# Shutdown master means breakdown the cluster's service?
**URL:** https://discuss.elastic.co/t/shutdown-master-means-breakdown-the-clusters-service/10259
**Category:** Elasticsearch
**Created:** [January 8, 2013, 8:59am UTC](https://discuss.elastic.co/t/shutdown-master-means-breakdown-the-clusters-service/10259 "2013-01-08T08:59:14Z")
**Posts on this page:** 9
**Page:** 1
### Author:  [@asoqa](https://discuss.elastic.co/u/asoqa)
#### Post date: [January 8, 2013, 8:59am UTC](https://discuss.elastic.co/t/shutdown-master-means-breakdown-the-clusters-service/10259/1 "2013-01-08T08:59:14Z")
I have one cluster with two nodes. I used the curl command to shutdown the
master node:
$ curl -XPOST '[http://localhost:9200/\_cluster/nodes/\_master/\_shutdown](http://localhost:9200/_cluster/nodes/_master/_shutdown)'
I thought es should automatically choose a new master node to replace the
failed one. But it seems things does not like what I think.
--
---
### Author:  [@dadoonet](https://discuss.elastic.co/u/dadoonet)
#### Post date: [January 8, 2013, 9:15am UTC](https://discuss.elastic.co/t/shutdown-master-means-breakdown-the-clusters-service/10259/2 "2013-01-08T09:15:19Z")
Why do you think that a new master has not been elected?
Can you share your other nodes logs?
Le 8 janvier 2013 à 09:59, asoqa [asoqa51@gmail.com](mailto:asoqa51@gmail.com) a écrit :
> I have one cluster with two nodes. I used the curl command to shutdown the
> master node:
>
> $ curl -XPOST '[http://localhost:9200/\_cluster/nodes/\_master/\_shutdown](http://localhost:9200/_cluster/nodes/_master/_shutdown)'
>
> I thought es should automatically choose a new master node to replace the
> failed one. But it seems things does not like what I think.
>
> --
--
David Pilato
[http://www.scrutmydocs.org/](http://www.scrutmydocs.org/)
[http://dev.david.pilato.fr/](http://dev.david.pilato.fr/)
Twitter : @dadoonet / @elasticsearchfr / @scrutmydocs
--
---
### Author:  [@asoqa](https://discuss.elastic.co/u/asoqa)
#### Post date: [January 9, 2013, 1:46am UTC](https://discuss.elastic.co/t/shutdown-master-means-breakdown-the-clusters-service/10259/3 "2013-01-09T01:46:34Z")
the left node's log:
[2013-01-09 09:39:30,517][INFO][discovery.zen] [Quasimodo]
master\_left [[Plague][\_6-qEiuHQ76mjwFdTDF9eQ][inet[/10.232.42.205:9300]]],
reason [shut\_down]
[2013-01-09 09:39:30,518][WARN][discovery.zen] [Quasimodo] not
enough master nodes after master left (reason = shut\_down), current nodes:
{[Quasimodo][ts0Kq\_c2RoCFwa-UKFI8Tg][inet[/10.232.42.204:9300]],}
[2013-01-09 09:39:30,518][INFO][cluster.service] [Quasimodo]
removed {[Plague][\_6-qEiuHQ76mjwFdTDF9eQ][inet[/10.232.42.205:9300]],},
reason: zen-disco-master\_failed
([Plague][\_6-qEiuHQ76mjwFdTDF9eQ][inet[/10.232.42.205:9300]])
the master node's log
[2013-01-09 09:39:27,365][INFO][action.admin.cluster.node.shutdown]
[Plague] [partial\_cluster\_shutdown]: requested, shutting down
[[\_6-qEiuHQ76mjwFdTDF9eQ]] in [1s]
[2013-01-09 09:39:28,370][INFO][action.admin.cluster.node.shutdown]
[Plague] shutting down in [200ms]
[2013-01-09 09:39:28,372][INFO][action.admin.cluster.node.shutdown]
[Plague] [partial\_cluster\_shutdown]: done shutting down
[[\_6-qEiuHQ76mjwFdTDF9eQ]]
[2013-01-09 09:39:28,572][INFO][action.admin.cluster.node.shutdown]
[Plague] initiating requested shutdown (using service)
[2013-01-09 09:39:30,345][INFO][node] [Plague]
{0.19.11-SNAPSHOT}[16001]: stopping ...
[2013-01-09 09:39:30,531][INFO][node] [Plague]
{0.19.11-SNAPSHOT}[16001]: stopped
[2013-01-09 09:39:30,531][INFO][node] [Plague]
{0.19.11-SNAPSHOT}[16001]: closing ...
[2013-01-09 09:39:30,539][INFO][node] [Plague]
{0.19.11-SNAPSHOT}[16001]: closed
And I can not search the index, e.g. curl -XGET
localhost:9200/twitter/\_search
在 2013年1月8日星期二UTC+8下午5时15分19秒，David Pilato写道：
> Why do you think that a new master has not been elected?
> Can you share your other nodes logs?
>
> Le 8 janvier 2013 à 09:59, asoqa \<[aso...@gmail.com](mailto:aso...@gmail.com) \\> a
> écrit :
>
> I have one cluster with two nodes. I used the curl command to shutdown
> the master node:
>
> $ curl -XPOST '[http://localhost:9200/\_cluster/nodes/\_master/\_shutdown](http://localhost:9200/_cluster/nodes/_master/_shutdown)'
>
> I thought es should automatically choose a new master node to replace
> the failed one. But it seems things does not like what I think.
>
> --
>
> --
> David Pilato
> [http://www.scrutmydocs.org/](http://www.scrutmydocs.org/)
> [http://dev.david.pilato.fr/](http://dev.david.pilato.fr/)
> Twitter : @dadoonet / @elasticsearchfr / @scrutmydocs
--
---
### Author:  [@Igor\_Motov](https://discuss.elastic.co/u/Igor_Motov)
#### Post date: [January 9, 2013, 1:55am UTC](https://discuss.elastic.co/t/shutdown-master-means-breakdown-the-clusters-service/10259/4 "2013-01-09T01:55:50Z")
You, probably, have discovery.zen.minimum\_master\_nodes setting set to 2.
So, when the cluster losses the second master eligible node it doesn't
elect the only remaining node as a master to avoid split brain. You can
either remove this setting or add the third master eligible node to your
cluster.
On Tuesday, January 8, 2013 8:46:34 PM UTC-5, asoqa wrote:
> the left node's log:
> [2013-01-09 09:39:30,517][INFO][discovery.zen] [Quasimodo]
> master\_left [[Plague][\_6-qEiuHQ76mjwFdTDF9eQ][inet[/10.232.42.205:9300]]],
> reason [shut\_down]
> [2013-01-09 09:39:30,518][WARN][discovery.zen] [Quasimodo]
> not enough master nodes after master left (reason = shut\_down), current
> nodes: {[Quasimodo][ts0Kq\_c2RoCFwa-UKFI8Tg][inet[/10.232.42.204:9300]],}
> [2013-01-09 09:39:30,518][INFO][cluster.service] [Quasimodo]
> removed {[Plague][\_6-qEiuHQ76mjwFdTDF9eQ][inet[/10.232.42.205:9300]],},
> reason: zen-disco-master\_failed
> ([Plague][\_6-qEiuHQ76mjwFdTDF9eQ][inet[/10.232.42.205:9300]])
>
> the master node's log
> [2013-01-09 09:39:27,365][INFO][action.admin.cluster.node.shutdown]
> [Plague] [partial\_cluster\_shutdown]: requested, shutting down
> [[\_6-qEiuHQ76mjwFdTDF9eQ]] in [1s]
> [2013-01-09 09:39:28,370][INFO][action.admin.cluster.node.shutdown]
> [Plague] shutting down in [200ms]
> [2013-01-09 09:39:28,372][INFO][action.admin.cluster.node.shutdown]
> [Plague] [partial\_cluster\_shutdown]: done shutting down
> [[\_6-qEiuHQ76mjwFdTDF9eQ]]
> [2013-01-09 09:39:28,572][INFO][action.admin.cluster.node.shutdown]
> [Plague] initiating requested shutdown (using service)
> [2013-01-09 09:39:30,345][INFO][node] [Plague]
> {0.19.11-SNAPSHOT}[16001]: stopping ...
> [2013-01-09 09:39:30,531][INFO][node] [Plague]
> {0.19.11-SNAPSHOT}[16001]: stopped
> [2013-01-09 09:39:30,531][INFO][node] [Plague]
> {0.19.11-SNAPSHOT}[16001]: closing ...
> [2013-01-09 09:39:30,539][INFO][node] [Plague]
> {0.19.11-SNAPSHOT}[16001]: closed
>
> And I can not search the index, e.g. curl -XGET
> localhost:9200/twitter/\_search
>
> 在 2013年1月8日星期二UTC+8下午5时15分19秒，David Pilato写道：
>
> > Why do you think that a new master has not been elected?
> > Can you share your other nodes logs?
> >
> > Le 8 janvier 2013 à 09:59, asoqa [aso...@gmail.com](mailto:aso...@gmail.com) a écrit :
> >
> > I have one cluster with two nodes. I used the curl command to shutdown
> > the master node:
> >
> > $ curl -XPOST '[http://localhost:9200/\_cluster/nodes/\_master/\_shutdown](http://localhost:9200/_cluster/nodes/_master/_shutdown)'
> >
> > I thought es should automatically choose a new master node to replace
> > the failed one. But it seems things does not like what I think.
> >
> > --
> >
> > --
> > David Pilato
> > [http://www.scrutmydocs.org/](http://www.scrutmydocs.org/)
> > [http://dev.david.pilato.fr/](http://dev.david.pilato.fr/)
> > Twitter : @dadoonet / @elasticsearchfr / @scrutmydocs
--
---
### Author:  [@asoqa](https://discuss.elastic.co/u/asoqa)
#### Post date: [January 9, 2013, 2:35am UTC](https://discuss.elastic.co/t/shutdown-master-means-breakdown-the-clusters-service/10259/5 "2013-01-09T02:35:43Z")
I didn't change the default setting. After reading your post, I change
the discovery.zen.minimum\_master\_nodes to 1 in config/elasticsearch.yml,
and restart elasticsearch.
It's strange I find es automatically reset
discovery.zen.minimum\_master\_nodes to 2. Here is the log:
[2013-01-09 10:31:06,107][INFO][node] [Wilson, Sam]
{0.19.11-SNAPSHOT}[32214]: started
[2013-01-09 10:31:10,396][INFO][cluster.service] [Wilson, Sam]
detected\_master [Warhawk][h6xkYH-JTT2\_g8pSpLKTkA][inet[/10
.232.42.205:9300]], added
{[Warhawk][h6xkYH-JTT2\_g8pSpLKTkA][inet[/10.232.42.205:9300]],}, reason:
zen-disco-receive(from master [[W
arhawk][h6xkYH-JTT2\_g8pSpLKTkA][inet[/10.232.42.205:9300]]])
[2013-01-09 10:31:10,425][INFO][discovery.zen.elect] [Wilson, Sam]
updating [discovery.zen.minimum\_master\_nodes] from [1] to
[2]
在 2013年1月9日星期三UTC+8上午9时55分50秒，Igor Motov写道：
> You, probably, have discovery.zen.minimum\_master\_nodes setting set to 2.
> So, when the cluster losses the second master eligible node it doesn't
> elect the only remaining node as a master to avoid split brain. You can
> either remove this setting or add the third master eligible node to your
> cluster.
>
> On Tuesday, January 8, 2013 8:46:34 PM UTC-5, asoqa wrote:
>
> > the left node's log:
> > [2013-01-09 09:39:30,517][INFO][discovery.zen] [Quasimodo]
> > master\_left [[Plague][\_6-qEiuHQ76mjwFdTDF9eQ][inet[/10.232.42.205:9300]]],
> > reason [shut\_down]
> > [2013-01-09 09:39:30,518][WARN][discovery.zen] [Quasimodo]
> > not enough master nodes after master left (reason = shut\_down), current
> > nodes: {[Quasimodo][ts0Kq\_c2RoCFwa-UKFI8Tg][inet[/10.232.42.204:9300]],}
> > [2013-01-09 09:39:30,518][INFO][cluster.service] [Quasimodo]
> > removed {[Plague][\_6-qEiuHQ76mjwFdTDF9eQ][inet[/10.232.42.205:9300]],},
> > reason: zen-disco-master\_failed
> > ([Plague][\_6-qEiuHQ76mjwFdTDF9eQ][inet[/10.232.42.205:9300]])
> >
> > the master node's log
> > [2013-01-09 09:39:27,365][INFO][action.admin.cluster.node.shutdown]
> > [Plague] [partial\_cluster\_shutdown]: requested, shutting down
> > [[\_6-qEiuHQ76mjwFdTDF9eQ]] in [1s]
> > [2013-01-09 09:39:28,370][INFO][action.admin.cluster.node.shutdown]
> > [Plague] shutting down in [200ms]
> > [2013-01-09 09:39:28,372][INFO][action.admin.cluster.node.shutdown]
> > [Plague] [partial\_cluster\_shutdown]: done shutting down
> > [[\_6-qEiuHQ76mjwFdTDF9eQ]]
> > [2013-01-09 09:39:28,572][INFO][action.admin.cluster.node.shutdown]
> > [Plague] initiating requested shutdown (using service)
> > [2013-01-09 09:39:30,345][INFO][node] [Plague]
> > {0.19.11-SNAPSHOT}[16001]: stopping ...
> > [2013-01-09 09:39:30,531][INFO][node] [Plague]
> > {0.19.11-SNAPSHOT}[16001]: stopped
> > [2013-01-09 09:39:30,531][INFO][node] [Plague]
> > {0.19.11-SNAPSHOT}[16001]: closing ...
> > [2013-01-09 09:39:30,539][INFO][node] [Plague]
> > {0.19.11-SNAPSHOT}[16001]: closed
> >
> > And I can not search the index, e.g. curl -XGET
> > localhost:9200/twitter/\_search
> >
> > 在 2013年1月8日星期二UTC+8下午5时15分19秒，David Pilato写道：
> >
> > > Why do you think that a new master has not been elected?
> > > Can you share your other nodes logs?
> > >
> > > Le 8 janvier 2013 à 09:59, asoqa [aso...@gmail.com](mailto:aso...@gmail.com) a écrit :
> > >
> > > I have one cluster with two nodes. I used the curl command to shutdown
> > > the master node:
> > >
> > > $ curl -XPOST '[http://localhost:9200/\_cluster/nodes/\_master/\_shutdown](http://localhost:9200/_cluster/nodes/_master/_shutdown)'
> > >
> > > I thought es should automatically choose a new master node to replace
> > > the failed one. But it seems things does not like what I think.
> > >
> > > --
> > >
> > > --
> > > David Pilato
> > > [http://www.scrutmydocs.org/](http://www.scrutmydocs.org/)
> > > [http://dev.david.pilato.fr/](http://dev.david.pilato.fr/)
> > > Twitter : @dadoonet / @elasticsearchfr / @scrutmydocs
--
---
### Author:  [@asoqa](https://discuss.elastic.co/u/asoqa)
#### Post date: [January 9, 2013, 9:53am UTC](https://discuss.elastic.co/t/shutdown-master-means-breakdown-the-clusters-service/10259/6 "2013-01-09T09:53:16Z")
Did I miss something or do something wrong?
在 2013年1月9日星期三UTC+8上午10时35分43秒，asoqa写道：
> I didn't change the default setting. After reading your post, I change
> the discovery.zen.minimum\_master\_nodes to 1 in config/elasticsearch.yml,
> and restart elasticsearch.
> It's strange I find es automatically reset
> discovery.zen.minimum\_master\_nodes to 2. Here is the log:
> [2013-01-09 10:31:06,107][INFO][node] [Wilson, Sam]
> {0.19.11-SNAPSHOT}[32214]: started
> [2013-01-09 10:31:10,396][INFO][cluster.service] [Wilson, Sam]
> detected\_master [Warhawk][h6xkYH-JTT2\_g8pSpLKTkA][inet[/10
> .232.42.205:9300]], added
> {[Warhawk][h6xkYH-JTT2\_g8pSpLKTkA][inet[/10.232.42.205:9300]],}, reason:
> zen-disco-receive(from master [[W
> arhawk][h6xkYH-JTT2\_g8pSpLKTkA][inet[/10.232.42.205:9300]]])
> [2013-01-09 10:31:10,425][INFO][discovery.zen.elect] [Wilson, Sam]
> updating [discovery.zen.minimum\_master\_nodes] from [1] to
> [2]
>
> 在 2013年1月9日星期三UTC+8上午9时55分50秒，Igor Motov写道：
>
> > You, probably, have discovery.zen.minimum\_master\_nodes setting set to 2.
> > So, when the cluster losses the second master eligible node it doesn't
> > elect the only remaining node as a master to avoid split brain. You can
> > either remove this setting or add the third master eligible node to your
> > cluster.
> >
> > On Tuesday, January 8, 2013 8:46:34 PM UTC-5, asoqa wrote:
> >
> > > the left node's log:
> > > [2013-01-09 09:39:30,517][INFO][discovery.zen] [Quasimodo]
> > > master\_left [[Plague][\_6-qEiuHQ76mjwFdTDF9eQ][inet[/10.232.42.205:9300]]],
> > > reason [shut\_down]
> > > [2013-01-09 09:39:30,518][WARN][discovery.zen] [Quasimodo]
> > > not enough master nodes after master left (reason = shut\_down), current
> > > nodes: {[Quasimodo][ts0Kq\_c2RoCFwa-UKFI8Tg][inet[/10.232.42.204:9300]],}
> > > [2013-01-09 09:39:30,518][INFO][cluster.service] [Quasimodo]
> > > removed {[Plague][\_6-qEiuHQ76mjwFdTDF9eQ][inet[/10.232.42.205:9300]],},
> > > reason: zen-disco-master\_failed
> > > ([Plague][\_6-qEiuHQ76mjwFdTDF9eQ][inet[/10.232.42.205:9300]])
> > >
> > > the master node's log
> > > [2013-01-09 09:39:27,365][INFO][action.admin.cluster.node.shutdown]
> > > [Plague] [partial\_cluster\_shutdown]: requested, shutting down
> > > [[\_6-qEiuHQ76mjwFdTDF9eQ]] in [1s]
> > > [2013-01-09 09:39:28,370][INFO][action.admin.cluster.node.shutdown]
> > > [Plague] shutting down in [200ms]
> > > [2013-01-09 09:39:28,372][INFO][action.admin.cluster.node.shutdown]
> > > [Plague] [partial\_cluster\_shutdown]: done shutting down
> > > [[\_6-qEiuHQ76mjwFdTDF9eQ]]
> > > [2013-01-09 09:39:28,572][INFO][action.admin.cluster.node.shutdown]
> > > [Plague] initiating requested shutdown (using service)
> > > [2013-01-09 09:39:30,345][INFO][node] [Plague]
> > > {0.19.11-SNAPSHOT}[16001]: stopping ...
> > > [2013-01-09 09:39:30,531][INFO][node] [Plague]
> > > {0.19.11-SNAPSHOT}[16001]: stopped
> > > [2013-01-09 09:39:30,531][INFO][node] [Plague]
> > > {0.19.11-SNAPSHOT}[16001]: closing ...
> > > [2013-01-09 09:39:30,539][INFO][node] [Plague]
> > > {0.19.11-SNAPSHOT}[16001]: closed
> > >
> > > And I can not search the index, e.g. curl -XGET
> > > localhost:9200/twitter/\_search
> > >
> > > 在 2013年1月8日星期二UTC+8下午5时15分19秒，David Pilato写道：
> > >
> > > > Why do you think that a new master has not been elected?
> > > > Can you share your other nodes logs?
> > > >
> > > > Le 8 janvier 2013 à 09:59, asoqa [aso...@gmail.com](mailto:aso...@gmail.com) a écrit :
> > > >
> > > > I have one cluster with two nodes. I used the curl command to
> > > > shutdown the master node:
> > > >
> > > > $ curl -XPOST '[http://localhost:9200/\_cluster/nodes/\_master/\_shutdown](http://localhost:9200/_cluster/nodes/_master/_shutdown)'
> > > >
> > > > I thought es should automatically choose a new master node to replace
> > > > the failed one. But it seems things does not like what I think.
> > > >
> > > > --
> > > >
> > > > --
> > > > David Pilato
> > > > [http://www.scrutmydocs.org/](http://www.scrutmydocs.org/)
> > > > [http://dev.david.pilato.fr/](http://dev.david.pilato.fr/)
> > > > Twitter : @dadoonet / @elasticsearchfr / @scrutmydocs
--
---
### Author:  [@Igor\_Motov](https://discuss.elastic.co/u/Igor_Motov)
#### Post date: [January 9, 2013, 11:37am UTC](https://discuss.elastic.co/t/shutdown-master-means-breakdown-the-clusters-service/10259/7 "2013-01-09T11:37:06Z")
This setting was probably updated using Update Cluster Setting API[http://www.elasticsearch.org/guide/reference/api/admin-cluster-update-settings.html](http://www.elasticsearch.org/guide/reference/api/admin-cluster-update-settings.html). What
do you get when you run this command:
curl localhost:9200/\_cluster/settings
If it says something like
{"persistent":{"discovery.zen.minimum\_master\_nodes":"2"},"transient":{}}
Run the following command to set number of master nodes to 1:
curl -XPUT localhost:9200/\_cluster/settings -d '{
"persistent" : {
"discovery.zen.minimum\_master\_nodes" : 1
}
}'
On Wednesday, January 9, 2013 4:53:16 AM UTC-5, asoqa wrote:
> Did I miss something or do something wrong?
>
> 在 2013年1月9日星期三UTC+8上午10时35分43秒，asoqa写道：
>
> > I didn't change the default setting. After reading your post, I change
> > the discovery.zen.minimum\_master\_nodes to 1 in config/elasticsearch.yml,
> > and restart elasticsearch.
> > It's strange I find es automatically reset
> > discovery.zen.minimum\_master\_nodes to 2. Here is the log:
> > [2013-01-09 10:31:06,107][INFO][node] [Wilson, Sam]
> > {0.19.11-SNAPSHOT}[32214]: started
> > [2013-01-09 10:31:10,396][INFO][cluster.service] [Wilson, Sam]
> > detected\_master [Warhawk][h6xkYH-JTT2\_g8pSpLKTkA][inet[/10
> > .232.42.205:9300]], added
> > {[Warhawk][h6xkYH-JTT2\_g8pSpLKTkA][inet[/10.232.42.205:9300]],}, reason:
> > zen-disco-receive(from master [[W
> > arhawk][h6xkYH-JTT2\_g8pSpLKTkA][inet[/10.232.42.205:9300]]])
> > [2013-01-09 10:31:10,425][INFO][discovery.zen.elect] [Wilson, Sam]
> > updating [discovery.zen.minimum\_master\_nodes] from [1] to
> > [2]
> >
> > 在 2013年1月9日星期三UTC+8上午9时55分50秒，Igor Motov写道：
> >
> > > You, probably, have discovery.zen.minimum\_master\_nodes setting set to 2.
> > > So, when the cluster losses the second master eligible node it doesn't
> > > elect the only remaining node as a master to avoid split brain. You can
> > > either remove this setting or add the third master eligible node to your
> > > cluster.
> > >
> > > On Tuesday, January 8, 2013 8:46:34 PM UTC-5, asoqa wrote:
> > >
> > > > the left node's log:
> > > > [2013-01-09 09:39:30,517][INFO][discovery.zen] [Quasimodo]
> > > > master\_left [[Plague][\_6-qEiuHQ76mjwFdTDF9eQ][inet[/10.232.42.205:9300]]],
> > > > reason [shut\_down]
> > > > [2013-01-09 09:39:30,518][WARN][discovery.zen] [Quasimodo]
> > > > not enough master nodes after master left (reason = shut\_down), current
> > > > nodes: {[Quasimodo][ts0Kq\_c2RoCFwa-UKFI8Tg][inet[/10.232.42.204:9300]],}
> > > > [2013-01-09 09:39:30,518][INFO][cluster.service] [Quasimodo]
> > > > removed {[Plague][\_6-qEiuHQ76mjwFdTDF9eQ][inet[/10.232.42.205:9300]],},
> > > > reason: zen-disco-master\_failed
> > > > ([Plague][\_6-qEiuHQ76mjwFdTDF9eQ][inet[/10.232.42.205:9300]])
> > > >
> > > > the master node's log
> > > > [2013-01-09 09:39:27,365][INFO][action.admin.cluster.node.shutdown]
> > > > [Plague] [partial\_cluster\_shutdown]: requested, shutting down
> > > > [[\_6-qEiuHQ76mjwFdTDF9eQ]] in [1s]
> > > > [2013-01-09 09:39:28,370][INFO][action.admin.cluster.node.shutdown]
> > > > [Plague] shutting down in [200ms]
> > > > [2013-01-09 09:39:28,372][INFO][action.admin.cluster.node.shutdown]
> > > > [Plague] [partial\_cluster\_shutdown]: done shutting down
> > > > [[\_6-qEiuHQ76mjwFdTDF9eQ]]
> > > > [2013-01-09 09:39:28,572][INFO][action.admin.cluster.node.shutdown]
> > > > [Plague] initiating requested shutdown (using service)
> > > > [2013-01-09 09:39:30,345][INFO][node] [Plague]
> > > > {0.19.11-SNAPSHOT}[16001]: stopping ...
> > > > [2013-01-09 09:39:30,531][INFO][node] [Plague]
> > > > {0.19.11-SNAPSHOT}[16001]: stopped
> > > > [2013-01-09 09:39:30,531][INFO][node] [Plague]
> > > > {0.19.11-SNAPSHOT}[16001]: closing ...
> > > > [2013-01-09 09:39:30,539][INFO][node] [Plague]
> > > > {0.19.11-SNAPSHOT}[16001]: closed
> > > >
> > > > And I can not search the index, e.g. curl -XGET
> > > > localhost:9200/twitter/\_search
> > > >
> > > > 在 2013年1月8日星期二UTC+8下午5时15分19秒，David Pilato写道：
> > > >
> > > > > Why do you think that a new master has not been elected?
> > > > > Can you share your other nodes logs?
> > > > >
> > > > > Le 8 janvier 2013 à 09:59, asoqa [aso...@gmail.com](mailto:aso...@gmail.com) a écrit :
> > > > >
> > > > > I have one cluster with two nodes. I used the curl command to
> > > > > shutdown the master node:
> > > > >
> > > > > $ curl -XPOST '[http://localhost:9200/\_cluster/nodes/\_master/\_shutdown](http://localhost:9200/_cluster/nodes/_master/_shutdown)'
> > > > >
> > > > > I thought es should automatically choose a new master node to
> > > > > replace the failed one. But it seems things does not like what I think.
> > > > >
> > > > > --
> > > > >
> > > > > --
> > > > > David Pilato
> > > > > [http://www.scrutmydocs.org/](http://www.scrutmydocs.org/)
> > > > > [http://dev.david.pilato.fr/](http://dev.david.pilato.fr/)
> > > > > Twitter : @dadoonet / @elasticsearchfr / @scrutmydocs
--
---
### Author:  [@asoqa](https://discuss.elastic.co/u/asoqa)
#### Post date: [January 10, 2013, 5:59am UTC](https://discuss.elastic.co/t/shutdown-master-means-breakdown-the-clusters-service/10259/8 "2013-01-10T05:59:10Z")
thank you for you patience. Problem was solved!
在 2013年1月9日星期三UTC+8下午7时37分06秒，Igor Motov写道：
> This setting was probably updated using Update Cluster Setting API[http://www.elasticsearch.org/guide/reference/api/admin-cluster-update-settings.html](http://www.elasticsearch.org/guide/reference/api/admin-cluster-update-settings.html). What
> do you get when you run this command:
>
> curl localhost:9200/\_cluster/settings
>
> If it says something like
>
> {"persistent":{"discovery.zen.minimum\_master\_nodes":"2"},"transient":{}}
>
> Run the following command to set number of master nodes to 1:
>
> curl -XPUT localhost:9200/\_cluster/settings -d '{
> "persistent" : {
> "discovery.zen.minimum\_master\_nodes" : 1
> }
> }'
>
> On Wednesday, January 9, 2013 4:53:16 AM UTC-5, asoqa wrote:
>
> > Did I miss something or do something wrong?
> >
> > 在 2013年1月9日星期三UTC+8上午10时35分43秒，asoqa写道：
> >
> > > I didn't change the default setting. After reading your post, I change
> > > the discovery.zen.minimum\_master\_nodes to 1 in config/elasticsearch.yml,
> > > and restart elasticsearch.
> > > It's strange I find es automatically reset
> > > discovery.zen.minimum\_master\_nodes to 2. Here is the log:
> > > [2013-01-09 10:31:06,107][INFO][node] [Wilson,
> > > Sam] {0.19.11-SNAPSHOT}[32214]: started
> > > [2013-01-09 10:31:10,396][INFO][cluster.service] [Wilson,
> > > Sam] detected\_master [Warhawk][h6xkYH-JTT2\_g8pSpLKTkA][inet[/10
> > > .232.42.205:9300]], added
> > > {[Warhawk][h6xkYH-JTT2\_g8pSpLKTkA][inet[/10.232.42.205:9300]],}, reason:
> > > zen-disco-receive(from master [[W
> > > arhawk][h6xkYH-JTT2\_g8pSpLKTkA][inet[/10.232.42.205:9300]]])
> > > [2013-01-09 10:31:10,425][INFO][discovery.zen.elect] [Wilson,
> > > Sam] updating [discovery.zen.minimum\_master\_nodes] from [1] to
> > > [2]
> > >
> > > 在 2013年1月9日星期三UTC+8上午9时55分50秒，Igor Motov写道：
> > >
> > > > You, probably, have discovery.zen.minimum\_master\_nodes setting set to
> > > > 2. So, when the cluster losses the second master eligible node it doesn't
> > > > elect the only remaining node as a master to avoid split brain. You can
> > > > either remove this setting or add the third master eligible node to your
> > > > cluster.
> > > >
> > > > On Tuesday, January 8, 2013 8:46:34 PM UTC-5, asoqa wrote:
> > > >
> > > > > the left node's log:
> > > > > [2013-01-09 09:39:30,517][INFO][discovery.zen]
> > > > > [Quasimodo] master\_left
> > > > > [[Plague][\_6-qEiuHQ76mjwFdTDF9eQ][inet[/10.232.42.205:9300]]], reason
> > > > > [shut\_down]
> > > > > [2013-01-09 09:39:30,518][WARN][discovery.zen]
> > > > > [Quasimodo] not enough master nodes after master left (reason = shut\_down),
> > > > > current nodes:
> > > > > {[Quasimodo][ts0Kq\_c2RoCFwa-UKFI8Tg][inet[/10.232.42.204:9300]],}
> > > > > [2013-01-09 09:39:30,518][INFO][cluster.service]
> > > > > [Quasimodo] removed
> > > > > {[Plague][\_6-qEiuHQ76mjwFdTDF9eQ][inet[/10.232.42.205:9300]],}, reason:
> > > > > zen-disco-master\_failed
> > > > > ([Plague][\_6-qEiuHQ76mjwFdTDF9eQ][inet[/10.232.42.205:9300]])
> > > > >
> > > > > the master node's log
> > > > > [2013-01-09 09:39:27,365][INFO][action.admin.cluster.node.shutdown]
> > > > > [Plague] [partial\_cluster\_shutdown]: requested, shutting down
> > > > > [[\_6-qEiuHQ76mjwFdTDF9eQ]] in [1s]
> > > > > [2013-01-09 09:39:28,370][INFO][action.admin.cluster.node.shutdown]
> > > > > [Plague] shutting down in [200ms]
> > > > > [2013-01-09 09:39:28,372][INFO][action.admin.cluster.node.shutdown]
> > > > > [Plague] [partial\_cluster\_shutdown]: done shutting down
> > > > > [[\_6-qEiuHQ76mjwFdTDF9eQ]]
> > > > > [2013-01-09 09:39:28,572][INFO][action.admin.cluster.node.shutdown]
> > > > > [Plague] initiating requested shutdown (using service)
> > > > > [2013-01-09 09:39:30,345][INFO][node] [Plague]
> > > > > {0.19.11-SNAPSHOT}[16001]: stopping ...
> > > > > [2013-01-09 09:39:30,531][INFO][node] [Plague]
> > > > > {0.19.11-SNAPSHOT}[16001]: stopped
> > > > > [2013-01-09 09:39:30,531][INFO][node] [Plague]
> > > > > {0.19.11-SNAPSHOT}[16001]: closing ...
> > > > > [2013-01-09 09:39:30,539][INFO][node] [Plague]
> > > > > {0.19.11-SNAPSHOT}[16001]: closed
> > > > >
> > > > > And I can not search the index, e.g. curl -XGET
> > > > > localhost:9200/twitter/\_search
> > > > >
> > > > > 在 2013年1月8日星期二UTC+8下午5时15分19秒，David Pilato写道：
> > > > >
> > > > > > Why do you think that a new master has not been elected?
> > > > > > Can you share your other nodes logs?
> > > > > >
> > > > > > Le 8 janvier 2013 à 09:59, asoqa [aso...@gmail.com](mailto:aso...@gmail.com) a écrit :
> > > > > >
> > > > > > I have one cluster with two nodes. I used the curl command to
> > > > > > shutdown the master node:
> > > > > >
> > > > > > $ curl -XPOST '[http://localhost:9200/\_cluster/nodes/\_master/\_shutdown](http://localhost:9200/_cluster/nodes/_master/_shutdown)'
> > > > > >
> > > > > > I thought es should automatically choose a new master node to
> > > > > > replace the failed one. But it seems things does not like what I think.
> > > > > >
> > > > > > --
> > > > > >
> > > > > > --
> > > > > > David Pilato
> > > > > > [http://www.scrutmydocs.org/](http://www.scrutmydocs.org/)
> > > > > > [http://dev.david.pilato.fr/](http://dev.david.pilato.fr/)
> > > > > > Twitter : @dadoonet / @elasticsearchfr / @scrutmydocs
--
---
### Author:  [@system](https://discuss.elastic.co/u/system)
#### Post date: [July 6, 2017, 2:57am UTC](https://discuss.elastic.co/t/shutdown-master-means-breakdown-the-clusters-service/10259/9 "2017-07-06T02:57:03Z")