Error: unexpected error happend during sending data to cache server.

I have setup PCCS in k8s cluster on a non-SGX machine. Then I configure a SGX device to communicate to this PCCS service. I have also installed PCKIDRetrievalTool as mentioned [here](https://www.intel.com/content/www/us/en/developer/articles/guide/intel-software-guard-extensions-data-center-attestation-primitives-quick-install-guide.html).

Things are all good and healthy with the setup and I am able to run the PCKIDRetrievalTool. But as more and more nodes are added with time, we start seeing the PCKIDRetrievalTool fail. Also the node which had succeeded earlier starts to fail with the following log -

```
ubuntu@myserver40:/opt/intel/sgx-pck-id-retrieval-tool$ sudo PCKIDRetrievalToolIntel(R) Software Guard Extensions PCK Cert ID Retrieval Tool Version 1.12.101.1
Warning: platform manifest is not available or current platform is not multi-package platform.
Error: unexpected error happend during sending data to cache server.
pckid_retrieval.csv has been generated successfully, however the data couldn't be sent to cache server!
```

Corresponding log seen on the PCCS is -
```
2022-01-06 05:27:32.536 [info]: Client Request-ID : de5abc9c21634101a833657761ffe127
2022-01-06 05:27:33.103 [info]: Request-ID is : 450d6aae91664f39ba589027383854f2
2022-01-06 05:27:33.106 [error]: Error: No cache data for this platform.
{{ at Proxy.getPckCertFromPCS (/opt/intel/pccs/services/logic/commonCacheLogic.js:86:11)}}
{{ at runMicrotasks ()}}
{{ at processTicksAndRejections (internal/process/task_queues.js:95:5)}}
{{ at async LazyCachingMode.registerPlatforms (/opt/intel/pccs/services/caching_modes/cachingMode.js:163:7)}}
{{ at async Proxy.registerPlatforms (/opt/intel/pccs/services/platformsRegService.js:107:3)}}
{{ at async postPlatforms (/opt/intel/pccs/controllers/platformsController.js:40:5)}}
2022-01-06 05:27:33.111 [info]: 10.0.0.249 - - [06/Jan/2022:05:27:33 +0000] "POST /sgx/certification/v3/platforms HTTP/1.1" 404 32 "" ""
```
Resetting the SGX on the device solves the issue at times and the PCKIDRetrievalTool can be run again.
I do not understand what could happen with time (usually over 24-30 hours of the PCCS running) that causes this failure.

These are production machines and hence we are able to succeed with the appropriate API keys initially.

**EDIT:**
Worth mentioning, I do see a restart of the PCCS service post which we get this observation. Its not very clear what causes the restart but its the AWS instance (the non-SGX machine mentioned above) that restarts & not the PCCS service alone.