[Bug] FE start failed on centos

[chenjienn]

Search before asking

• [X] I had searched in the issues and found no similar issues.

Version

2.0-beta

What's Wrong?

按照部署文档，通过./bin/start_fe.sh --daemon命令启动后，在fe.log里先是报了下面的错：

Failed to init stats cleaner java.lang.RuntimeException: Table: internal.default_cluster:__internal_schema.column_statistics not exists at org.apache.doris.statistics.util.StatisticsUtil.findTable(StatisticsUtil.java:329) ~[doris-fe.jar:1.2-SNAPSHOT] at org.apache.doris.statistics.StatisticsCleaner.init(StatisticsCleaner.java:104) ~[doris-fe.jar:1.2-SNAPSHOT] at org.apache.doris.statistics.StatisticsCleaner.clear(StatisticsCleaner.java:82) ~[doris-fe.jar:1.2-SNAPSHOT] at org.apache.doris.statistics.StatisticsCleaner.runAfterCatalogReady(StatisticsCleaner.java:78) ~[doris-fe.jar:1.2-SNAPSHOT] at org.apache.doris.common.util.MasterDaemon.runOneCycle(MasterDaemon.java:58) ~[doris-fe.jar:1.2-SNAPSHOT] at org.apache.doris.common.util.Daemon.run(Daemon.java:116) ~[doris-fe.jar:1.2-SNAPSHOT]

然后就一直重复报下面的错：

Statistics storage initiated failed, will try again later org.apache.doris.common.DdlException: errCode = 2, detailMessage = System has no available disk capacity or no available BE nodes at org.apache.doris.system.SystemInfoService.checkAvailableCapacity(SystemInfoService.java:815) ~[doris-fe.jar:1.2-SNAPSHOT] at org.apache.doris.datasource.InternalCatalog.createTable(InternalCatalog.java:1072) ~[doris-fe.jar:1.2-SNAPSHOT] at org.apache.doris.catalog.InternalSchemaInitializer.createTbl(InternalSchemaInitializer.java:81) ~[doris-fe.jar:1.2-SNAPSHOT] at org.apache.doris.catalog.InternalSchemaInitializer.run(InternalSchemaInitializer.java:72) ~[doris-fe.jar:1.2-SNAPSHOT]

磁盘空间还剩40多G，BE还没启动，用datagrip连接fe的mysql，报上面的错，连不上

What You Expected?

通过./bin/start_fe.sh --daemon命令能正常启动

How to Reproduce?

No response

Anything Else?

No response

Are you willing to submit PR?

• [ ] Yes I am willing to submit a PR!

Code of Conduct

• [X] I agree to follow this project's Code of Conduct

[tbwtd]

我目前2.01版本也是这个报错 这个不是bug 是配置问题吗？

[zzccctv]

我目前也是这个报错

[sqnczzl]

目前我也遇到同样的问题，怎么解决？

[zzccctv]

问题已经解决了，需要修改fe.conf中的priority_networks = 这个配置，改成机器的实际网段

[sqnczzl]

fe,conf已经配置上priority_networks =，并且这个已经是机器的实际网段地址，但是fe 仍然是启动不起来

[zzccctv]

配置好之后，启动be，异常就没有了。

[shaotianfeng]

我也是这样（fe,conf已经配置上priority_networks =，并且这个已经是机器的实际网段地址，但是fe 仍然是启动不起来），be可以启动起来，但fe启动不起来，不知道哪里出了问题， 相同配置1.2.6版本没问题

[zzccctv]

一样的错，改网段后，启动fe然后通过mysql进行be配置,最后启动be就可以了.

[shaotianfeng]

网段也改了，是之前1.2.6版本的网段，然后起不起来。。。

[zzccctv]

配置完成后删除doris-meta试试

[khan-lau]

我目前2.01版本也是这个报错 这个不是bug 是配置问题吗？

多网卡指定 priority_networks 能确保线路正确; 如果是fe, 则需删除 doris-meta目录后重启

[xinghalo321]

2023-12-11 11:32:32,699 INFO (Thread-48|97) [InternalCatalog.createDb():423] create database[default_cluster:__internal_schema] which already exists
2023-12-11 11:32:32,700 WARN (Thread-48|97) [InternalSchemaInitializer.run():76] Statistics storage initiated failed, will try again later
org.apache.doris.common.DdlException: errCode = 2, detailMessage = System has no available disk capacity or no available BE nodes
    at org.apache.doris.system.SystemInfoService.checkAvailableCapacity(SystemInfoService.java:817) ~[doris-fe.jar:1.2-SNAPSHOT]
    at org.apache.doris.datasource.InternalCatalog.createTable(InternalCatalog.java:1085) ~[doris-fe.jar:1.2-SNAPSHOT]
    at org.apache.doris.catalog.InternalSchemaInitializer.createTbl(InternalSchemaInitializer.java:130) ~[doris-fe.jar:1.2-SNAPSHOT]
    at org.apache.doris.catalog.InternalSchemaInitializer.run(InternalSchemaInitializer.java:74) ~[doris-fe.jar:1.2-SNAPSHOT]

配置了priority_networks，也删除了doris-meta，还是不行；甚至重装了主机都无法修复.... 是不是ip_v6的网段识别有问题： priority_networks = FDBE:DC05:3::/48

[s-luping]

[s-luping]

[s-luping]

需要把BE节点添加进去才行

[TanLingxiao]

The right way to start fe is

1. ./bin/start_fe.sh --helper xx.xx.xx.xx:9010 --daemon
2. ALTER SYSTEM ADD FOLLOWER "xx.xx.xx.xx:9010";

Do not use ./bin/start_fe.sh to start it directly.

document

[baimafeima-yf]

fe,conf已经配置上priority_networks =，并且这个已经是机器的实际网段地址，但是fe 仍然是启动不起来

你好 你解决了吗，我也遇到了这个问题 2.0.11版本

[baimafeima-yf]

2023-12-11 11:32:32,699 INFO (Thread-48|97) [InternalCatalog.createDb():423] create database[default_cluster:__internal_schema] which already exists
2023-12-11 11:32:32,700 WARN (Thread-48|97) [InternalSchemaInitializer.run():76] Statistics storage initiated failed, will try again later
org.apache.doris.common.DdlException: errCode = 2, detailMessage = System has no available disk capacity or no available BE nodes
  at org.apache.doris.system.SystemInfoService.checkAvailableCapacity(SystemInfoService.java:817) ~[doris-fe.jar:1.2-SNAPSHOT]
  at org.apache.doris.datasource.InternalCatalog.createTable(InternalCatalog.java:1085) ~[doris-fe.jar:1.2-SNAPSHOT]
  at org.apache.doris.catalog.InternalSchemaInitializer.createTbl(InternalSchemaInitializer.java:130) ~[doris-fe.jar:1.2-SNAPSHOT]
  at org.apache.doris.catalog.InternalSchemaInitializer.run(InternalSchemaInitializer.java:74) ~[doris-fe.jar:1.2-SNAPSHOT]

配置了priority_networks，也删除了doris-meta，还是不行；甚至重装了主机都无法修复.... 是不是ip_v6的网段识别有问题： priority_networks = FDBE:DC05:3::/48

您好 ，您解决了吗 ，我也是一样的问题，配置了priority_networks，也删除了doris-meta，还是不行

[AiTian1211]

你们其他FE节点第一次启动时是不是没有指定第一个FE节点？./fe/bin/start_fe.sh --helper 10.10.10.16:9010 --daemon 我也遇到这个问题，但我是其他节点第一次启动时忘记指定第一个启动的FE节点了，把doris-meta目录删除后通过./fe/bin/start_fe.sh --helper 10.10.10.16:9010 --daemon启动就解决了