How to resolve the Memory Spikes in Dataverse?

[Mansoor Koormath]

Dear Team,

I want to resolve the memory spikes in Dataverse. The server has 16 GB of RAM, and it shows 12 GB of Memory used.  As per the ChatGPT suggestion, I have to replace the 

asadmin delete-jvm-options -Xmx7765m with the lines 

asadmin create-jvm-options "-Xms6g"
asadmin create-jvm-options "-Xmx6g"

Could you please let me know whether I can proceed with these settings? Are there any risks associated with making these changes to the settings?

I am looking forward to hearing from you. 

Regards

Mansoor Koormath

Digital Library Services

Indian Institute of Management Bangalore

[Don Sizemore]

Hello,

You should be fine with those settings, but the risk is that you'll attempt to perform some action (or your server may find itself under heavy load) and you'll run out of heap space.

The end-to-end test suite fails in a JVM heap smaller than 8GB, and Dataverse won't even deploy in fewer than 4GB.

I hope this helps,

Don

--
You received this message because you are subscribed to the Google Groups "Dataverse Users Community" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dataverse-commu...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/dataverse-community/CAC8Un-3geUc2RmxRD9%2BDJhT9KqpYUT-HboY5mvO5JVuCFpZ-vA%40mail.gmail.com.

[Mansoor Koormath]

Dear Don,

  Thank you for your response. 🙂

Regards

Mansoor Koormath

To view this discussion visit https://groups.google.com/d/msgid/dataverse-community/CAPfMOayEWvHQzy6cMC8eSq2F9O-EgNtAT1r8Ea0PSv0sR0FOSg%40mail.gmail.com.

[o.be...@fz-juelich.de]

Hi there!

In addendum to what Don said above: while I am able to deploy Dataverse in less than 4GB of RAM for development purposes, for any production scenario using only 4GB of heap memory is cutting it very close!

In case you want to keep an eye on actual heap usage, please have a look at the metrics exposed by Payara. You can find the docs here: https://docs.payara.fish/community/docs/6.2025.10/Technical%20Documentation/MicroProfile/Metrics/Metrics%20Rest%20Endpoint.html You will need something like Prometheus or similar tools to keep an inventory of the statistics, which will allow you to gather intel how much heap is actually in use.

You may also want to look into tuning your GC setup.

On a related note, but you already seem to have a sense of distrust (good!): please be very careful in how you ask ChatGPT questions with regards to running a server. I found on multiple occasions that it will have hallucinations about this kind of stuff, very much depending on how you ask, what you provide as context, etc. Getting hard evidence by measuring is key here - as always in ops.

Cheers,
Oliver

[Mansoor Koormath]

Dear Mr Oliver,

Thank you for your comments. I got some misleading information from ChatGPT before. Therefore, I am not always trusting ChatGPT's answers as solutions. Hence, I posted my query in this community and sought advice from the experts. Thank you very much to you and Mr Don for your valuable response.

Regards

Mansoor Koormath

Digital Library Services

Indian Institute of Management Bangalore

To view this discussion visit https://groups.google.com/d/msgid/dataverse-community/52f58dea-9c6b-404a-b1ff-bd3462aa50efn%40googlegroups.com.