I am getting a Broken pipe exception while invoking the Enhanced Smart Tags training workflow from my local 6.5 author instance. The smart tags workflow is working fine, however it is always throwing a broken pipe exception on my error logs whenever I invoke a training workflow..
The similarity search JMX configuration is all passed while invoking validateConfigs. Any thoughts why the socket/pipe is getting closed prematurely. Is it the organisation network that causes this issue.
Caused by: java.io.IOException: Broken pipe
at sun.nio.ch.FileDispatcherImpl.writev0(Native Method)
at org.eclipse.jetty.io.ChannelEndPoint.flush(ChannelEndPoint.java:264) [org.apache.felix.http.jetty:4.0.8]
I tried with multiple scenarios. By invoking on a folder with one image which is around 100KB. Also on a folder with 20-25 images similar sized images all images less than 250Kb
However the log says the below information as well.
03.12.2019 10:10:08.307 *INFO* [JobHandler: /var/workflow/instances/server0/2019-12-03_1/smart-tags-training_232:/content/dam/smart-tags/time] com.day.cq.dam.similaritysearch.internal.impl.TransferTrainingDataServiceImpl Successfully Completed Training Data Transfer for [/content/dam/smart-tags/time]
On subsequent run,
03.12.2019 12:39:32.665 *INFO* [JobHandler: /var/workflow/instances/server0/2019-12-03_1/smart-tags-training_233:/content/dam/smart-tags/time] com.day.cq.dam.similaritysearch.internal.impl.TransferTrainingDataServiceImpl Completed Training Data Transfer for [/content/dam/smart-tags/time]. No tags were transferred
But my smart tag training report is empty and the broken type issue is still there.