Star

Get neighbors failed报错问题

用户问答 使用问题
#1

nebula 版本：v1-nightly
nebula-console 版本：v1-nightly
部署方式（分布式 / 单机 / Docker / DBaaS）：Docker
硬盘：机械硬盘100g
内存：15g

Graph500测试数据，三度查询。
Graphd报错如下

E0209 09:10:05.939831 16 StorageClient.inl:123] Request to [:44500] failed: N6apache6thrift9transport19TTransportExceptionE: Timed Out
E0209 09:10:05.941138 29 ExecutionPlan.cpp:80] Execute failed: Get neighbors failed

storage报错如下：

E0209 08:21:17.252593 50 MetaClient.cpp:524] Send request to [:45500], exceed retry limit
E0209 08:21:17.252986 1 MetaClient.cpp:58] Heartbeat failed, status:RPC failure in MetaClient: N6apache6thrift9transport19TTransportExceptionE: AsyncSocketException: connect failed, type = Socket not open, errno = 111 (Connection refused): Connection refused
报错后storage 还活着，没挂

#2

修改一下，不是集群部署，是单节点部署

#3

你好，能提供下完整的graph和storage日志吗？以及看一下运行query后show hosts的截图。
如果服务都ok，看graph中报的错误是storage那边请求超时了，graphd向storaged请求的超时时间是一分钟，你可以在graphd的配置文件里面增加， 修改完配置文件需要重启graphd服务

--storage_client_timeout_ms=120

Ref: storaged超时导致Get neighbors failed?

#4

公司安全问题，不能上传文件。我提供的是error日志，请问你需要的是info的日志吗

#5

是的，粘贴上来就行，最好用论坛内置的markdown format一下，这样看起来比较方便

#6

我刚刚按照你的方法改了配置=180，运行报错如下。
show hosts:
image
Graphd
Error

引用
E0210 05:32:28.760882 15 MetaClient.cpp:524] Send request to [172.28.1.1:45500], exceed retry limit
E0210 05:32:28.761380 1 MetaClient.cpp:58] Heartbeat failed, status:RPC failure in MetaClient: N6apache6thrift9transport19TTransportExceptionE: AsyncSocketException: connect failed, type = Socket not open, errno = 111 (Connection refused): Connection refused

Info

引用
Log file created at: 2021/02/10 05:39:35
Running on machine: 54e7200cee9a
Log line format: [IWEF]mmdd hh:mm:ss.uuuuuu threadid file:line] msg
I0210 05:39:35.698618 1 GraphDaemon.cpp:101] Starting Graph HTTP Service
I0210 05:39:35.729967 9 WebService.cpp:142] Web service started on HTTP[13000], HTTP2[13002]
I0210 05:39:35.730114 1 GraphDaemon.cpp:115] Number of networking IO threads: 4
I0210 05:39:35.730149 1 GraphDaemon.cpp:124] Number of worker threads: 4
I0210 05:39:35.730522 1 MetaClient.cpp:44] Create meta client to [172.28.1.1:45500]
I0210 05:39:35.733669 1 StatsManager.cpp:112] registerHisto, bucketSize: 1000, min: 1, max: 1000000
I0210 05:39:35.735211 1 GflagsManager.cpp:125] Prepare to register 5 gflags to meta
I0210 05:39:35.757643 1 MetaClient.cpp:2198] Register gflags ok 5
I0210 05:39:35.758539 1 MetaClient.cpp:87] Register time task for heartbeat!
I0210 05:39:35.769598 1 StatsManager.cpp:112] registerHisto, bucketSize: 1000, min: 1, max: 1000000
I0210 05:39:35.769675 1 GraphDaemon.cpp:160] Starting nebula-graphd on 0.0.0.0:3699
I0210 05:40:24.653935 31 GraphService.cpp:50] Authenticating user user from 10.25.85.67:51988
I0210 05:40:28.208401 31 StatsManager.cpp:112] registerHisto, bucketSize: 1000, min: 1, max: 1000000
I0210 05:40:28.212683 31 StatsManager.cpp:112] registerHisto, bucketSize: 1000, min: 1, max: 1000000
I0210 05:40:28.217664 31 StatsManager.cpp:112] registerHisto, bucketSize: 1000, min: 1, max: 1000000

Storage
Error

引用
E0210 05:32:28.929567 49 MetaClient.cpp:524] Send request to [172.28.1.1:45500], exceed retry limit
E0210 05:32:28.929922 1 MetaClient.cpp:58] Heartbeat failed, status:RPC failure in MetaClient: N6apache6thrift9transport19TTransportExceptionE: AsyncSocketException: connect failed, type = Socket not open, errno = 111 (Connection refused): Connection refused

Info 比较大 截取一部分

引用
53, term = 4, lastLogId = 509327, lastLogTerm = 3, candidateIP = 172.28.2.1, candidatePort = 44501)
I0210 05:33:25.074435 67 RaftPart.cpp:1124] [Port: 44501, Space: 12, Part: 87] Sending out an election request (space = 12, part = 87, term = 4, lastLogId = 492575, lastLogTerm = 3, candidateIP = 172.28.2.1, candidatePort = 44501)
I0210 05:33:25.074498 67 RaftPart.cpp:1084] [Port: 44501, Space: 12, Part: 87] Partition is elected as the new leader for term 4
I0210 05:33:25.074470 66 RaftPart.cpp:1084] [Port: 44501, Space: 12, Part: 53] Partition is elected as the new leader for term 4
I0210 05:33:25.074532 67 RaftPart.cpp:1179] [Port: 44501, Space: 12, Part: 87] The partition is elected as the leader
I0210 05:33:25.074606 67 InMemoryLogBuffer.h:23] [Port: 44501, Space: 12, Part: 87] InMemoryLogBuffer ctor, firstLogId 492576
I0210 05:33:25.074570 66 RaftPart.cpp:1179] [Port: 44501, Space: 12, Part: 53] The partition is elected as the leader
I0210 05:33:25.074646 67 RaftPart.cpp:991] [Port: 44501, Space: 12, Part: 99] Start leader election, reason: lastMsgDur 33559, term 3
I0210 05:33:25.074674 67 RaftPart.cpp:1124] [Port: 44501, Space: 12, Part: 99] Sending out an election request (space = 12, part = 99, term = 4, lastLogId = 507561, lastLogTerm = 3, candidateIP = 172.28.2.1, candidatePort = 44501)
I0210 05:33:25.074689 66 InMemoryLogBuffer.h:23] [Port: 44501, Space: 12, Part: 53] InMemoryLogBuffer ctor, firstLogId 509328
I0210 05:33:25.074695 67 RaftPart.cpp:1084] [Port: 44501, Space: 12, Part: 99] Partition is elected as the new leader for term 4
I0210 05:33:25.074760 67 RaftPart.cpp:1179] [Port: 44501, Space: 12, Part: 99] The partition is elected as the leader
I0210 05:33:25.074800 67 InMemoryLogBuffer.h:23] [Port: 44501, Space: 12, Part: 99] InMemoryLogBuffer ctor, firstLogId 507562
I0210 05:33:25.074206 68 InMemoryLogBuffer.h:23] [Port: 44501, Space: 12, Part: 98] InMemoryLogBuffer ctor, firstLogId 494451

#7

请问这个问题稳定复现吗？如果是两度查询还有报错吗？

#8

两度查询没有报错，测试节点10个，运行到3301167 节点的时候一直报错。查询如下
GO 1 TO 3 STEPS FROM 3301167 OVER ed WHERE ed._dst != 3301167 YIELD ed._dst AS vid | GROUP BY $-.vid YIELD 1 AS id | GROUP BY $-.id YIELD COUNT(id)
返回结果应该是 2022448，前两个成功的结果大概1550000左右。

#9

我是通过接口调用的，需不需要把该查询直接在console运行一下呢

#10

由于我们曾经用过graph500测试的，按理3度查询是没有问题的，感觉主要原因还是因为使用HHD，你可以试试按照 System Requirement - Nebula Graph Database Manual 先设置下参数。

Nebula在设计之初就是针对SSD的，HDD并没有做过大量的测试有时会存在一些问题。

#11

好的，非常感谢，还想请问下，目前最新版本支持大概百分之多少的Cypher查询呢。

#12

这个比较难给出一个量化，相关开发还在进行中，也是我们开发的一个重点。目前2.0 RC1 已经支持了 大部分表达式和match，欢迎试用 :handshake:。

也欢迎提出你的需求，pm小姐姐将会对其评审。

#14

docker> ./scripts/nebula.service restart graphd

#15

我换了ssd,之前部署了2.0，现在准备部署1.0的时候报错了。。。
vesoft/nebula-storaged

image
image903×97 13.9 KB

这个之前一直都是好的，yaml文件用的也是之前的，不知道为什么这次就没有cluster.id了。

麻烦帮忙看下，谢谢

#16

@Aiee

#17

这个我开了一个新帖子问了，这里不用回复啦，谢谢

1赞

浙ICP备20010487号