[ZBX-11402] Mail escalations wrongly sent on SNMP dependent triggers Created: 2016 Oct 27 Updated: 2024 Apr 10 Resolved: 2017 Feb 17 |
|
Status: | Closed |
Project: | ZABBIX BUGS AND ISSUES |
Component/s: | Server (S) |
Affects Version/s: | 3.0.4 |
Fix Version/s: | 3.0.8rc1, 3.2.4rc1, 3.4.0alpha1 |
Type: | Incident report | Priority: | Major |
Reporter: | Piotr Ciechoński | Assignee: | Unassigned |
Resolution: | Fixed | Votes: | 2 |
Labels: | triggers | ||
Remaining Estimate: | Not Specified | ||
Time Spent: | Not Specified | ||
Original Estimate: | Not Specified |
Attachments: | actions.png host_icmp_unavailable.png triggerA.png triggerB.png triggerB_eventDetail.png |
Team: | Team A |
Sprint: | Sprint 1 |
Story Points: | 0.5 |
Description |
There are two SNMP triggers on different SNMP items. Trigger B is dependent on Trigger A. Escalations are set to be sent immediately. In log there is an entry at 22-10-2016 09:58:51 that states: "temporarily disabling SNMP agent check on host". |
Comments |
Comment by Aleksandrs Saveljevs [ 2016 Oct 31 ] |
Could you please show some screenshots from the frontend (e.g., "Monitoring" -> "Events", event details, action configuration, and action log)? |
Comment by Piotr Ciechoński [ 2016 Nov 10 ] |
I have attached screenshots |
Comment by Aleksandrs Saveljevs [ 2016 Dec 05 ] |
Looking at process_escalations() and check_escalation_trigger() functions in escalator.c, it seems that escalations for dependent triggers are delayed. That is, if trigger B in the discussion above depends on trigger A and trigger A is in PROBLEM state, then escalation for trigger B is delayed. That would explain why notification for trigger B was not sent immediately, because even though it triggered before trigger A, escalations are processed in friendly bunches later (after events are generated and after escalations are initially inserted into the database), and at that time both triggers were PROBLEM. The next question is why did it suddenly send a notification at 09:58:53 on October 22? My guess would be that trigger A came into an UNKNOWN state, because its host was unreachable. Since it was in an UNKNOWN state, it was not seen as "blocking" trigger B's escalation (see DCconfig_check_trigger_dependencies_rec() function in dbconfig.c), so trigger B's escalation was finally processed. |
Comment by Aleksandrs Saveljevs [ 2016 Dec 05 ] |
The above comment seems to explain the behavior, so changing issue state to "Open". |
Comment by Viktors Tjarve [ 2017 Jan 12 ] |
Fixed in development branch svn://svn.zabbix.com/branches/dev/ZBX-11402 Notifications will no longer be sent from dependent trigger (Trigger B) when it depends on another trigger (Trigger A) which was in PROBLEM state and the host of this trigger (Trigger A) is unreachable. |
Comment by Alexander Vladishev [ 2017 Jan 13 ] |
(1) Documentation viktors.tjarve RESOLVED
All check by martins-v
REOPENED viktors.tjarve RESOLVED please review VSI
VSI Looks like all requested changes are done. CLOSED |
Comment by Sergejs Paskevics [ 2017 Jan 19 ] |
Successfully tested |
Comment by Viktors Tjarve [ 2017 Jan 23 ] |
Released in:
|
Comment by Rostislav Palivoda [ 2017 Feb 13 ] |
VSI please review documentation and mark as done or assign to me. |