On Thu, Oct 24, 2019 at 9:14 AM Chris Kilding
<
chris+...@chriskilding.com> wrote:
> is the master being overwhelmed by either the sheer number of steps as they enter the job queue, or the parallelism of those steps when they execute
Quite possibly. The server sets a cap on the number of active elastic
agents, which is fine, but pushing up against that limit seems to
actually cause things to explode, rather than tasks simply waiting
politely in the queue. No one with sufficient access has (to my
knowledge) yet tried to track down whether the problem is in Jenkins
core, Pipeline, the Azure agent plugin we use, or something else.
https://github.com/jenkinsci/bom/blob/f384f08edeb3da9a2ba2c5b83be3fee4b5e0c628/Jenkinsfile#L40-L43