Elukey has uploaded a new change for review. (
https://gerrit.wikimedia.org/r/363154 )
Change subject: role::analytics_cluster::hadoop::master: add more monitors to
HDFS metrics
......................................................................
role::analytics_cluster::hadoop::master: add more monitors to HDFS metrics
Bug: T163908
Change-Id: I22910c9e216b3af5211a7c2d7053e75bd4fc745f
---
M modules/role/manifests/analytics_cluster/hadoop/master.pp
1 file changed, 33 insertions(+), 0 deletions(-)
git pull ssh://gerrit.wikimedia.org:29418/operations/puppet
refs/changes/54/363154/1
diff --git a/modules/role/manifests/analytics_cluster/hadoop/master.pp
b/modules/role/manifests/analytics_cluster/hadoop/master.pp
index c078465..00c5990 100644
--- a/modules/role/manifests/analytics_cluster/hadoop/master.pp
+++ b/modules/role/manifests/analytics_cluster/hadoop/master.pp
@@ -92,6 +92,39 @@
],
}
+ # Alert if the HDFS space consumption raises above a safe threshold.
+ monitoring::graphite_threshold { 'hadoop-hdfs-percent-used':
+ description => 'HDFS capacity used percentage',
+ metric =>
"Hadoop.NameNode.${::hostname}_eqiad_wmnet_9980.Hadoop.NameNode.NameNodeInfo.PercentUsed.mean",
+ from => '30min',
+ warning => 70,
+ critical => 80,
+ percentage => '60',
+ contact_group => 'analytics',
+ }
+
+ # Alert in case of HDFS currupted or missing blocks. In the ideal state
+ # these values should always be 0.
+ monitoring::graphite_threshold { 'hadoop-hdfs-corrupt-blocks':
+ description => 'HDFS missing blocks',
+ metric =>
"Hadoop.NameNode.${::hostname}_eqiad_wmnet_9980.Hadoop.NameNode.NameNodeInfo.PercentUsed.mean",
+ from => '30min',
+ warning => 70,
+ critical => 80,
+ percentage => '60',
+ contact_group => 'analytics',
+ }
+
+ monitoring::graphite_threshold { 'hadoop-hdfs-missing-blocks':
+ description => 'HDFS corrupted blocks',
+ metric =>
"Hadoop.NameNode.${::hostname}_eqiad_wmnet_9980.Hadoop.NameNode.FSNamesystem.MissingBlocks.mean",
+ from => '180min',
+ warning => 2,
+ critical => 5,
+ percentage => '60',
+ contact_group => 'analytics',
+ }
+
# Java heap space used alerts.
# The goal is to get alarms for long running memory leaks like T153951.
# Only include heap size alerts if heap size is configured.
--
To view, visit https://gerrit.wikimedia.org/r/363154
To unsubscribe, visit https://gerrit.wikimedia.org/r/settings
Gerrit-MessageType: newchange
Gerrit-Change-Id: I22910c9e216b3af5211a7c2d7053e75bd4fc745f
Gerrit-PatchSet: 1
Gerrit-Project: operations/puppet
Gerrit-Branch: production
Gerrit-Owner: Elukey <ltosc...@wikimedia.org>
_______________________________________________
MediaWiki-commits mailing list
MediaWiki-commits@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/mediawiki-commits
