Bug: Missing entries in the journal

[cdprete]

Spring Boot Admin Server information

• Version: 3.5.0

• Spring Boot version: 3.5.0

• Configured Security: None

• Webflux or Servlet application: WebFlux

Client information

• Spring Boot versions: 3.5.0

• Used discovery mechanism: Eureka

• Webflux or Servlet application: WebFlux

Description

Hello.

I start by saying this may not be a bug but, maybe, just me not clearly understanding how the journal and its events work.

With that being said, I can see that in the journal the entries for my application are not updated since the 27th of May:

But, given that I've ProcessInfoContributor enabled as well as another custom contributor (see #4253), I would expect to have them update at least once a day or more.
On the other end, if I inspect the payload that's shown in the last entry, it reports correctly the latest available information (at least, for my custom contributor that updates once a day).

I wonder if this may be a bug or, maybe, given that I'm using Hazelcast for clustering (see #4346), if the latter may be the cause.

[ulischulte]

Hi @cdprete,
I'm not sure if I understand you correctly. Are you missing audit events or process info?
The journal view shows data from the auditevents actuator endpoint. ProcessInfoContributor contributes additional info to the info endpoint and won't show up here as long as you don't publish an event yourself.

[cdprete]

Hi @ulischulte

Basically I can't explain to myself why there are those gaps as well as potential missed updates in the journal.

[ulischulte]

Just to clarify things - the journal view shows data from the auditevents actuator endpoint, not from the info endpoint. Are you missing/expecting data from the info endpoint in the journal view?

[cdprete]

Not directly. I mentioned it, because the event is named INFO_CHANGED and its payload is the composed by info returned by the info endpoint.
At least, for what I can see in the UI.

So, my assumption was that every time the info change a new INFO_CHANGED event should be visible.
But, of course, I can easily be wrong.

As said, what I would like to understand is why there are events missing.

[ulischulte]

Hi @cdprete
I see. You're right, any InstanceInfoChangedEvent should trigger the InfoUpdater and result in a new entry in the journal.
Since we're not able to reproduce your issue, could you please provide a minimal sample project?
Regards,
Uli

[cdprete]

Hi @cdprete I see. You're right, any InstanceInfoChangedEvent should trigger the InfoUpdater and result in a new entry in the journal. Since we're not able to reproduce your issue, could you please provide a minimal sample project? Regards, Uli

Hi @ulischulte.
Just enable the ProcessInfoContributor from your side and, ideally, the info should change on every refresh since the memory informations displayed in there change.

[SteKoe]

InfoUpdater in backend is called in whatever spring.boot.admin.monitor.info-interval (default: 1m) is set to. This is mainly used to check every minute wether the instance is still up. In addition spring.boot.admin.monitor.info-lifetime (default: 1m) sets the lifetime of the last info data update. Since it is both set to 1m by default, you see a InfoChangedEvent just every 2nd minute.

If you set the lifetime to 0s, then on any info update in backend the information will trigger an InfoChangeEvent, as long as information from actuator have changed.

[SteKoe]

In this example I have set lifetime to 0s and interval to 10s. As you can see, every ten seconds, the info is updated.

[cdprete]

InfoUpdater in backend is called in whatever spring.boot.admin.monitor.info-interval (default: 1m) is set to. This is mainly used to check every minute wether the instance is still up. In addition spring.boot.admin.monitor.info-lifetime (default: 1m) sets the lifetime of the last info data update. Since it is both set to 1m by default, you see a InfoChangedEvent just every 2nd minute.

If you set the lifetime to 0s, then on any info update in backend the information will trigger an InfoChangeEvent, as long as information from actuator have changed.

Good to know this, but it doesn't explain why there are gaps in the events.

[cdprete]

In this example I have set lifetime to 0s and interval to 10s. As you can see, every ten seconds, the info is updated.

Mmm, have you also tried with the same setup:

• fetch the instances through Eureka
• configure an embedded Hazelcast in each SBA instance
• have 2 SBA instances
  ?