[dev.icinga.com #9986] Make log replication toggleable

[icinga-migration]

This issue has been migrated from Redmine: https://dev.icinga.com/issues/9986

Created by mfrosch on 2015-08-24 08:48:44 +00:00

Assignee: (none)
Status: Rejected (closed on 2015-08-25 12:14:19 +00:00)
Target Version: (none)
Last Update: 2015-08-25 12:14:19 +00:00 (in Redmine)

Backport?: Not yet backported
Include in Changelog: 1

---

We need an intermediate change in 2.3 until we can improve the behavior how replay logs are stored (#9730)

The idea is to allow a user to disable log replication per Endpoint/Zone, so a big Icinga master won't iterate over the log for every client.

The long term fix would be to split replay logs properly (#9730)

For now (2.3) the solution could look like this:

object Endpoint "someagent" {
  host = "1.2.3.4"
}
object Zone "someagent" {
  endpoints = [ "someagent"]

  replay_logs = false
}

Usually on a normal agent or a satellite the logs to be replayed would be tiny or non existing.

What causes us problems is that Icinga 2 has to read it's local log to evaluate what has to be sent to the client.

This flag can disable the behavior, and would not send any "old" logs from the master to that zone. For a simple agent, either in checker or command_endpoint usage, this would not matter.

---

Relations:

• relates #9983
• relates #9976
• relates #9730

[icinga-migration]

Updated by mfrosch on 2015-08-24 08:49:01 +00:00

• Relates set to 9983

[icinga-migration]

Updated by mfrosch on 2015-08-24 08:49:12 +00:00

• Relates set to 9976

[icinga-migration]

Updated by mfrosch on 2015-08-24 08:49:17 +00:00

• Relates set to 9730

[icinga-migration]

Updated by mfrosch on 2015-08-24 08:49:45 +00:00

This should fix bugs like #9983 and #9976

[icinga-migration]

Updated by mfrosch on 2015-08-24 08:58:08 +00:00

• Target Version changed from 2.3.9 to Backlog

[icinga-migration]

Updated by mfrosch on 2015-08-25 12:14:19 +00:00

• Status changed from New to Rejected
• Assigned to deleted gbeutner
• Target Version deleted Backlog

Feedback: That flag is not needed, one can set log_duration to 0 for having the same effect.

This may help on problems with connections to agents waiting for a long log read on the master.