Jump to content

User:Razzi/2021-07-30

From Wikitech

Doing stuff with https://phabricator.wikimedia.org/T283067

Have to run restarts for aqs. Luca says:

razzi: re:  > "restbase" is the aqs cassandra cluster, right?
razzi: Nope :) In the cassandra cluster's dashboard you need to select the analytics datasource (we have a separate prometheus instance) and then you'll see "aqs" as cluster (that should also be present among the option of the cassandra cookbook)
razzi: the restbase cluster is the one managed by SRE
for sre.aqs.roll-restart aqs
we use the canary basically to test safely the new druid mw history snapshot 
(so the cookbook depools one aqs node, restart nodejs and ask to the operator to test locally)
if you have doubts/etc.. ping me anytime!
# Druid analytics-eqiad (non public) servers.
# These power internal backends and queries.
# https://wikitech.wikimedia.org/wiki/Analytics/Data_Lake#Druid
node /^druid100[123]\.eqiad\.wmnet$/ {
    role(druid::analytics::worker)
}
node /^an-druid100[12]\.eqiad\.wmnet$/ {
    role(druid::analytics::worker)
}

# new an-druid nodes T274163
node /^an-druid100[345]\.eqiad\.wmnet$/ {
    role(insetup)
}

node /^an-test-druid1001\.eqiad\.wmnet$/ {
    role(druid::test_analytics::worker)
}

# Druid public-eqiad servers.
# These power AQS and wikistats 2.0 and contain non sensitive datasets.
# https://wikitech.wikimedia.org/wiki/Analytics/Data_Lake#Druid
node /^druid100[4-8]\.eqiad\.wmnet$/ {
    role(druid::public::worker)
}

Ok, so there are still some insetup druid nodes, also an-druid and druid are both analytics druid; druid1004-8 is public. A bit confusing to not have a single node name pattern associated with a single role.