Skip to main content

Long mergeout every day

  • February 28, 2018
  • 2 replies
  • 5 views

sergey_h
Forum|alt.badge.img+1

Hi guys,
I have problem witch mergeout in my vertica cluster. Once a day start big mergeout at the same time and working at 3-4 hours. I don't understand why it happens?

Statistic mergeout:

 DATE_TRUNC           min_ros_gb  max_ros_gb  sum_ros_gb  count_merg  
 -------------------  ----------  ----------  ----------  ----------  
 2018-02-28 13:00:00  42.72       43.12       171.96      4           
 2018-02-28 12:00:00  26.95       43.25       506.74      14          
 2018-02-28 11:00:00  27.01       27.24       162.85      6           
 2018-02-28 05:00:00  129.69      129.82      259.51      2           
 2018-02-28 04:00:00  129.53      129.79      1296.59     10      
 2018-02-28 03:00:00  84.69       84.92       1018.06     12
 2018-02-27 23:00:00  54.33       54.33       54.33       1           
 2018-02-27 22:00:00  54.31       54.61       326.70      6           
 2018-02-27 21:00:00  35.71       54.66       482.44      11          
 2018-02-27 20:00:00  36.00       36.21       216.71      6           
 2018-02-27 14:00:00  39.61       39.68       79.29       2           
 2018-02-27 13:00:00  39.54       39.76       158.68      4           
 2018-02-27 12:00:00  26.28       41.05       199.38      7                   

Every day at 3.00 - 6.00 running big mergeout. Objects are the same - two big flex table.
How can I fix / optimize it?

2 replies

caliskurahmet
  • New Participant
  • March 2, 2018

hey sergey , how's your tm resource pool values look like? Have you tried to increase the memorysize of tm pool?


sergey_h
Forum|alt.badge.img+1
  • Author
  • Participating Frequently
  • March 2, 2018

Hi, caliskurahmet
My tm pool:

-[ RECORD 1 ]-------+---------
pool_name           | tm
memory_size_kb      | 4194304
max_memory_size_kb  | 16777216
planned_concurrency | 4
max_concurrency     | 4
query_budget_kb     | 1048576

I has increased the memorysize to 8G - no effect.