[07:32:47] <elukey>	 bonjour
[07:33:08] <elukey>	 so judging from https://grafana.wikimedia.org/d/000000418/cassandra?orgId=1&var-datasource=eqiad%20prometheus%2Fanalytics&var-cluster=aqs&var-keyspace=local_group_default_T_editors_bycountry&var-table=data&var-quantile=99p it seems that the JVMs are not horribly under pressure
[07:33:20] <elukey>	 but there is the problem of the 16G of heap size
[07:33:30] <elukey>	 we are running to close to the limit
[07:33:50] <elukey>	 the GC activity is fine, but as soon as we cross a certain threshold we go trashing + OOM
[07:41:53] <elukey>	 but at the same time, we cannot really reduce further page cache, since cassandra memory maps sstables when reading
[07:41:57] <elukey>	 (IIRC)
[07:51:21] <elukey>	 so we really need the new boxes with 128G of ram :D
[07:51:56] <elukey>	 Now I am wondering if we made a mistake to get only 128G of ram
[07:51:59] <elukey>	 instead of say 256
[07:52:59] <elukey>	 anyway, nothing on fire :)
[08:59:46] <joal>	 ack elukey - thanks for reviweing
[15:33:44] <wikibugs>	 10Analytics, 10Quarry, 10Security-Team, 10Security, and 2 others: Reflected Cross-Site scripting (XSS) vulnerability in analytics-quarry-web - https://phabricator.wikimedia.org/T270195 (10Framawiki) I have just read this ticket, sorry for the code I added and thanks for the remediation.