Hi.
Do you think you could record a full network trace of the flow where the issue happens and share with us as a HAR file? So it will be the first attempt with the huge delay you see, and a few subsequent attempts with no delays, we need both results for comparison, you can put them into a single HAR.
You can use steps listed [here](https://www.inflectra.com/support/knowledgebase/kb254.aspx) - please use Firefox for that task, Chrome's HARs are flawed. Also don't forget to set "Persist log" and "Disable cache" checkboxes in the console to save everything, not just the recently loaded page.
Before proceeding with this, please also increase verbosity for IDP's logs by editing "Logging level shortcuts" section in `/opt/shibboleth-idp/conf/logback.xml` file inside `oxshibboleth`container like this (preserve the original file to roll-back the changes later):
```
<variable name="idp.loglevel.idp" value="${idp.loglevel.idp:-DEBUG}" />
<variable name="idp.loglevel.ldap" value="${idp.loglevel.ldap:-DEBUG}" />
<variable name="idp.loglevel.messages" value="${idp.loglevel.messages:-DEBUG}" />
<variable name="idp.loglevel.encryption" value="${idp.loglevel.encryption:-INFO}" />
<variable name="idp.loglevel.opensaml" value="${idp.loglevel.opensaml:-DEBUG}" />
<variable name="idp.loglevel.props" value="${idp.loglevel.props:-DEBUG}" />
<variable name="idp.loglevel.httpclient" value="${idp.loglevel.httpclient:-DEBUG}" />
```
Tail the log inside `oxshibboleth`container for some time to make sure changes are applied (some DEBUG entries are appearing): `# tail -F /opt/shibboleth-idp/logs/idp-process.log`
When it's applied, collect the traces (HAR file) and the `idp-process.log` related to the same time period, and share with us.