We found some curious cases in TD 14.10. The DB has 3 active nodes and 1 HSN. DBA switched one active node and HSN in Feb, after swithced:
1. In view point, it shows one node down. but the query and ETL job can, should be not actually down. If it's issue of view point?
2. After several days laer, the daily loading table size ( refer to dbc.tablesize ) is larger day by day, e.g., day is 6G, day 2 is 7 G..., but we ensure the real data size is almost the same in these days. This just happened in some table and some AMPs. Then encountered a DB auto restart later, DBA perform all table scan, then the size become normal again. Why this happened?
3. We checked log file in /var/log/sa, found the number of packets per second reduced by 80%. If the switching will change packet size?