[10:11:01] !log tools T196137 `aborrero@tools-clushmaster-01:~$ clush -w@all 'sudo wc -l /var/log/exim4/paniclog 2>/dev/null | grep -v ^0 && sudo rm -rf /var/log/exim4/paniclog && sudo service prometheus-node-exporter restart || true'` [10:11:04] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/SAL [10:11:04] T196137: toolforge: prometheus issue is filling up email queue - https://phabricator.wikimedia.org/T196137 [11:46:00] !log wikilabels 7d3af05 is going to staging [11:46:02] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Wikilabels/SAL [11:48:22] It seems great, moving to prod [11:48:58] !log wikilabels 7d3af05 is going to prod [11:48:58] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Wikilabels/SAL [17:29:15] tools.wmflabs.org/bash is taking a whole loooot time to load, any issues on the servers? :) [21:37:41] If I'm setting up my own Redis instance on a Cloud VPS, and I want the instance to be directly accessible from, say, Toolforge, but not the Internet in general, how should I configure the Redis server? [21:47:27] !log deployment-prep deploying ores 6ee8775 [21:47:30] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Deployment-prep/SAL [21:54:39] hare, that's complicated [21:54:56] all the labs instances sit side by side in the same network [21:55:12] I think it's difficult to make it accessible to tools and not other labs instances [21:55:48] allowing connections from the whole of labs and then only distributing the passwords etc. to tools instances might be enough [21:55:53] otherwise, maybe you could do something with security groups [21:56:07] i think i would be fine with that [21:56:16] in that case, 10.0.0.0/8 is the normal range to use [21:56:17] accessible to all of labs is better than accessible to the entirety of the internet [21:56:31] so I would make it accessible to labs and set a really good password and that should cover my bases? [21:56:57] (technically that includes the prod private hosts but I don't think most of prod can talk to labs private stuff anyway, and I wouldn't be worried about prod when you're letting random labs instances connect) [21:57:08] probably [22:45:28] is redis protocol encrypted? Would AUTH just send the pasword plaintext acros sthe internet? [22:52:45] ebernhardson, not sure but I'm not sure that matters much between labs hosts [22:53:19] it won't be across the internet [22:54:46] it'll be across the wikimedia private lan in eqiad [22:57:33] Krenair: oh silly me, i read cloud vps as generic "cloud" and not wmf cloud :) [22:58:16] lol [22:58:24] :)