Hi!
Can someone pls explain to me why some shards dont get relocated to the correct node in my cluster? I want no shard on node-4 at the moment.
Infos:
GET _cat/nodeattrs
node-4 172.21.2.174 172.21.2.174 ml.machine_memory 17179869184
node-4 172.21.2.174 172.21.2.174 ml.max_open_jobs 20
node-4 172.21.2.174 172.21.2.174 xpack.installed true
node-4 172.21.2.174 172.21.2.174 store_type warm
node-4 172.21.2.174 172.21.2.174 transform.node true
node-1 172.21.2.120 172.21.2.120 ml.machine_memory 33620926464
node-1 172.21.2.120 172.21.2.120 xpack.installed true
node-1 172.21.2.120 172.21.2.120 transform.node true
node-1 172.21.2.120 172.21.2.120 ml.max_open_jobs 20
node-1 172.21.2.120 172.21.2.120 store_type hot
node-3 172.21.3.14 172.21.3.14 ml.machine_memory 17179869184
node-3 172.21.3.14 172.21.3.14 ml.max_open_jobs 20
node-3 172.21.3.14 172.21.3.14 xpack.installed true
node-3 172.21.3.14 172.21.3.14 store_type hot
node-3 172.21.3.14 172.21.3.14 transform.node true
node-2 172.21.2.86 172.21.2.86 ml.machine_memory 33613492224
node-2 172.21.2.86 172.21.2.86 ml.max_open_jobs 20
node-2 172.21.2.86 172.21.2.86 xpack.installed true
node-2 172.21.2.86 172.21.2.86 store_type hot
node-2 172.21.2.86 172.21.2.86 transform.node true
shards on node-4:
ppr-ecs-test-2020.09 0 p STARTED 249 261.5kb 172.21.2.174 node-4
index settings:
{
"ppr-ecs-test-2020.09" : {
"settings" : {
"index" : {
"routing" : {
"allocation" : {
"include" : {
"store_type" : "hot"
}
}
},
"number_of_shards" : "1",
"provided_name" : "ppr-ecs-test-2020.09",
"creation_date" : "1599654541084",
"number_of_replicas" : "1",
"uuid" : "CS-UYQWcSCKdi-NAFGTsdQ",
"version" : {
"created" : "7080099"
}
}
}
}
}
lzukel
(Lance)
September 10, 2020, 1:08pm
2
Try the following
{
"settings" : {
"index" : {
"routing" : {
"allocation" : {
"require" : {
"store_type" : "hot"
}
}
},
"number_of_shards" : "1",
"number_of_replicas" : "1"
}
}
}
}
The difference between include and require:
Include means nodes may have that attribute
Require means nodes must have that attribute
Can i overwrite the setting somehow? or delete the old one?
Because now i have this:
{
"ppr-ecs-test-2020.09" : {
"settings" : {
"index" : {
"routing" : {
"allocation" : {
"include" : {
"store_type" : "hot"
},
"require" : {
"store_type" : "hot"
}
}
},
"number_of_shards" : "1",
"provided_name" : "ppr-ecs-test-2020.09",
"creation_date" : "1599654541084",
"number_of_replicas" : "1",
"uuid" : "CS-UYQWcSCKdi-NAFGTsdQ",
"version" : {
"created" : "7080099"
}
}
}
}
}
lzukel
(Lance)
September 10, 2020, 2:00pm
4
"include" : {
"store_type" : ""
}
setting null value will essentially remove the setting
I would also recommend using index template for configuring this setting for all newly created indices where this setting is required.
OK, now with:
{
"ppr-ecs-test-2020.09" : {
"settings" : {
"index" : {
"routing" : {
"allocation" : {
"include" : {
"store_type" : ""
},
"require" : {
"store_type" : "hot"
}
}
},
"number_of_shards" : "1",
"provided_name" : "ppr-ecs-test-2020.09",
"creation_date" : "1599654541084",
"number_of_replicas" : "1",
"uuid" : "CS-UYQWcSCKdi-NAFGTsdQ",
"version" : {
"created" : "7080099"
}
}
}
}
}
Its the same. Shard is on node-4 pretty much ignoring the setting.
Yeah for the new ones i do it via the template.
lzukel
(Lance)
September 10, 2020, 2:12pm
6
Can you post the output of:
GET /_cat/recovery/ppr-ecs-test-2020.09
ppr-ecs-test-2020.09 0 576ms peer done 172.21.2.120 node-1 172.21.3.14 node-3 n/a n/a 0 0 0.0% 0 0 0 0.0% 0 0 0 100.0%
ppr-ecs-test-2020.09 0 2.5s peer done 172.21.2.120 node-1 172.21.2.174 node-4 n/a n/a 16 16 100.0% 16 267821 267821 100.0% 267821 0 0 100.0%
lzukel
(Lance)
September 10, 2020, 2:30pm
8
Can you share response of:
GET /allocation/explain?pretty
Can also try:
POST /_cluster/reroute?retry_failed=true
GET /_cat/allocation/explain?pretty
EMPTY response
POST /_cluster/reroute?retry_failed=true
...
"ppr-ecs-test-2020.09" : {
"shards" : {
"0" : [
{
"state" : "STARTED",
"primary" : false,
"node" : "foRZv0ngQieXX3uwSXrZZg",
"relocating_node" : null,
"shard" : 0,
"index" : "ppr-ecs-test-2020.09",
"allocation_id" : {
"id" : "NP22yUdkTkmpCM4vvzrF7A"
}
},
{
"state" : "STARTED",
"primary" : true,
"node" : "zlcwsGyVSxOSE4c8I7MeVw",
"relocating_node" : null,
"shard" : 0,
"index" : "ppr-ecs-test-2020.09",
"allocation_id" : {
"id" : "On_WcGlVT9KJd-_tM06MYA"
}
}
]
}
}
...
After the upgrade to 7.9.1 the shards got assigned on the correct node
system
(system)
Closed
October 9, 2020, 6:26am
11
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.