Hello all. We're dealing with some regular (~daily) client timeouts and resulting ingest job failures. Reviewing logs it all points to slow disks (hosts are getting old). Sadly the spark connector doesn't expose the defaultOperationTimeoutMs in AsyncKuduClient so we're stuck with 30s default unless we build a custom one.
So we are considering increasing the service queue from 50 to 100 or so to smooth out ingest spikes from different jobs. What would be the tradeoffs of doing this? thanks, -- Mauricio Aristizabal Architect - Data Pipeline https://impact.com <https://www.linkedin.com/company/impact-partech/> <https://www.facebook.com/ImpactParTech/> <https://twitter.com/impactpartech> <https://www.youtube.com/c/impactpartech>
