[10:58:42] is it possible to update the ruby version used? [10:58:46] toolforge seems to be on 2.3...? [10:58:58] $ ruby -v [10:58:58] ruby 2.3.3p222 (2016-11-21) [x86_64-linux-gnu] [10:59:10] no rbenv/rvm afaics? [11:02:16] proc: there should be a container in toolforge kubernetes for webservices running ruby 2.5.5 [11:02:37] https://wikitech.wikimedia.org/wiki/Help:Toolforge/Kubernetes#Available_container_types [11:02:56] What about for non webservices? [11:03:18] This is for a job I want to run on the grid, for context [11:08:33] in the grid you can try with rbenv, but I'm not sure, I'm not familiar with that stack [11:09:12] today the grid runs Debian Stretch, which is a bit outdated. We are planning on upgrading it to Debian Buster soon, which should have more updated packages for several frameworks and langs [11:09:53] T141388 might have some helpful pointers [11:09:54] T141388: Add a easy way to run a ruby webservice on tools - https://phabricator.wikimedia.org/T141388 [11:16:45] !log toolsbeta aborrero@cloudcontrol1005:~ $ sudo wmcs-openstack --os-project-id=toolsbeta port create --network lan-flat-cloudinstances2b toolsbeta-redis-vip (T153810) [11:16:49] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Toolsbeta/SAL [11:16:49] T153810: Make switching Redis server simpler - https://phabricator.wikimedia.org/T153810 [11:21:58] !log toolsbeta allowed VIP address from the new port 172.16.3.26 into the ports of toolsbeta-redis-[1-3] (T153810) [11:22:01] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Toolsbeta/SAL [11:22:01] T153810: Make switching Redis server simpler - https://phabricator.wikimedia.org/T153810 [11:58:03] i cant seem to run "rbenv install 2.7.3", command just times out. i've had it running for like 20 mins. have interrupted and will try again tho [11:58:29] https://www.irccloud.com/pastebin/tZ1gBeun/ [12:04:18] yeh same issue again [12:20:50] !log tools.wikibugs Updated channels.yaml to: e55fede038ce3d79e7208cfd85fa4c7c8ef6886b Remove old config that I do not need anymore [12:20:52] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools.wikibugs/SAL [13:26:34] !log tools.github-pr-closer updating to https://gerrit.wikimedia.org/r/c/labs/tools/github-pr-closer/+/691201 [13:26:36] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools.github-pr-closer/SAL [13:47:18] can a sysadmin try to install the ruby v using rbenv on my local user? as i say, it's just timing out for me [14:34:35] proc: any user session on a bastion is given very low resource limits to protect the bastion from being overloaded. I would recommend trying to run your rbenv install as a grid job. That should make it run faster. It still may take a while as there are NFS bandwidth limits even on the grid. [14:38:25] seems to just stop as soon as it runs. any way to view the output of a job so i can see what the error is? [14:38:29] bd808 [14:38:42] (ran using jsub rbenv install 2.7.3) [14:40:25] proc: look for rbenv.out & rbenv.err in your $HOME. Those would be the default stdout & stderr log files generated I think [14:40:34] proc: by default they go to the tools home directory, jobname.{out,err} [14:40:55] Are others having trouble ssh:ing to bastion.wmcloud.org ? [14:41:28] ruby-build: TMPDIR=/tmp/2562175.1.task cannot hold executables (partition possibly mounted with `noexec`) [14:41:29] hmm [14:41:44] MichaelCochez: works ok for me, so the host is probably ok. [14:42:13] guess i can set another tmpdir [14:43:27] MichaelCochez: https://wikitech.wikimedia.org/wiki/Reporting_a_connectivity_issue might have some things to help you figure out if there is a routing problem from your location to our eqiad datacenter where the bastion.wmcloud.org instance lives [14:44:42] can the grid read variables in the session that starts the job? if not how are env variables supplied with jsub? [14:46:02] MichaelCochez: I think the sisue is generally network access to recommender.schematreerecommender.eqiad1.wikimedia.cloud [14:46:46] (tried jsub -v TMPDIR=~/tmp rbenv install 2.7.3) [14:47:43] perhaps the security groups are wrong homehow [14:48:56] proc: try making a shell script to wrap it all together and submit that. Otherwise I think you would have to use /usr/bin/env as the main process which works but can be a pain to get just right. [14:49:54] the grid can access my files right? so /data/project/... is a valid path for a submitted job? [14:50:27] proc: yes [14:50:53] addshore, MichaelCochez: I'm trying that host with my root key and it does seem like something is keeping the session from being established. [14:52:15] woo! it's not just us then! :) [14:52:38] * bd808 peeks at config in Horizon [14:53:05] The "Security Groups" look "default" to me :/ but also different to what I see on other instances [14:54:07] yeah... I've got a hunch that OpenStack is making weird initial security groups again. This hits us occasionally after an OpenStack upgrade. [14:56:43] I am still getting : $ ssh bastion.wmcloud.org [14:56:44] ssh: Could not resolve hostname bastion.wmcloud.org: Temporary failure in name resolution [14:58:10] addshore, MichaelCochez: I added the missing ssh allow rule to your project and I can ssh in now. [14:59:32] MichaelCochez: that sounds like a local problem? `dig @8.8.8.8 bastion.wmcloud.org` returns an ip for me which should mean that any resolver can find it. [15:06:08] bd808: looks like that tip (make a shell script) is working. ty :) [15:06:23] ^ hopefully didnt jinx it [15:07:05] :) [15:09:18] Hello, I have issues with resolving DNS from bastion, see [15:09:19] https://www.irccloud.com/pastebin/ZF8s8lGi/ [15:10:03] using 8.8.8.8 works, so this sounds to be an issue in Toolforge infra? https://www.irccloud.com/pastebin/ct5otpGc/ [15:10:51] I once again can't resolve bastion.wmfcloud.org now :P [15:10:57] *wmcloud [15:11:07] that one works from toolforge [15:11:23] ...and on my home resolver [15:12:05] also can't resolve bastion.wmflabs.org from here, what is going on today, DNS as always having some fun [15:12:11] we're investigating the dns issues [15:12:28] cool :-). Is there a task to watch? :) [15:12:40] not yet [15:12:43] * Majavah makes one [15:14:22] Urbanecm: T282877 [15:14:22] T282877: Cloud VPS DNS issues, 2021-05-14 - https://phabricator.wikimedia.org/T282877 [15:14:28] thx [15:33:28] These were DNS issues apparently were bigger than just cloud and should be resolving now. [15:33:36] *that apparently [15:36:13] thanks bstorm [15:54:30] Thank you, this works now. [15:55:07] https://www.irccloud.com/pastebin/vkXdsNLi/ [15:55:10] any ideas? [15:55:34] https://www.irccloud.com/pastebin/QfPmtSTJ/ [15:58:24] proc: I get that same failure from my local system, so I don't think this is a cloud-vps/toolforge issue; someone at rubygems.org will probably need to fix it [15:58:46] (or, possible alternative explanation: there is no longer anyone at rubygems.org) [16:00:10] andrewbogott: it's working for me from my local system [16:00:38] PING rubygems.org (151.101.65.227) [16:01:57] https://www.irccloud.com/pastebin/tfrl5U8W/ [16:02:02] hmm [16:02:17] oh, you're right, I had a typo [16:02:28] can i change DNS servers on toolforge? [16:02:38] so I guess I don't know what's happening, best to open a ticket [16:02:54] bd808: Majavah: is this related to T282877? [16:02:55] T282877: Cloud VPS DNS issues, 2021-05-14 - https://phabricator.wikimedia.org/T282877 [16:03:18] might be [16:03:45] does cloudflare host dns for that domain? [16:03:50] could be yeah. I think we are having some edge network problems related to UDP? [16:04:14] Majavah: dyn [16:04:26] slash dnsimple [16:05:04] is it possible to run "bundle install" while specifying 8.8.8.8 as the DNS server to use? i gather i can't change DNS servers locally [16:05:27] I thought the issues on our side were resolved [16:05:33] anyways, /me afk [16:06:15] well, it seems to work if i specify google's DNS servers in the `host` command [16:06:33] so i'd guess it's an issue with toolforge's default DNS servers? [16:13:37] ill try a mirror i guess [16:25:42] `dig NS dnsimple.com` doesn't even show the glue records for that domain somehow [16:27:19] proc: you may be hitting some kind of internet weather event, dns seems weird all over. If you don't want to look for a workaround it might be worth just waiting a few hours. [16:39:36] well, looks like i got it all working (at last) [16:39:47] last Q, is it fine to store acct credentials in the .sh that runs the app? [16:47:23] !log tools deleting log files older than 14 days on toolserver-proxy-01 [16:47:27] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/SAL [16:49:13] proc: yes, but only if you have made that file unreadable by other users. `chmod o= ` [16:55:34] !log tools rebooting toolserver-proxy-01 to clear up stray files [16:55:37] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/SAL [16:56:08] /dev/vda2 19G 3.7G 14G 21% / [17:23:45] Hi, I read about the OSI-approved licensing restriction for Toolforge tools. I was wondering if it's still possible to license under CC0, although it's not OSI approved? [17:24:43] !help [17:24:43] 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-kanban [17:24:52] Ahh, okay, I'll wait [17:25:13] epicpupper-is-tr: typically cc licenses are intended for content and not for code. There are equivalent OSI-approved licenses though [17:25:28] CC0 is also intended for code, I believe [17:25:36] (essentially public domain) [17:25:38] https://opensource.org/faq#cc-zero [17:26:19] Ah okay, I think I'll license under Apache instead. Thank you for the link [17:27:12] Apache is good! I believe that MIT has even fewer restrictions: https://opensource.org/licenses/MIT [17:27:21] (tbh I see apache getting used more though) [17:32:42] apache2 has patent indemnification and some other niceties for the project. And of course it is required as the license if you are transferring the project to apache [17:33:14] * bd808 is fond of copyleft personally [17:40:22] * andrewbogott is too but the conversation started at cc0 so that seems like a stretch [17:46:42] speaking of CC0, Kat is currently looking for software projects that use a CC license instead of one actually intended for software: https://mastodon.social/@mindspillage/106230200641775418 [17:47:12] all gadgets? (re @wmtelegram_bot: speaking of CC0, Kat is currently looking for software projects that use a CC license instead of one actually intended for software: https://mastodon.social/@mindspillage/106230200641775418) [17:47:35] chicocvenancio: I already mentioned that and Lua templates :) [18:36:52] if you just need CC0 equivalent for sw - maybe wtfpl? https://spdx.org/licenses/WTFPL.html [18:37:47] not OSI approved, which would be required [18:38:01] legoktm: if you count cc0, https://github.com/liberapay/liberapay.com#license (yeah, no LICENSE file sucks, but blame the maintainer) [18:38:44] but zero-clause bsd is approved, which is effectively pd with the no warranty clause from bsd licenses [19:18:54] !log tools adjusting the rate limits for bastions nfs_write upward a lot to make NFS writes faster now that the cluster is finally using 10Gb on the backend and frontend T218338 [19:18:58] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/SAL [19:18:58] T218338: labstore: Re-evaluate traffic shaping settings - https://phabricator.wikimedia.org/T218338