Repair all keyspaces (#15)

Author: tiago-loureiro

README.md

@@ -5,7 +5,7 @@ Ansible role to install an Apache Cassandra cluster supervised by systemd. Inclu
 * Some OS tuning options such as installing jemalloc, setting max_map_count and tcp_keepalive, disabling swap.
 * Bootstraps nodes using the IPs of the servers in the `cassandra_seed` (configurable) inventory group.
 * Weekly scheduled repairs via cron jobs that are non-overlapping (see `cassandra_repair_slots`).
-    * requires setting `cassandra_keyspaces` (default `[]` will have no effect)
+    * Note that **all** keyspaces will be scheduled for repairs
 * Incremental and full backup scripts as well as a restore script. (disabled by default, optional) (NOTE: needs better testing)
     * backup/restore requires access to S3.
 * prometheus-style metrics using jmx-exporter
@@ -42,12 +42,6 @@ Give your cluster a better name:
 cassandra_cluster_name: default
 ```
 
-You should override the keyspaces to match keyspaces on which you wish to run weekly repairs:
-
-```yaml
-cassandra_keyspaces: []
-```
-
 You may wish to override the following defaults to enable backups:
 
 ```yaml
@@ -108,8 +102,6 @@ Then the following should work and start your cluster:
   vars:
     # set cluster_name before running the playbook for the first time; never change it afterwards
     cassandra_cluster_name: my_cluster
-    cassandra_keyspaces:
-      - my_keyspace1
     # set installed java package version manually. required when using Ubuntu 18.04. see: [A note on Java 8 and Ubuntu 18.04](#a-note-on-Java-8-and-Ubuntu-18.04)
     java_packages: openjdk-8-jdk
   roles:

defaults/main.yml

@@ -45,7 +45,6 @@ cassandra_env: dev
 cassandra_cluster_name: default
 cassandra_role: 'cassandra_{{ cassandra_cluster_name }}'
 cassandra_seed_role: '{{ cassandra_role }}_seed'
-cassandra_keyspaces: []
 cassandra_metrics_prefix: 'cassandra_{{ cassandra_cluster_name }}'
 
 cassandra_snitch: '{% if is_aws_environment %}Ec2Snitch{% else %}SimpleSnitch{% endif %}'

templates/cassandra_repair.j2

@@ -11,45 +11,41 @@ metrics_file=${metrics_dir}/cassandra_repair.prom
 metrics_tmpl='
 # TYPE cassandra_repair counter
 # HELP cassandra_repair Counts the number of repairs and the status
-cassandra_repair{status="%s",errno="%d",keyspace="%s"} 1 %d
+cassandra_repair{status="%s",errno="%d"} 1 %d
 
 # TYPE cassandra_repair_duration_seconds gauge
 # HELP cassandra_repair_duration_seconds Track the duration of the repair task, in seconds
-cassandra_repair_duration_seconds{status="%s",keyspace="%s"} %d %d
+cassandra_repair_duration_seconds{status="%s"} %d %d
 '
 
 started_at=$(date +%s%3N)
 
 function report_metrics {
     status=$1
     errno=$2
-    keyspace=$3
 
     now=$(date +%s%3N)
 
     # nb. the date must be milliseconds since the epoch
     printf "$metrics_tmpl" \
-        $status $errno $keyspace $now $status $keyspace $(((now-started_at)/1000)) $now > ${metrics_file}.$$
+        $status $errno $now $status $(((now-started_at)/1000)) $now > ${metrics_file}.$$
     mv -v ${metrics_file}.$$ ${metrics_file} || true
 }
 
-{% for keyspace in cassandra_keyspaces %}
 # Check if a repair is still running
-if ps -u cassandra -f | grep "repair {{ keyspace }}" > /dev/null
+if ps -u cassandra -f | grep "repair" > /dev/null
 then
-    report_metrics "aborted" 255 "{{ keyspace }}"
+    report_metrics "aborted" 255
     exit 255;
 fi
 
 # Run repair
 su - {{ cassandra_user }} \
-    -c '/opt/cassandra/bin/nodetool repair {{ keyspace }} -full -pr'
+    -c '/opt/cassandra/bin/nodetool repair -full -pr'
 
 if [ $? -ne 0 ]; then
   st="failed"
 else
   st="succeeded"
 fi
-report_metrics $st $? "{{ keyspace }}"
-
-{% endfor %}
+report_metrics $st $?