do you think - confusing or helpful? |
Updated by mjbrooks on 2012-09-02 08:40:41 +00:00 Wicked awesome job on this guys. @dnsmichi would it be much trouble to do a mock-up of what it would look like with the color? I like the idea, but it might be too busy visually taking one's eyes away from the checkboxes which are key to this page. Perhaps a column of empty appropriately colored squares to the left of the checkbox might work better? |
Updated by mfriedrich on 2012-09-03 17:35:04 +00:00
hm. i'll leave the coloring for an extra issue. my initial request is more than resolved, thank you ricardo. i think people should get to know the extra tick column, as well as the difference on rescheduling passive checks or not. and not getting confused by the coloring, etc. we can discuss the coloring schema in a different issue if needed. for me, it's resolved. |
Updated by mfriedrich on 2014-12-08 09:27:20 +00:00
|
This issue has been migrated from Redmine: https://dev.icinga.com/issues/2697
Created by mfriedrich on 2012-06-20 08:04:04 +00:00
Assignee: ricardo
Status: Resolved (closed on 2012-09-03 17:35:04 +00:00)
Target Version: 1.8
Last Update: 2014-12-08 09:27:19 +00:00 (in Redmine)
we are using the compound commands quite extensively, but in combination with check_mk (which got a special action_url to run the check_mk related command itsself).
now the thing is - the long list of "mark all" tick also marks the passive hosts/services, and this is totally fine when you want to acknowledge a problem.
but the specific command "reschedule checks for these services" also affects the passive hosts/services - which normally leads into a call on the command which is defined for freshness checking only. guess what happens - everything fucked up on the current check schedule.
i know that pre-filtering in the gui is difficult, but i'd like to propose a dedicated filter for cmd.cgi
since we already rewrote the tac.cgi in order to show "only passive hosts/services", it think it's worth adding such an enhancement filter as well.
what do others think?
Attachments
Changesets
2012-08-17 23:33:05 +00:00 by ricardo 449580e
2012-08-20 14:57:53 +00:00 by ricardo 2949ee4
2012-08-31 12:53:31 +00:00 by mfriedrich 372daa3
2012-09-18 23:18:25 +00:00 by ricardo 192d429
2012-09-24 21:20:09 +00:00 by ricardo 700d5e0
The text was updated successfully, but these errors were encountered: