Upping defaults because helps our ITBLL runs on cluster go further.
HBASE-19359 changed the client retries. It made them 10 rather than 35. 10 doesn't seem to be enough to make it across a server crash when there are 40-odd WALs to split. Upping it until we do more work on MTTR.
When we hit the blocking store file limit we stop all writes for 90 seconds. Can happen a few times back-to-back. Kills the running ITBLL. This pause for 90 seconds needs reexamination. It was put in place a long time ago when things were very different.