data_dist_detect.py

import requests
import json
# detect table status about data

db = 'disk_test'
ns_leader = '172.24.4.27:7123'

url = f'http://{ns_leader}/NameServer/ShowTable'
res = requests.get(url, json={"show_all": True})
tables = json.loads(res.text)['table_info']

tablet2partition = {}
tablet2count = {}
tablet2mem = {}

def collect(parts, field):
    dist = {}
    for part in parts:
        for replica in part['partition_meta']:
            if replica['endpoint'] not in dist:
                dist[replica['endpoint']] = 0
            dist[replica['endpoint']] += replica[field] if field else 1
    return dist

def add_merge(dist, dist2):
    for key, value in dist2.items():
        dist[key] = dist.get(key, 0) + value
    return dist

for table in tables:
    if db:
        if table['db'] != db:
            continue
    print(table['db'], table['name'])
    parts = table['table_partition']
    part_dist = collect(parts,'')
    print('partition dist(include replica)', part_dist)
    print('partition size:', len(parts))
    count_dist = collect(parts, 'record_cnt')
    mem_dist = collect(parts, 'record_byte_size')
    print('record count dist(include replica)', count_dist)
    print('mem dist(include replica)', mem_dist)
    add_merge(tablet2partition, part_dist)
    add_merge(tablet2count, count_dist)
    add_merge(tablet2mem, mem_dist)

print('total')
print('tablet2partition', tablet2partition)
print('tablet2count', tablet2count)
print('tablet2mem', tablet2mem)