[00:20:23] Ok, so as of now T239670 is working right for the proxy but not internally to the cluster. That will break some things but not others :-/ [00:20:23] T239670: Toolforge kubernetes kube-proxy broken by recent production changes - https://phabricator.wikimedia.org/T239670 [00:20:32] Overall, it might not be much impact on webservice users [00:38:29] Ok, now it is fixed on all the worker nodes as well [11:38:40] Hello. On a Toolforge account, how can I access the DB replicas from a CGI (Bash) script when /usr/bin/mysql is not available? It used to be available there and obviously still is from an interactive shell. [12:04:14] hm, so it’s not available in your webservice? [12:04:34] that sounds like a bug to me… maybe something went wrong when rebuilding one of the servers or something [12:15:47] Lucas_WMDE: I believe it used to work, but if I now let a CGI script list files in /usr/bin, there is a lot of files missing compared to what I can see through SSH, and that includes /usr/bin/mysql. [12:17:19] Any idea how this can be solved/reported and/or how I can circumvent it? I need a webservice that performs a query on the DB replicas and returns its result. Yes, I could probably code it in PHP, for instance, but I took Bash because I thought it meant the simplest code. [12:21:14] !help missing /usr/bin binaries in webservices, including /usr/bin/mysql [12:21:14] Lucas_WMDE: If you don't get a response in 15-30 minutes, please create a phabricator task -- https://phabricator.wikimedia.org/maniphest/task/edit/form/1/?projects=wmcs-team [12:21:22] Blahma: ^ you can create a Phabricator task there [12:23:15] OK, I'll wait and if there is no response, I will create the task. [12:23:21] I don’t know how to work around it, though [12:23:58] I guess I'll just run the query manually now for this particular case, but that's of course not a long-time solution for external users of my tool. [12:29:25] blahma: what type of webservice image are you running [12:39:55] Is there a way to get the amount of watchers for a page from the replica databases? [12:42:56] I would be surprised if that’s available, since on-wiki it’s redacted if < 30 [12:43:07] unless there’s some value on the replicas that’s likewise redacted [12:45:53] and as far as I can tell, MediaWiki gets the number of watchers (for ?action=info) directly from COUNT(*) on the watchlist table, there doesn’t seem to be a place where the aggregate count is stored [12:47:03] I think you’ll have to go through the API (action=query, prop=info, inprop=watchers) [12:47:20] :/ [12:47:24] thanks anyways [13:37:15] phamhi: Mine is just a regular tool account with "webservice start" where I put a .cgi file in the public_html with #!/bin/bash in it and the Bash code tries to run a mysql command and return its output. Something like Quarry, but with a fixed SQL query and no UI. [17:32:59] !log deployment-prep deploying ores 6dd1fef87fcd5b76eb52e4a64087fbe8a982c54f [17:33:03] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Deployment-prep/SAL