One of our use cases make a lot (~30) lookups in parallel, and it seems that they either all succeed or all timeout depending on how we tweek the timeout for the ExtendedResolver. We use a 5 second timeout as the default, but can make the lookups succeed if we raise it a bit.
Our theory is that the lookups are resolved in the ForkJoin common pool, which is also busy with other work so all lookups will take a bit longer to complete, making them all fall outside the timeout limit.
Should we just raise the timeout, or could the behavior of dnsjava be improved?
We read #206 but were not sure if that would affect the behavior of multiple unrelated lookups or not.
We have a reproducer just using dnsjava which I will include.
One of our use cases make a lot (~30) lookups in parallel, and it seems that they either all succeed or all timeout depending on how we tweek the timeout for the ExtendedResolver. We use a 5 second timeout as the default, but can make the lookups succeed if we raise it a bit.
Our theory is that the lookups are resolved in the ForkJoin common pool, which is also busy with other work so all lookups will take a bit longer to complete, making them all fall outside the timeout limit.
Should we just raise the timeout, or could the behavior of dnsjava be improved?
We read #206 but were not sure if that would affect the behavior of multiple unrelated lookups or not.
We have a reproducer just using dnsjava which I will include.