Elukey has submitted this change and it was merged.
Change subject: Decommission the old AQS cluster
......................................................................
Decommission the old AQS cluster
Cleaning up puppet from old references of aqs100[123] nodes
in the AQS/Cassandra configurations. The aqs100[123] are also assigned
to the role spare to be decommissioned.
Bug: T147461
Change-Id: I248ee4f77083345ee823eb285a2ca25983af22e8
---
M conftool-data/nodes/eqiad.yaml
M hieradata/hosts/aqs1004.yaml
M hieradata/hosts/aqs1005.yaml
M hieradata/hosts/aqs1006.yaml
M hieradata/role/common/aqs.yaml
M manifests/site.pp
6 files changed, 33 insertions(+), 127 deletions(-)
Approvals:
Elukey: Looks good to me, approved
Ottomata: Looks good to me, but someone else must approve
jenkins-bot: Verified
diff --git a/conftool-data/nodes/eqiad.yaml b/conftool-data/nodes/eqiad.yaml
index 4e08a96..01e05a5 100644
--- a/conftool-data/nodes/eqiad.yaml
+++ b/conftool-data/nodes/eqiad.yaml
@@ -298,9 +298,6 @@
phabricator:
iridium-vcs.eqiad.wmnet: [git-ssh]
aqs:
- aqs1001.eqiad.wmnet: [aqs]
- aqs1002.eqiad.wmnet: [aqs]
- aqs1003.eqiad.wmnet: [aqs]
aqs1004.eqiad.wmnet: [aqs]
aqs1005.eqiad.wmnet: [aqs]
aqs1006.eqiad.wmnet: [aqs]
diff --git a/hieradata/hosts/aqs1004.yaml b/hieradata/hosts/aqs1004.yaml
index 1f3804b..3d2168d 100644
--- a/hieradata/hosts/aqs1004.yaml
+++ b/hieradata/hosts/aqs1004.yaml
@@ -1,12 +1,4 @@
-# Analytics Query Service - Testing node
-
-cassandra::seeds:
- - aqs1004-a.eqiad.wmnet
- - aqs1004-b.eqiad.wmnet
- - aqs1005-a.eqiad.wmnet
- - aqs1005-b.eqiad.wmnet
- - aqs1006-a.eqiad.wmnet
- - aqs1006-b.eqiad.wmnet
+# Analytics Query Service - rack1 instances
cassandra::rack: 'rack1'
@@ -21,28 +13,3 @@
listen_address: 10.64.0.127 #aqs1004-b.eqiad.wmnet
rpc_address: 10.64.0.127
rpc_interface: eth0
-
-cassandra::cluster_name: "Analytics Query Service Storage"
-
-cassandra::target_version: '2.2'
-
-# Special compaction settings, following suggesions in:
-#
https://docs.datastax.com/en/cassandra/2.1/cassandra/configuration/configCassandra_yaml_r.html
-# All values are divided by two since we have two instances running on each
node
-# Assumption: 32 cores with ht on each host, so 16 * 1.5 = 24 cores considered
-# in the calculations.
-cassandra::compaction_throughput_mb_per_sec: 256
-cassandra::concurrent_compactors: 12
-cassandra::concurrent_writes: 64
-cassandra::concurrent_reads: 64
-
-#
-# AQS service
-#
-aqs::seeds:
- - aqs1004-a.eqiad.wmnet
- - aqs1004-b.eqiad.wmnet
- - aqs1005-a.eqiad.wmnet
- - aqs1005-b.eqiad.wmnet
- - aqs1006-a.eqiad.wmnet
- - aqs1006-b.eqiad.wmnet
diff --git a/hieradata/hosts/aqs1005.yaml b/hieradata/hosts/aqs1005.yaml
index 497cadf..a0f7435 100644
--- a/hieradata/hosts/aqs1005.yaml
+++ b/hieradata/hosts/aqs1005.yaml
@@ -1,12 +1,4 @@
-# Analytics Query Service - Testing node
-
-cassandra::seeds:
- - aqs1004-a.eqiad.wmnet
- - aqs1004-b.eqiad.wmnet
- - aqs1005-a.eqiad.wmnet
- - aqs1005-b.eqiad.wmnet
- - aqs1006-a.eqiad.wmnet
- - aqs1006-b.eqiad.wmnet
+# Analytics Query Service - rack2 instances
cassandra::rack: 'rack2'
@@ -20,29 +12,4 @@
jmx_port: 7190
listen_address: 10.64.32.190 #aqs1005-b.eqiad.wmnet
rpc_address: 10.64.32.190
- rpc_interface: eth0
-
-cassandra::cluster_name: "Analytics Query Service Storage"
-
-cassandra::target_version: '2.2'
-
-# Special compaction settings, following suggesions in:
-#
https://docs.datastax.com/en/cassandra/2.1/cassandra/configuration/configCassandra_yaml_r.html
-# All values are divided by two since we have two instances running on each
node
-# Assumption: 32 cores with ht on each host, so 16 * 1.5 = 24 cores considered
-# in the calculations.
-cassandra::compaction_throughput_mb_per_sec: 256
-cassandra::concurrent_compactors: 12
-cassandra::concurrent_writes: 64
-cassandra::concurrent_reads: 64
-
-#
-# AQS service
-#
-aqs::seeds:
- - aqs1004-a.eqiad.wmnet
- - aqs1004-b.eqiad.wmnet
- - aqs1005-a.eqiad.wmnet
- - aqs1005-b.eqiad.wmnet
- - aqs1006-a.eqiad.wmnet
- - aqs1006-b.eqiad.wmnet
+ rpc_interface: eth0
\ No newline at end of file
diff --git a/hieradata/hosts/aqs1006.yaml b/hieradata/hosts/aqs1006.yaml
index 942a83c..1484eec 100644
--- a/hieradata/hosts/aqs1006.yaml
+++ b/hieradata/hosts/aqs1006.yaml
@@ -1,12 +1,4 @@
-# Analytics Query Service - Testing node
-
-cassandra::seeds:
- - aqs1004-a.eqiad.wmnet
- - aqs1004-b.eqiad.wmnet
- - aqs1005-a.eqiad.wmnet
- - aqs1005-b.eqiad.wmnet
- - aqs1006-a.eqiad.wmnet
- - aqs1006-b.eqiad.wmnet
+# Analytics Query Service - rack3 instances
cassandra::rack: 'rack3'
@@ -20,29 +12,4 @@
jmx_port: 7190
listen_address: 10.64.48.149 #aqs1006-b.eqiad.wmnet
rpc_address: 10.64.48.149
- rpc_interface: eth0
-
-cassandra::cluster_name: "Analytics Query Service Storage"
-
-cassandra::target_version: '2.2'
-
-# Special compaction settings, following suggesions in:
-#
https://docs.datastax.com/en/cassandra/2.1/cassandra/configuration/configCassandra_yaml_r.html
-# All values are divided by two since we have two instances running on each
node
-# Assumption: 32 cores with ht on each host, so 16 * 1.5 = 24 cores considered
-# in the calculations.
-cassandra::compaction_throughput_mb_per_sec: 256
-cassandra::concurrent_compactors: 12
-cassandra::concurrent_writes: 64
-cassandra::concurrent_reads: 64
-
-#
-# AQS service
-#
-aqs::seeds:
- - aqs1004-a.eqiad.wmnet
- - aqs1004-b.eqiad.wmnet
- - aqs1005-a.eqiad.wmnet
- - aqs1005-b.eqiad.wmnet
- - aqs1006-a.eqiad.wmnet
- - aqs1006-b.eqiad.wmnet
+ rpc_interface: eth0
\ No newline at end of file
diff --git a/hieradata/role/common/aqs.yaml b/hieradata/role/common/aqs.yaml
index d2c8064..50e2168 100644
--- a/hieradata/role/common/aqs.yaml
+++ b/hieradata/role/common/aqs.yaml
@@ -15,20 +15,31 @@
cassandra::metrics::graphite_host: 'graphite-in.eqiad.wmnet'
cassandra::start_rpc: 'false'
+cassandra::target_version: '2.2'
+
# TODO: set up a cluster variable similar to MySQL clusters to share
# cassandra cluster configs between cassandra & clients
cassandra::seeds:
- - aqs1001.eqiad.wmnet
- - aqs1002.eqiad.wmnet
- - aqs1003.eqiad.wmnet
+ - aqs1004-a.eqiad.wmnet
+ - aqs1004-b.eqiad.wmnet
+ - aqs1005-a.eqiad.wmnet
+ - aqs1005-b.eqiad.wmnet
+ - aqs1006-a.eqiad.wmnet
+ - aqs1006-b.eqiad.wmnet
cassandra::max_heap_size: 16g
# 1/4 heap size, no more than 100m/thread
cassandra::heap_newsize: 2048m
-cassandra::compaction_throughput_mb_per_sec: 60
-cassandra::concurrent_compactors: 10
-cassandra::concurrent_writes: 18
-cassandra::concurrent_reads: 18
+
+# Special compaction settings, following suggesions in:
+#
https://docs.datastax.com/en/cassandra/2.1/cassandra/configuration/configCassandra_yaml_r.html
+# All values are divided by two since we have two instances running on each
node
+# Assumption: 32 cores with ht on each host, so 16 * 1.5 = 24 cores considered
+# in the calculations.
+cassandra::compaction_throughput_mb_per_sec: 256
+cassandra::concurrent_compactors: 12
+cassandra::concurrent_writes: 64
+cassandra::concurrent_reads: 64
# The CassandraAuthorizer Auth mandates non trivial checks for
# each read/write operation to make sure that permissions are honored.
@@ -51,15 +62,18 @@
aqs::cassandra_password: "%{passwords::aqs::aqs_user}"
cassandra::dc: "%{::site}"
-cassandra::cluster_name: "Analytics Query Service Cassandra"
+cassandra::cluster_name: "Analytics Query Service Storage"
#
# AQS service
#
aqs::seeds:
- - aqs1001.eqiad.wmnet
- - aqs1002.eqiad.wmnet
- - aqs1003.eqiad.wmnet
+ - aqs1004-a.eqiad.wmnet
+ - aqs1004-b.eqiad.wmnet
+ - aqs1005-a.eqiad.wmnet
+ - aqs1005-b.eqiad.wmnet
+ - aqs1006-a.eqiad.wmnet
+ - aqs1006-b.eqiad.wmnet
aqs::logstash_host: logstash1001.eqiad.wmnet
aqs::cassandra_default_consistency: localOne
aqs::cassandra_local_dc: "%{::site}"
@@ -70,9 +84,6 @@
# Used for ferm firewall rules
aqs_hosts:
- - aqs1001.eqiad.wmnet
- - aqs1002.eqiad.wmnet
- - aqs1003.eqiad.wmnet
- aqs1004.eqiad.wmnet
- aqs1005.eqiad.wmnet
- aqs1006.eqiad.wmnet
diff --git a/manifests/site.pp b/manifests/site.pp
index 9865d9a..4efd417 100644
--- a/manifests/site.pp
+++ b/manifests/site.pp
@@ -136,15 +136,12 @@
include base::firewall
}
-# Analytics Query Service (RESTBase & Cassandra)
+# Old Analytics Query Service (now decommissioned)
node /aqs100[123]\.eqiad\.wmnet/ {
- role(aqs)
+ role(spare::system)
}
-# Analytics Query Service - Testing
-# These nodes are not part of the official AQS cluster now because we are
-# currently testing Cassandra configurations on top of them. Hiera variables
-# have been placed for each host to override the role's default values.
+# Analytics Query Service
node /aqs100[456]\.eqiad\.wmnet/ {
role(aqs)
}
--
To view, visit https://gerrit.wikimedia.org/r/314542
To unsubscribe, visit https://gerrit.wikimedia.org/r/settings
Gerrit-MessageType: merged
Gerrit-Change-Id: I248ee4f77083345ee823eb285a2ca25983af22e8
Gerrit-PatchSet: 2
Gerrit-Project: operations/puppet
Gerrit-Branch: production
Gerrit-Owner: Elukey <[email protected]>
Gerrit-Reviewer: Eevans <[email protected]>
Gerrit-Reviewer: Elukey <[email protected]>
Gerrit-Reviewer: Joal <[email protected]>
Gerrit-Reviewer: Ottomata <[email protected]>
Gerrit-Reviewer: jenkins-bot <>
_______________________________________________
MediaWiki-commits mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/mediawiki-commits
