Skip to main content
Declined

Fin du downtime, faux positifs

Related products:Infra Monitoring - Monitoring
  • Jun 5, 2023 1:27 PM UTC
  • 10 replies
  • 118 views

Forum|alt.badge.img+2

Bonjour, nous mettons en downtime des services groups lors d'indisponibilités programmées de ces services (ex : pendant une sauvegarde, une livraison applicative, un upgrade, ...). Notre soucis est que lorsqu'on met fin à ce downtime (=remise des services en "uptime"), le statut de ces services est celui du check (qui est souvent logiquement KO) précédent la remise en "uptime" (check toutes les 5 minutes). Existe t il une option pour que le moteur de supervision attende le prochain check d'un service lorsqu'on remet ce service en uptime, avant que le service soit effectivement "uptime" et pour éviter des faux positifs ? Merci. (NB : nos services groups contiennent parfois plusieurs centaines de services, donc nous ne souhaitons pas forcer un check sur l'ensemble du service group juste avant la remise en uptime)

Exemple :

minute 0 : service en downtime, service KO car upgrade en cours, check supervision = KO, état service Centreon = KO.

minute 1 : service en downtime, service OK car upgrade terminé, état service Centreon = KO.

minute 2 : FIN DU DOWNTIME (=service en uptime), service OK, état service Centreon = KO.

minute 5 : service en uptime, service OK, check supervision = OK, état service Centreon = OK.

--> Entre la minute 2 et la minute 5, nous avons un faux positif.

10 replies

lpinsivy
Centreonian
Forum|alt.badge.img+21
  • Centreonian
  • Jun 13, 2023 3:35 PM UTC

Hi @nicolasV maybe the downtime is not long enough.
In fact, the service(s) remain monitored for the duration of the downtime, using the regular check period.
If the downtime is long enough, the service(s) will be rechecked for the end of the downtime and will be in an OK status, not generating a false positive.

​


rchauvel
Centreonian
Forum|alt.badge.img+18
  • Centreonian
  • Jun 16, 2023 12:36 PM UTC
New→Discussion ongoing

Forum|alt.badge.img+2
  • Author
  • Steward *
  • Jun 16, 2023 2:47 PM UTC

Hi @Laurent ,

We automatically activate the downtime by script for a duration of 24h by default.

The downtime is automatically deactivated by script at the end of the operation, for example only 30 minutes after starting.

So we find ourselves in this situation of false positive.

One solution would be to add a 5mn wait to our script which deactivates the downtime, but this solution does not please us...


lpinsivy
Centreonian
Forum|alt.badge.img+21
  • Centreonian
  • Jun 19, 2023 8:05 AM UTC

Hi @nicolasV , If you use automation to start/stop downtimes, why you don’t re-schedule a check for services before to stop/cancel the downtime to be sure to have correct status for all services?

Regards,


Forum|alt.badge.img+2
  • Author
  • Steward *
  • Jun 19, 2023 9:03 AM UTC

Hi @Laurent ,

We thought about it, but we apply downtime on services groups, and some of which have several hundred of services, so it’s hard to force checks on it at the same time :
given the number of checks to be performed, this could take several seconds or tens of seconds. This would limit the duration of false positives, but would not fundamentally solve our problem
 
Thank you for trying to help me, but our best solution would be for the Centreon engine to return a service to uptime only if a check is passed after the end of downtime request…
 
Regards,

 


rchauvel
Centreonian
Forum|alt.badge.img+18
  • Centreonian
  • Jun 28, 2023 1:53 PM UTC

So if we understand correctly your ask is to perform a check when the downtime is over to know the current status of resources?


Forum|alt.badge.img+2
  • Author
  • Steward *
  • Jul 21, 2023 9:36 AM UTC

The only solution would be:

0- downtime a service

1- end of downtime request to the Centreon central server

2- the Centreon central server puts this request on hold until a new check is effective

3- Only after a new check is effective, the Centreon central server ends the downtime


rchauvel
Centreonian
Forum|alt.badge.img+18
  • Centreonian
  • Jul 26, 2023 1:16 PM UTC
Discussion ongoing→Needs Votes

rchauvel
Centreonian
Forum|alt.badge.img+18
  • Centreonian
  • Nov 22, 2024 9:57 AM UTC
Needs Votes→Declined

dagabard
Forum|alt.badge.img+5
  • Builder *
  • Nov 22, 2024 1:44 PM UTC

 

shift in time your downtime :-)