Looking across the fleet, the `purged_event_lag` metric looks really bad on four hosts, which are now depooled: cp2033, cp1087, cp2037, cp2039.
I stumbled across this after being forwarded some user reports on Twitter: https://twitter.com/john_overholt/status/1276276247602044933
The reporter was on the US East Coast (actually just a few miles from me), so I looked at eqiad:
```鉁旓笍 cdanis@cumin1001.eqiad.wmnet ~ 馃暀馃嵑 sudo cumin 'A:cp-text and A:eqiad' 'curl -s --resolve en.wikipedia.org:443:127.0.0.1 https://en.wikipedia.org/wiki/Samuel_Osborne | grep Legacy||echo FAIL'
8 hosts will be targeted:
cp[1075,1077,1079,1081,1083,1085,1087,1089].eqiad.wmnet
Confirm to continue [y/n]? y
===== NODE GROUP =====
(7) cp[1075,1077,1079,1081,1083,1085,1089].eqiad.wmnet
----- OUTPUT of 'curl -s --resolv...egacy||echo FAIL' -----
<h3><span class="mw-headline" id="Legacy">Legacy</span><span class="mw-editsection"><span class="mw-editsection-bracket">[</span><a href="/w/index.php?title=Samuel_Osborne&action=edit&section=1" title="
Edit section: Legacy">edit</a><span class="mw-editsection-bracket">]</span></span></h3>
===== NODE GROUP =====
(1) cp1087.eqiad.wmnet
----- OUTPUT of 'curl -s --resolv...egacy||echo FAIL' -----
FAIL
================
```
https://w.wiki/VDv
{F31906076}