These messages got logged in the serverless logs starting on 9/22:
[WorkloadAggregator]: TypeError: Cannot read properties of undefined (reading 'resurrectTimeout')
Failed to load list of active kibana nodes: Cannot read properties of undefined (reading 'resurrectTimeout')
Failed to poll for work: Cannot read properties of undefined (reading 'resurrectTimeout')
That field is part of the ES library we use directly and indirectly (via SO APIs) in Task Manager. Not clear if maybe this is a bug in that code, or perhaps we are not dealing with our clients correctly.
The error hasn't been seen in ECH yet.
Doing a little searching, seems like this is from the @elastic/transport package, so I'll open an issue there as well. Guessing this started happening due to the mget task poller changes and frequency of the ES calls that are made by the pollers.
These messages got logged in the serverless logs starting on 9/22:
That field is part of the ES library we use directly and indirectly (via SO APIs) in Task Manager. Not clear if maybe this is a bug in that code, or perhaps we are not dealing with our clients correctly.
The error hasn't been seen in ECH yet.
Doing a little searching, seems like this is from the
@elastic/transport
package, so I'll open an issue there as well. Guessing this started happening due to the mget task poller changes and frequency of the ES calls that are made by the pollers.