[00:19:29] How can a test fail in 1.35, but not in 1.36, but no obvious related changes... [00:21:18] (under php 8.0) [00:22:13] Numerous Failed asserting that exception of type "PHPUnit\Framework\Error\Deprecated" is thrown. [00:25:17] how can i render ~~~~ instead of a timestamp? [00:25:31] for the purposes of explaining how to use the timestamp mechanism [00:25:39] ~~~~ [00:26:02] awesome! thanks Reedy [00:28:39] >This is due to https://wiki.php.net/rfc/tostring_exceptions. PHP now throws Error exception instead of E_RECOVERABLE_ERROR. [00:29:30] not quite [00:32:26] Bets it's the back compat traits? [00:33:14] I guess it's PHPUnit4And6Compat::setExpectedException [00:35:47] but that's exception not deprecation [00:38:55] buuut [00:38:55] 01:13:31 Failed asserting that exception of type "PHPUnit\Framework\Error\Deprecated" is thrown. [02:52:00] Hey all, anyone use wikibase [02:52:01] ? [02:52:14] I am having difficulties linking one item to another item through a property. [02:52:28] I was hoping someone had some info or a video on how it works. [02:56:17] might need to wait a bit, its getting to that time where many of the more knowledgable ones are going to bed, or are just about to wake up [02:57:30] I figured as much. [02:57:47] I am trying to watch some videos online, but most videos show how to install wikibase, which I've already done. [02:57:56] Not much info on actually managing the data. [08:45:36] After a good night's sleep and 3 mugs of coffee and some thinking I have realized the following about my problem wikis: The "ghost"-revision about "backups made and verfied" from my log-article, that shows up where-ever I do one edit, and that is "not supposed to exist" in the database dump, is showing up because I was drafting that revision while the mysqldump was running, thinking it will not affect the state of the backup, as long as I don' [08:45:38] t save it before I have verified the backups on the verification server and I was wrong on this [09:01:07] Always use the --single-transaction arg on mysqldump! [09:01:54] Can someone corroborate or disprove my current my hypothesis, that because of editing while mysqldump was running, the "edit that is not supposed to be in the dump" exists in the database in some form, but is not a part of a completed edit and is messing up the restored wikis as soon as one tries to do one edit in them [09:05:06] Editing but *not saving* an edit, doesn't impact the database, other than stashing the edit in the cache. You said you were using Memcached exclusively. If that's true, it won't impact the database, since the stash will be on memcached and not on the database [09:18:59] Vulpix: hmmmmmm, thanks for the info, though it seems to contradict what I think I am observing ... could there be some reason why the edit I drafted while mysqldump was running made its way to the dump only partially? [09:20:44] There's no way unless you actually save your edit [09:21:28] ok thank you for the info [09:21:30] Last night I spun up a new server from the latest snapshot from 2021-04-22 and after adjusting the conf I was able to serially edit the temporary "develop"-wiki. Then I realized I'm paying +20% on servers for 7 day round robin backups and converted a few very recent backups to snapshots. I'm trying to evaluate which way I want to recover with the help of the combination of snapshot & backups [09:26:51] intuitively the easiest and logically arguable route would be to spin up a new server from the 2021-05-14 or 2021-05-15 and then mysqldump (with --single-transaction) on that machine and try importing those to the production machine [10:02:14] I made a new server from snapshot, dumped the database from that one and dropped, recreated and loaded the database on the production server and it seems that the MariaDB is on the production server is broken / hacked, because I restored database from 2021-05-14 and once again an edit from 2021-05-15 is showing up "out of nowhere" and in the wrong place [10:07:14] Iamthehuman: what about editing on the snapshot and see problem is reproduced there? If not, just make the snapshot the new production (or restore it entirely) [10:09:05] hi all, I will try again, who can help me with a ldap related question? [10:28:00] Last night I ascertained that with a little conf tweaking the 2021-04-22 snapshot works, I'll do the same moves for the 2021-05-14 snapshot-from-backup. Then we know [10:42:10] * Iamthehuman waiting for DNS propagation to happen [10:43:55] Iamthehuman: you can edit your /etc/hosts file to point to the new IP without waiting for the DNS to propagate. Just remember to remove it when you're done [10:44:28] You may need to restart your browser after the change just in case it cached the IP somehow [10:44:29] yeah, I know about /etc/hosts [10:45:12] there are different workflows that must be followed depending on if I want to restore in-place or move to new server from snapshot. soon I will know if the 2021-05-14 snapshot is in functioning state or not [11:08:45] hmm.. techincal difficulty.. the temp subdomain is using a wrong wildcard cert, disabling the other domains fixed this type of situation last night, but I'm wondering why it is using the wrong cert when the right one is in the -directive unchanged [11:52:47] that was my bad, now sorted out, proceeding to make edits in the temporary wiki on the temporary server from snapshots [11:58:44] Yes, the one wiki I adapted to be served from the new server from recent snapshot is in functioning state. I can work from this by one of two different routes: #1 In-place restore and not change anything in DNS, or #2 Spin new server, adjust the apache conf and make sure everything is ok and then change the DNS for the sites. [12:02:10] these differ in my ability to safely get the most recent changes of the sites that do not seem to be affected by the "Ghost in the MariaDB". As there is a ghost in the MariaDB of the current production server I'm going to change ssh keys and relevant passwords, whichever route I go [17:34:49] Remember Bitcoin in 2008??? Pi is a new digital currency developed by Stanford PhDs, To claim your piece of pi goto https://minepi.com and use "ilkde" as your invitation code. Get your piece of the pi now ! [21:35:50] * bd808 waves to libera across the bridge [21:36:05] legoktm: ^ it's alive [21:36:19] thank you! [21:40:18] I guess that takes care of logging somewhat [21:40:39] probably [21:45:04] wow, is there an official announcement? [21:52:44] see https://meta.wikimedia.org/wiki/Wikimedia_Forum#Freenode_(IRC) [21:53:11] there is a message from Az1568 saying the group contacts decided to move [21:53:45] oh, and there's a box at the top of https://meta.wikimedia.org/wiki/IRC [21:54:22] Vulpix: I was waiting for the bridge to be set up before officially announcing it [21:54:52] oh, intresting, I just wanted to get cloak, so, it is late? xD [21:55:02] cool [21:55:10] https://meta.wikimedia.org/wiki/IRC/Migrating_to_Libera_Chat#Cloaks [21:57:06] I noted at https://meta.wikimedia.org/wiki/Wikimedia_Forum#Stable_irc_network_hostname, we should have an official chat hostname under a wmf-owned domain [21:57:22] in case switching to a different network is needed at a later point [21:57:54] that would be nice, like how irc.debian.org is a CNAME to OFTC [21:58:21] Platonides: the accounts would still be unstable, so I'm not sure I see that as a huge win and actually more difficult to perform a migration. [21:58:24] not sure which would be the right domain, though [21:58:37] Vulpix thx :) [21:58:49] bd808: why more difficult? [21:59:14] a later migration would still have some pain points [21:59:16] Platonides: how would we do half-in half-out like we are doing right now? [21:59:53] "change all your config to and then change back in 2 weeks?" [21:59:54] if we wanted to use the old network after changing the CNAME? [21:59:55] I hope someday everyone will move to discord :) [22:00:05] by explicitely using the actual network [22:00:12] Re: (TLS certificates could be a problem, but hopefully the new network will be willing to add that hostname to their server certificate): I think it would be a problem for libera.chat if every project asks to list their own hostname on the certificate [22:01:01] SANs can contain a good number of entries [22:02:38] that burden could be lessed as well if the project doing such request is willing to host a server ;) [22:03:02] I assuming they use a similar scheme as freenode? [22:17:39] Platonides: I read your note on wiki again, and I think I grok what you are saying now. My understanding is that it would make it much easier to deal with documentation updates which makes sense. I was thinking more about user accounts and client configs. [22:18:27] it would somewhat help with client configs [22:18:35] specially for people which join once in a blue moon [22:18:46] it won't help with user accounts [22:19:04] feel free to clarify my edit [22:21:08] https://lists.wikimedia.org/hyperkitty/list/mediawiki-l@lists.wikimedia.org/thread/5VVH5LTZUIN63WNAJZFREMD66LFQLTFT/ [23:25:32] Don't you know the Libera will have a web interface? [23:26:02] Iniquity: you can use https://kiwiirc.com/nextclient/irc.libera.chat/?#mediawiki for now [23:26:32] Okay, I'm thinking of changing the links on the local wiki, thanks! :) [23:28:28] we can always change it again later :) [23:31:11] xD