nebualGraph 2.5.1 和spark -exchange 2.5.1 导入数据直接让storaged,graphd 全部挂掉,重启立刻挂掉

nebula-storaged.WARNING (616 字节)
nebula-storaged.ERROR (507 字节)
nebula-storaged.INFO (407.7 KB)

改了级别日志还是一样的吗没看到其他日志信息

我发现问题了。是这边创建图shecma 就直接把graph,storage 搞挂了

啥意思?

我刚换其他集群重新测试了一下,我没导入数据只是先创建schema 结果服务就挂了,是我这边的创建语句太多了吗,2000 条语句
用的python client 2.0.0 部分语句如下
[CREATE TAG IF NOT EXISTS Thing();]
[2022-03-18 18:17:22,137][shell_script_operator.py][line:185][INFO] success to create schema, ngql is [ALTER TAG Thing ADD (Thing_name string DEFAULT “NULL”);]
[2022-03-18 18:17:22,138][shell_script_operator.py][line:185][INFO] success to create schema, ngql is [ALTER TAG Thing ADD (Thing_title string DEFAULT “NULL”);]
[2022-03-18 18:17:22,138][shell_script_operator.py][line:185][INFO] success to create schema, ngql is [ALTER TAG Thing ADD (Thing_nameCh string DEFAULT “NULL”);]
[2022-03-18 18:17:22,138][shell_script_operator.py][line:185][INFO] success to create schema, ngql is [ALTER TAG Thing ADD (Thing_nameEn string DEFAULT “NULL”);]
[2022-03-18 18:17:22,138][shell_script_operator.py][line:185][INFO] success to create schema, ngql is [ALTER TAG Thing ADD (Thing_abbreviation string DEFAULT “NULL”);]
[2022-03-18 18:17:22,138][shell_script_operator.py][line:185][INFO] success to create schema, ngql is [ALTER TAG Thing ADD (Thing_tag string DEFAULT “NULL”);]
[2022-03-18 18:17:22,139][shell_script_operator.py][line:185][INFO] success to create schema, ngql is [ALTER TAG Thing ADD (Thing_alias string DEFAULT “NULL”);]
[2022-03-18 18:17:22,139][shell_script_operator.py][line:185][INFO] success to create schema, ngql is [ALTER TAG Thing ADD (Thing_abstract string DEFAULT “NULL”);]
[2022-03-18 18:17:22,139][shell_script_operator.py][line:185][INFO] success to create schema, ngql is [ALTER TAG Thing ADD (Thing_image string DEFAULT “NULL”);]
[2022-03-18 18:17:22,139][shell_script_operator.py][line:185][INFO] success to create schema, ngql is [ALTER TAG Thing ADD (Thing_video string DEFAULT “NULL”);]
[2022-03-18 18:17:22,140][shell_script_operator.py][line:185][INFO] success to create schema, ngql is [ALTER TAG Thing ADD (Thing_audio string DEFAULT “NULL”);]
[2022-03-18 18:17:22,140][shell_script_operator.py][line:185][INFO] success to create schema, ngql is [ALTER TAG Thing ADD (Thing_gmtCreated string DEFAULT “NULL”);]
[2022-03-18 18:17:22,140][shell_script_operator.py][line:185][INFO] success to create schema, ngql is [ALTER TAG Thing ADD (Thing_gmtModified string DEFAULT “NULL”);]
[2022-03-18 18:17:22,140][shell_script_operator.py][line:185][INFO] success to create schema, ngql is [ALTER TAG Thing ADD (Thing_popularity double DEFAULT 0.0);]
[2022-03-18 18:17:22,140][shell_script_operator.py][line:185][INFO] success to create schema, ngql is [ALTER TAG Thing ADD (Thing_prior bool DEFAULT false);]
[2022-03-18 18:17:22,141][shell_script_operator.py][line:185][INFO] success to create schema, ngql is [ALTER TAG Thing ADD (Thing_dataSource string DEFAULT “NULL”);]
[2022-03-18 18:17:22,141][shell_script_operator.py][line:185][INFO] success to create schema, ngql is [ALTER TAG Thing ADD (Thing_urls string DEFAULT “NULL”);]

这个跟之前挂不一定是同个原因的,还需要分析日志的

这个我这边测试结果就是创建scheme 失败,服务挂了,导致服务再也启动不了了

我减少创建语句,就成功了,服务没挂

那个session参数你挂的节点上都改了吗?

改了

这边反复测试了2000 建图语句直接把服务搞挂,500个就没有挂,可以判断服务挂掉就是这个问题怎么,不是导入数据问题,这个要怎么处理

建图语句具体是啥?只是create space吗?

方便的话把测试的代码或脚本贴一下,谢谢。

日志级别改了以后日志会详细一些,所有日志的级别v都改成4,感觉日志还不是很详细,确认下改成4后重启麻烦再把日志发一下?

create_space_ngql.txt (126.7 KB)
就是用python client 循环执行这个文件里面的语句,执行完服务就搞挂了
client 版本
from nebula2.gclient.net import ConnectionPool
from nebula2.Config import Config

config = Config()
config.max_connection_pool_size = 10
connection_pool = ConnectionPool()
ok = connection_pool.init([(first_graph[0], int(first_graph[1]))], config)
client = connection_pool.get_session(nebula_user, nebula_password)
client.execute(“use " +nebula_space+”;")
with open(’create_space_ngql.txt ‘, ‘r’) as f:
ngqls = f.readlines()
for temp in ngqls:
ngql = temp.strip()
if len(ngql) != 0:
response = client.execute(ngql)
if response.is_succeeded() or str(response.error_msg())==“Existed!” :
print(“success to create schema, ngql is [” + str(ngql) + “]”)
else:
print("fail to create schema, response.error_code = " + str(response.is_succeeded()) + ", response.error_msg is " + str(response.error_msg()) + “, ngql is [” + str(ngql) + “]”)
exit(1)

老哥,你按照 @Henry 的回复,先把日志调整下,然后把调整好日志级别、重启之后的日志发给他来看看,方便定位下是否哪里有问题。

crashes after executing multiple alter tags (stable reproduction) · Issue #3889 · vesoft-inc/nebula · GitHub 这个问题是一个已知的issue,在v3.0版本中已经修复了,如果之前导数据也是这个语句,那这两个问题都是同一个原因导致的。另外这个创建逻辑建议改成直接create tag的时候将属性写进去以绕过这个bug,而不是后面用alter 增加

之前是create tag (xxx,xxx) ,但是现在有新增字段的需求,所以改成了这个alter ,现在暂时不升级3.0 等3.0 稳定了再升级
没有别的办法吗, 这边可以一个session少执行一些alter ,在relaese session ,再 重新获取一个session 再执行其他alter一批,这样可以吗

可以试试看,如果成功了来论坛反馈一下啊

我刚看了上面issue ,说是版本号达到256 了,这边alter 2000次,版本号是一下就超256 了吧,和重新创建session 没关系,估计不行