The API get error 503 since HA deployment

[Alain]

Hello,

We use the API to retrieve the Probsum, Probsumc and list of clients from Prognosis Manager Node

Since I deployed the HA function in to the Manager Node the API we use get error 503 Service Unavailable Errors if the WebUI HA page is open.

I found in the log the HA do a API GET GetReplicationStatus every 5 sec. and this overload the API.

What is the API load supported and can we change the polling delays for the HA.

Best regards,

Alain,

[David_Sun]

Hi Alain,

The HTTP 503 error indicates the server is overloaded. Please check the post below about the possible solution. You might check if the server spec is up to the recommended specifications.

https://community.ir.com/t5/Transact-Infrastructure/Web-Interface-Error-code-503-Service-Temporarily...

There is another post about this error:

https://community.ir.com/t5/General/Service-Unavailable-error-Http-Error-503/m-p/11585

In addition, you can also check the HA heartbeat threshold timing configuration to see if it is set to short interval. Details can be found on Online Help:

HA Automatic Failover and Failback (prognosis.com)

If the above does not resolve the issue, please log a support case with irfax from Primary, Secondary and Management servers for our investigation.

Thanks

David 

[Alain · Accepted solution]

Hi David_Sun

Thank you to have pointed some points to check.

I have already identify the http 503 is for the server overloaded caused by the HA WebUI page open.

I know the VM server is under the spec recommendation.
Finally the Threshold HA Automatic Failover and Failback is not started.

I have a lot of database and script that run on that server and I found that something gradualy block the memmory available on the server.
After restarted Prognosis service the memory was released and error http 503 desapear.

Bestregards,

[David_Sun]

Hi @Alain ,

Thank you for the update, and good to know the server spec and the memory utilization may have played roles in this issue. 

THe first step to improve this would be to up the server specification, the following link provides details on how to check and resolve the server specification issue.

Solved: How to identify if you have incorrect core socket ... - IR Community

If after this server upgrade the problem still happens, it is likely the resource utilization issue that needs a closer look into the actual utilization of memory by processes. Please log a support case for our investigation. 

Regards

David