弹性搜索批量索引超时错误!错误:3000ms后请求超时 [英] Elastic search bulk index timeout err! Error: Request Timeout after 30000ms
问题描述
最近,我想将旧的索引数据滚动到新的基于月的索引。存储的数据从2015/07至今为止。每个月几乎有3万条记录。按照滚动和批量方法/api-reference-2-2.html#api-scroll-2-2rel =nofollow> 2.2 API ,我完成了如下代码。
For recently, I want to scroll through the old index data to new monthly-based indices. The stored data begin from 2015/07 until now. and it is almost 30,000 records for every month. Follow the scroll and bulk methods provided in 2.2 API, I finish the code as follows.
file main.coffee
logger = require 'graceful-logger'
elasticsearch = require 'elasticsearch'
setMonthlyIndices = require './es-test-promise'
client = new elasticsearch.Client
host:
host: 'localhost'
port: 9200
protocol: 'http'
setMonthlyIndices client, 'es_test_messages', 'talk_messages_v2', 'messages', 2015, 6
文件 es-test-promise.coffee
logger = require 'graceful-logger'
elasticsearch = require 'elasticsearch'
config = require 'config'
setIndice = (client, prefix, index, type, year, month) ->
allDocs = []
count = 0
prevYear = year + ''
# with leading '0' for month less than 10
prevMonth = ("0" + month).slice(-2)
nextDate = new Date(year, month)
nextYear = nextDate.getFullYear().toString()
nextMonth = ("0" + (nextDate.getMonth()+1)).slice(-2)
minDate = "#{prevYear}-#{prevMonth}-01"
maxDate = "#{nextYear}-#{nextMonth}-01"
indice_name = "#{prefix}_#{prevYear}_#{prevMonth}"
q =
filtered:
filter:
range:
createdAt:
gte: minDate
lt: maxDate
format: "yyyy-MM-dd"
client.search
index: index
type: type
scroll: '1m'
body:
query: q
sort: ['_doc']
size: 1000
, callback = (err, response) ->
console.log "indice_name 1", indice_name
return logger.err err.stack if err
return unless response.hits?.total
allDocs = []
response.hits.hits.forEach (hit)->
action =
index:
_id: hit._id
allDocs.push(action)
allDocs.push(hit._source)
count = count + allDocs.length
client.bulk
index: indice_name
type: type
body: allDocs
, (err, resp) ->
console.log "indice_name 2", indice_name
return logger.err err.stack if err
if response.hits.total *2 != count
client.scroll
scrollId: response._scroll_id
scroll: '1m'
, callback
else
logger.info "Finish indicing #{indice_name}"
setMonthlyIndices = (client, prefix, index, type, year, month) ->
current = new Date()
currentYear = current.getFullYear()
currentMonth = current.getMonth() + 1
processYear = year or currentYear
processMonth = month or 0
processDate = new Date(processYear, processMonth)
currentDate = new Date(currentYear, currentMonth)
processDate = new Date(2015, 6)
currentDate = new Date(2015, 9)
while processDate <= currentDate
year = processDate.getFullYear()
month = processDate.getMonth() + 1
setIndice(client, prefix, index, type, year, month)
processDate.setMonth(processDate.getMonth() + 1)
module.exports = setMonthlyIndices
我想知道是否由于打开太多的客户端请求,因为在文件 es-test-promise.coffee ,所有这些搜索请求都同时运行。这只是一个猜测,然后我也试图用承诺实现,以确保请求可以逐个执行。最后,我无法理解,放弃。
I am wondering whether it is due to open too many client request, because in file es-test-promise.coffee, all these search request is running simultaneously. This is just a guess, and then I have also tried to implement with promise to make sure the request could be executed one by one. Finally, I can't figure it out and give up.
你有什么建议,我认为应该是源码问题,但是我不知道检查...
Do you have any suggestion, I think it should be the source issues , but I don't know where to check...
推荐答案
只需将requestTimeout放入您的配置。
Just put the requestTimeout to your config.
例如:
new elasticsearch.Client({host:"localhost", requestTimeout : Infinity});
您可以将 Infinity
替换为您所需的限制在'ms'。
You can replace Infinity
by your desired limit in 'ms' .
这篇关于弹性搜索批量索引超时错误!错误:3000ms后请求超时的文章就介绍到这了,希望我们推荐的答案对大家有所帮助,也希望大家多多支持IT屋!