[07:05:13] Evrifaessa: You can install the desired version of python in your home directory, rather than depend on the system-installed version. [07:06:32] can I? I don't think I have permissions to run apt and install Python [07:18:44] Evrifaessa: well you can always compile it yourself, but I'd strongly suggest tweaking your code so it's compatible with Python 3.5 (if it's a webservice, use 3.7) [07:19:03] there are ways you can run bots on the webservice system to get access to 3.7, but it's a bit complicated [07:49:48] legoktm: thanks [08:52:34] [telegram] 🔥Q系列： [08:52:34] [telegram] 1：私人营销Q，每天可加人加群（不限IP，永久售后）【买5个送1个】---158元🔥🔥🔥 [08:52:36] [telegram] 2：Q群主号：自带几百个老群----联系客服 [08:52:37] [telegram] 3：8位靓号秒绑手机3A-4A豹子号任选（不占卡位）-----188元（每天可加人加群） [08:52:38] [telegram] 4：8位普号Q龄14年以上----198元 [08:52:40] [telegram] 5：8位皇冠----688元 [08:52:41] [telegram] 6：单太阳Q号16-31级---25元 [08:52:42] [telegram] 7：双太阳Q好32-40级---45元 [08:52:43] [telegram] 8：三太阳48-49级4-7年----75元 [08:52:45] [telegram] 支付宝 陌陌 抖音 微信 企业微信 云控 等业务 [08:52:46] [telegram] 联系Q：319··078··084 [08:53:00] wat [12:04:22] !log wmde-templates-alpha create mediawiki1001 instance [12:04:24] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Wmde-templates-alpha/SAL [12:19:30] Is role::labs::mediawiki_vagrant the recommended way to set up a MediaWiki server? [12:32:28] awight: I think so, yes! [12:33:12] arturo: Thanks for the note :-) [12:33:58] :-) [13:52:13] I get the NFS error every time I try to `vagrant up`: [13:52:25] > mount.nfs: mount to NFS server '192.168.122.1:/srv/mediawiki-vagrant' failed: RPC Error: Unable to receive [14:13:51] rpcbind won't start [14:26:40] > rpcbind.socket: Socket service rpcbind.service not loaded, refusing. [14:27:00] I've tried both the stretch and buster images. [14:31:48] Okay, I learned in T241710 that rpcbind is deprecated, so something else is wrong with NFS. [14:57:48] awight: mostly the Vagrant code for managing LXC + NFS setup is just a bit flakey. I've been deep into the rabbit hole of figuring out why and never come back with a convincing answer. [15:00:27] He has brunched with the Mad Hatter. [15:00:39] on many occasions :) [15:01:17] bd808: I've seen lots of local failures when using vagrant+virtualbox+NFS, but I'm getting 100% failure on my WMCS instance. Am I normal? [15:01:20] Strangely it seems to become "eventually stable" in my tests in the past. Sometimes it takes 5 reboots, but once it starts working it seems to stay working [15:01:25] O_O [15:02:04] I haven't setup a new mwv vm in several months but that was my past experience [15:02:21] And just to triple-check, role::labs::mediawiki_vagrant is currently the lowest-maintenance method for running a mediawiki server? [15:03:06] Perhaps unsurprisingly I haven't put work onto mwv for a long time now. I got really tired of maintaining something that only got complaints [15:03:57] awight: that is personal preference I guess. It is the only role I know of that tries to setup a full MediaWiki stack [15:07:53] :-) Sorry to contribute to the "complaints" column. I've also benefitted enormously from mwv, it's probably saved me 100 hours and two handfuls of hair. [15:16:56] I had a qn. about node versions on labs vms. on onf the vms, we have node v6 just like production. what is the recommended way of upfrading it to say v12? [15:17:48] subbu: "labs" is gone [15:17:59] oh sorry :) [15:18:00] cloud vm [15:18:08] old habits die hard. [15:19:47] so, back to the real question... I guess the "recommended" way would be to find a deb source which provides the version you want or compile it yourself [15:20:35] https://packages.debian.org/search?keywords=nodejs -- shows v12 in Debian sid [15:21:12] ok .. i suppose i can actually install whatever i want on a cloud vm. [15:21:12] It might be possible for someone good at deb packaging to backport that [15:21:31] yes, as long as the things you install are FOSS [15:21:35] right, right. [15:23:00] subbu: if this is for a project that you hope to take to prod at some point, you might want to try and get moritzm or someone else from the SRE team interested in your backport problem [15:23:18] no, this is primarily testing only. [15:23:28] if it is just for some messing around, then smash it into your instance however you can I guess [15:23:40] and production uses will depend on everyone else's uses of node. [15:25:17] looks like all I need is to get npm installed there .. and then i can install the n package with which to manage node versions. [15:27:11] bd808, thanks. [16:00:14] bd808: This is probably irrelevant, but I noticed that vagrant is launching its nfs client at v3, but recent tasks mention that the vagrant hosts will support nfs v4 only. [16:00:48] awight: that sounds like a fun mismatch [16:01:12] Large breadcrumb: when I remove "-o udp", the client is able to nfs mount. [16:02:02] -- but only with mount.nfs4 and not `mount.nfs -o vers3...` [16:02:49] awight: it sounds like you have some data that is worthy of phab task. There may be something we can do with the settings[:nfs_shares] collection in Vagrantfile? [16:04:27] Good idea. I'll enphabricate [16:17:38] !log admin icinga downtime cloudcontrol[1003-1005].wikimedia.org for 1h for galera database movements [16:17:39] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Admin/SAL [16:19:19] Thanks for the pointers, I've extruded my notes into T257855. [16:19:20] T257855: Problems launching a mediawiki-vagrant instance in Cloud VPS - https://phabricator.wikimedia.org/T257855 [16:20:15] Oh look at that: root_share_options[:mount_options] << 'vers=3' if settings[:nfs_force_v3] [17:22:10] “enphabricate” sounds like a perfectly cromulent word :3 [17:30:46] I think it needs a meme :) [17:43:52] so, i am trying to run apt-get update on my cloud VM (parsing-qa-01) .. and i run into signature errors because i believe jessie repos no longer exist. as it turns out, all the sources already use stretch except one: /etc/apt/sources.list.d/openstack-mitaka-jessie.list ... [17:44:09] subbu: that [17:44:13] so, do i just blindly replace 'jessie' with 'stretch' in that file (and possibly rename it) .. or is there something else i do? [17:44:18] grr.. that's a bug. Rm the file [17:44:24] ok. [17:44:39] we tried to clean it up several times, but apparently keep missing instances [17:45:19] that fixed it. [17:45:20] thanks. [17:45:23] yw [20:05:12] what is the best way to dist upgrade from stretch to buster on a VM? run dist-upgrade on the VM or spin up a new VM and discard the old one? [20:07:50] subbu: the best way is to treat VMs as throwaway and discard and start a fresh one. doing in-place upgrade is also possible but these will look like they are still stretch at some point to cleanup scripts because they are still based on the stretch image from openstack point of view [20:10:48] ok. [20:14:22] I already got a node v10. So, only reason to upgrade is to get npm .. but, as i am looking more, the VM in question parsing-qa-01 is a custom flavor for parsoid QA testing and we don't have enough capacity to spin up another instance with that flavor .. so, will have to drop first and then build a new one .. given that, I think i should try the distro upgrade ... or just continue with stretch and install npm using a different route ... or, [20:14:22] continue using branches in the visualdiff repo to check in node modules so i don't have to run npm on the VM ... status quo is less work ... hmm :) [20:26:45] too many choices. i'll go with status quo for today .. and pick this up another day. i've done enough yak shaving for now. :) [20:41:23] subbu: or ask for quota bump to be able to create one more [20:41:29] ssastry@parsing-qa-01:~$ node -v [20:41:30] v10.15.2 [20:41:30] ssastry@parsing-qa-01:~$ npm -v [20:41:30] 5.8.0 [20:41:30] ssastry [20:41:52] the npm and node packages working together and using backports is also in that testreduce class [20:42:00] npm -g install npm :) [20:42:16] i just used the distro update without doing a full distro upgrade yet .. just switched repos in sources and reinstalled nodejs and npm [20:42:23] repos from stretch to buster [20:42:58] mutante you added one more choice to my list of choices (reg ask for quota bump) :-) [20:43:07] yea, editing sources.list , apt-get update, upgrade, dist-upgrade, reboot ... should do it. the do-dist-upgrade -p thing is Ubuntu-only [20:44:04] subbu: is this its own project or deployment-prep? maybe you can find an old one to delete instead [20:44:13] * mutante stops adding more options now [20:44:17] lol [20:46:07] subbu: you can also use https://gerrit.wikimedia.org/g/integration/npm to get an updated npm [20:48:02] i got npm installed there now, so, all good. the problem was that i needed node v10 and that isn't available in stretch .. so i added stretch-backports on moritz's recommendation but then npm wasn't installable because it conflicted with other packges .. so i had to upgrade to buster before i could install npm. and that chain is now complete. :) [20:48:43] ACK, that's also what the testreduce class fixes with the APT pinning config and stuff [21:04:15] !log create codesearch codesearch.wmcloud.org dns proxy (T256206) [21:04:15] Amir1: Unknown project "create" [21:04:16] T256206: Switch codesearch to codesearch.wmcloud.org - https://phabricator.wikimedia.org/T256206 [21:04:29] !log codesearch create codesearch.wmcloud.org dns proxy (T256206) [21:04:31] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Codesearch/SAL [22:12:36] !log redirects Add codesearch.wmflabs.org (T256206) [22:12:38] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Redirects/SAL [22:12:39] T256206: Switch codesearch to codesearch.wmcloud.org - https://phabricator.wikimedia.org/T256206 [22:25:41] legoktm, Amir1: https://codesearch.wmflabs.org should redirect now (might take a few minutes for DNS to propagate if you have done the lookup recently) [22:28:07] woot :D [22:28:09] works for me [23:06:41] Thank you so much!