No. My client is python …direct… I can get the same behavior via curl.
I current belief is that the volumes attached to the dgraph-server instances are somehow in an inconsistent state. I think this behavior has happened because the server instances ran out of memory during inserts.
I change the backing nodes of the k8s cluster to have more memory and increased the requested memory for each pod. While things were more stable, the server still degraded over time.
I am currently recreating the cluster and database contents from scratch to see if this can be reproduced.
It would be great to understand how to debug the source of “Unable to find log file” errors. I saw nothing in the server or zero logs that were traceable to that response to the client. In fact, the log files were silent.