I have a running phoenix table with index.

when too much data write in the same time, index write can not catch up, 
finally cause a regionserver down.


Then I dropped index, and restart region server.


Here comes the problem:
regionserver will recover edits, and found index data, it tried to replay these 
index writes,
but since index is dropped, it can not replay updates, a dead loop.


May be replay log can be improved to handle case that index table is gone?






jinzhuan

Reply via email to