猪抛出不兼容的类型错误答案

【问题标题】：Pig throwing incompatible type error猪抛出不兼容的类型错误
【发布时间】：2015-03-27 10:26:23
【问题描述】：

我正在使用以下代码通过在 datafu 中使用 sessionize UDF 在 pig 中生成 sessionId。

 SET       mapred.min.split.size 1073741824   
SET       mapred.job.queue.name 'marathon'
 SET       mapred.output.compress true;
--SET       avro.output.codec snappy;
--SET       pig.maxCombinedSplitSize 536870912;



page_view_pre = LOAD '/data/tracking/PageViewEvent/' USING LiAvroStorage('date.range','start.date=20150226;end.date=20150226;error.on.missing=true');  -----logic is currently for 2015-02-26,will later replace them with date parameters
p_key = LOAD '/projects/dwh/dwh_dim/dim_page_key/#LATEST' USING LiAvroStorage();


page_view_pre = FILTER page_view_pre  BY (requestHeader.userAgent != 'CRAWLER' and requestHeader.browserId != 'CRAWLER') and NOT IsTestMemberId(header.memberId);


page_view_pre = FOREACH page_view_pre GENERATE 
            (int)   (header.memberId <0 ? -9 : header.memberId )           as member_sk,
            (chararray)  requestHeader.browserId                               as browserId,
    --(chararray)       requestHeader.sessionId                                    as sessionId,
(chararray)         UnixToISO(header.time)                                     as pageViewTime,
            header.time                                                    as pv_time,
    (chararray) requestHeader.path                                 as path,
    (chararray)     requestHeader.referer                                  as referer,
    (chararray)  epochToFormat(header.time, 'yyyyMMdd', 'America/Los_Angeles') as tracking_date,
(chararray)         requestHeader.pageKey                                      as pageKey,
    (chararray)        SUBSTRING(requestHeader.trackingCode, 0, 500)           as trackingCode,  
        FLATTEN(botLookup(requestHeader.userAgent, requestHeader.browserId))   as (is_crawler, crawler_type),
    (int)   totalTime                                                                  as totalTime,
    ((int) totalTime < 20 ? 1 :0)                                                    as bounce_flag;    

page_view_pre = FILTER page_view_pre BY is_crawler == 'N' ;

p_key = FILTER p_key By is_aggregate ==1;

page_view_agg = JOIN page_view_pre by pageKey ,p_key by page_key;

page_view_agg = FOREACH page_view_agg GENERATE
                 (chararray)page_view_pre::member_sk as member_sk,
                 (chararray)page_view_pre::browserId as browserId,
                --page_view_pre::sessionId as sessionId,
                (chararray)page_view_pre::pageViewTime as pageViewTime,
                (long)page_view_pre::pv_time as pv_time,
                (chararray)page_view_pre::tracking_date as tracking_date,
                (chararray)page_view_pre::path as path,
                (chararray)page_view_pre::referer as referer,
                (chararray)page_view_pre::pageKey as pageKey,
                (int)p_key::page_key_sk as page_key_sk,
                (chararray)page_view_pre::trackingCode as trackingCode,
                (int)page_view_pre::totalTime as totalTime,
                (int)page_view_pre::bounce_flag as bounce_flag;

page_view_agg = FILTER page_view_agg By (member_sk is NOT null) OR (browserId IS NOT NULL) ;        

pvs_by_member_browser_pair = GROUP page_view_agg BY (member_sk,browserId);



***session_groups = FOREACH pvs_by_member_browser_pair {
                                visits = ORDER page_view_agg BY pv_time;
                                GENERATE FLATTEN(Sessionize(visits)) AS (
                                pageViewTime,member_sk, pv_time,tracking_date, pageKey,page_key_sk,browserId,referer ,path, trackingCode,totalTime, sessionId
                                                                        );
                                                    }***

粗体部分给我以下错误：

错误 1031：不兼容的架构：左边是“pageViewTime:NULL,member_sk:NULL,pv_time:NULL,tracking_date:NULL,pageKey:NULL,page_key_sk:NULL,browserId:NULL,referer:NULL,path:NULL,trackingCode: NULL,totalTime:NULL,sessionId:NULL”，右边是“datafu.pig.sessions.sessionize_visits_43::member_sk:chararray,datafu.pig.sessions.sessionize_visits_43::browserId:chararray,datafu.pig.sessions.sessionize_visits_43::pageViewTime :chararray,datafu.pig.sessions.sessionize_visits_43::pv_time:long,datafu.pig.sessions.sessionize_visits_43::tracking_date:chararray,datafu.pig.sessions.sessionize_visits_43::path:chararray,datafu.pig.sessions.sessionize_visits_43: :referer:chararray,datafu.pig.sessions.sessionize_visits_43::pageKey:chararray,datafu.pig.sessions.sessionize_visits_43::page_key_sk:int,datafu.pig.sessions.sessionize_visits_43::trackingCode:chararray,datafu.pig.sessions。 sessionize_visits_43::totalTime:int,datafu.pig.sessions.sessionize_visits_43::bounce_flag:int,datafu.pig.sessions.sessionize_visits_ 43::session_id:chararray"

我最初认为这与 null 成员或浏览器 ID 有关。我也对其进行了过滤，但错误仍然存在。我已经在这里停留了几个小时。非常感谢一些指针或解决方案来解决这个问题。

谢谢

【问题讨论】：

标签： hadoop hive apache-pig bigdata

【解决方案1】：

这是 Schema 不匹配的经典案例：

page_view_pre = LOAD '/data/tracking/PageViewEvent/' USING LiAvroStorage('date.range','start.date=20150226;end.date=20150226;error.on.missing=true');  -----logic is currently for 2015-02-26,will later replace them with date parameters

只需在此行后面加上illustrate page_view_pre 即可找出架构。

【讨论】：