| [12:47:21] | * v20th has quit (Quit: Leaving) |
| [15:51:53] | * s-mutin has quit (Read error: Connection reset by peer) |
| [15:52:17] | * s-mutin has joined #aegir |
| [15:54:38] | * s-mutin has quit (Read error: Connection reset by peer) |
| [15:55:05] | * s-mutin has joined #aegir |
| [16:15:36] | * s-mutin has quit (Read error: Connection reset by peer) |
| [16:16:02] | * s-mutin has joined #aegir |
| [16:19:45] | * s-mutin has quit (Read error: Connection reset by peer) |
| [16:20:08] | * s-mutin has joined #aegir |
| [16:21:06] | * s-mutin has quit (Read error: Connection reset by peer) |
| [16:21:33] | * s-mutin has joined #aegir |
| [19:26:27] | * roycroft has quit (Ping timeout: 240 seconds) |
| [00:21:46] | * ergonlog1c has joined #aegir |
| [00:36:40] | * v20th has joined #aegir |
| [00:46:48] | * roycroft has joined #aegir |
| [01:07:38] | * shaneonabike has joined #aegir |
| [01:10:33] | * shaneonabike has left #aegir () |
| [01:47:55] | * ybabel has joined #aegir |
| [02:18:12] | * theMusician has joined #aegir |
| [03:05:21] | * theMusician has quit (Ping timeout: 268 seconds) |
| [03:08:22] | * theMusician has joined #aegir |
| [03:47:50] | * fatguylaughing has joined #aegir |
| [04:03:28] | <colan> | helmo: still around? |
| [04:04:21] | <colan> | on one upgrade, it's hanging on "Platforms path /var/aegir/platforms is writable." |
| [04:05:36] | <colan> | not too sure how to debug that. running "env DPKG_DEBUG=yes sudo apt full-upgrade" doesn't actually produce additional output. |
| [04:05:47] | <colan> | ergonlog1c: ^^^ |
| [04:07:32] | <colan> | bgm: ^^^ |
| [04:12:29] | <bgm> | strace the process? |
| [04:12:48] | <bgm> | although i've run into that, usually some weird drupal cache or inifite loop :( |
| [04:13:03] | <bgm> | ex: when hosting_https would cause infinite loops |
| [04:13:09] | <bgm> | but haven't seen that on recent upgrades |
| [04:13:57] | * ybabel has quit (Quit: ybabel) |
| [04:15:44] | <colan> | bgm: what does this mean? |
| [04:15:44] | <colan> | pselect6(40, [0 37 39], NULL, NULL, {0, 50000000}, {[], 8}) = 0 (Timeout) |
| [04:15:44] | <colan> | wait4(2727, 0x7ffe05af7cec, WNOHANG, NULL) = 0 |
| [04:15:57] | <colan> | Just keeps repeating over & over. |
| [04:16:12] | * colan isn't much of a sysadmin |
| [04:22:26] | * theMusician has quit (Quit: theMusician) |
| [04:24:49] | * theMusician has joined #aegir |
| [04:37:25] | <bgm> | hmm pselect() is for I/O on file descriptors |
| [04:38:01] | <bgm> | so i guess it's not an infinite loop in the code, although really hard to tell |
| [04:50:00] | <colan> | "sudo lsof -p 2982 -ad 0,37,39" gives me: |
| [04:50:00] | <colan> | COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME |
| [04:50:01] | <colan> | apt 2982 root 0u CHR 136,3 0t0 6 /dev/pts/3 |
| [04:50:01] | <colan> | apt 2982 root 37u CHR 5,2 0t0 86 /dev/ptmx |
| [04:50:01] | <colan> | apt 2982 root 39r FIFO 0,10 0t0 21343 pipe |
| [04:50:01] | <colan> | ...which I'm not sure what to do with. |
| [05:06:03] | <viashimo> | is apt waiting for user input for some reason? |
| [05:48:18] | <helmo42[m]> | Could that be hosting-pause? It could be waiting for a task to complete |
| [06:09:12] | <ergonlog1c> | colan: I think you need "env DPKG_DEBUG=developer ..." |
| [06:09:41] | <ergonlog1c> | someone else metioned a similar issue recently on upgrading to 3.9 |
| [06:10:15] | <ergonlog1c> | so this might be caused by a change in the OS |
| [06:13:51] | <colan> | viashimo: nope. |
| [06:14:01] | <colan> | helmo: i never used it so i doubt it. |
| [06:14:15] | <colan> | ergonlog1c: still no extra output. |
| [06:15:01] | <colan> | what's weird is this worked fine on 5 aegirs, including staging for the prod site i'm working on. it's low traffic, but man, that's annoying. |
| [06:15:23] | <ergonlog1c> | http://docs.aegirproject.org/en/3.x/install/#8-troubleshooting-the-install |
| [06:15:28] | <ergonlog1c> | try apt-get |
| [06:17:42] | * theMusician has quit (Quit: theMusician) |
| [06:20:56] | <colan> | ergonlog1c: that's doesn't work at all. i get: |
| [06:20:56] | <colan> | The following packages have unmet dependencies: |
| [06:20:56] | <colan> | aegir : Depends: aegir-provision (>= 1.12) but it is not going to be installed |
| [06:20:56] | <colan> | Depends: aegir-hostmaster (>= 1.12) but it is not going to be installed |
| [06:21:41] | <colan> | "env DPKG_DEBUG=developer sudo apt-get dist-upgrade" and "env DPKG_DEBUG=developer sudo apt full-upgrade" will re-run, but then get stuck at the same spot, with no extra output. |
| [06:31:40] | <colan> | ergonlog1c: the last command the shows up in /tmp/drush_blah is "SHOW TABLES". So looks similar to https://www.drupal.org/node/2773223 |
| [06:31:41] | <hefring> | https://www.drupal.org/node/2773223 => Aegir Upgrade Stuck After Platforms Path Is Writable [#2773223] => 4 comments, 1 IRC mention |
| [06:41:34] | <roycroft> | hey folks |
| [06:41:37] | <roycroft> | a quick question, if i may |
| [06:41:47] | * theMusician has joined #aegir |
| [06:41:51] | <roycroft> | on a hosting only server what do i need to instal for the maegir master to push sites to it? |
| [06:42:07] | <ergonlog1c> | colan: "aegir3", rather than jut "aegir" (the docs need an update there |
| [06:42:09] | <roycroft> | aegir3 and aegir3-rpovision? |
| [06:42:15] | <roycroft> | provision, rather |
| [06:42:18] | <roycroft> | or just aegir3? |
| [06:42:45] | <ergonlog1c> | roycroft: aegir3-clusternode, iirc |
| [06:42:51] | <roycroft> | oh |
| [06:42:54] | <roycroft> | so none of the above :) |
| [06:43:26] | <ergonlog1c> | roycroft: sorry, that's aegir3-cluster-slave |
| [06:43:41] | <viashimo> | colan: I guess one thing looks kinda weird to me with the unmet dependencies: usually the new packages are aegir3-X right? |
| [06:43:41] | <roycroft> | right, i just determined that with apt-cache search |
| [06:44:12] | <roycroft> | that in addition to aegir3 and aegir3-provision? |
| [06:44:30] | <ergonlog1c> | roycroft: no |
| [06:44:31] | <roycroft> | i know i'm not going to install aegir3-hostmaster on the production machines |
| [06:44:34] | <roycroft> | ok, thanks |
| [06:44:41] | * roycroft will be doing that stuff after lunch |
| [06:44:48] | <ergonlog1c> | you don't need much more that a web server on remote, generally |
| [06:44:51] | <roycroft> | right |
| [06:44:58] | <roycroft> | web server, mysql, rsync |
| [06:45:12] | <roycroft> | and an aegir account for the hostmaster use for pushing things |
| [06:45:35] | <roycroft> | and php of course :) |
| [06:45:37] | <ergonlog1c> | you can pre-seed the SSH key: http://cgit.drupalcode.org/provision/plain/debian/aegir3-cluster-slave.t... |
| [06:46:04] | <colan> | ergonlog1c: ok, thanks, runs with aegir3, but still no extra output. tried both "yes" and "developer". |
| [06:46:07] | <roycroft> | probably drush as well would be good |
| [06:46:22] | <ergonlog1c> | roycroft: not really |
| [06:46:52] | <ergonlog1c> | colan: well, that in itself is odd |
| [06:47:05] | <roycroft> | my boss likes to change the admin password for websites and then forget it |
| [06:47:41] | <roycroft> | drush can reset it just like it does on the aegir master, can't it? |
| [06:47:53] | <roycroft> | i know i can do that in the database as well, but drush is easier |
| [06:48:31] | <roycroft> | and since i have to reset them for him all the time i like easy :) |
| [06:48:57] | <roycroft> | but i do like to keep production machines as lean as possible |
| [06:49:07] | <roycroft> | so maybe i'll skip drush, at least for now |
| [06:49:51] | <ergonlog1c> | roycroft: Aegir runs a hub&spoke network model. So all the tasks are run locally and then pushed out tot he remote |
| [06:49:58] | <ergonlog1c> | the remotes are "dumb" |
| [06:50:21] | <colan> | ergonlog1c: is it possible that env won't work on stable / non-dev installs? |
| [06:50:32] | <ergonlog1c> | for example, when you build a platform on a remote, Aegir will build it locally on the master, then rsync it out to the remote |
| [06:51:11] | <ergonlog1c> | colan: no... it's an option passed to the debian package that should make everything much more verbose |
| [06:51:53] | <ergonlog1c> | roycroft: so, generally, you don't want to grant access to remotes, since changes there may well get overwritten |
| [06:52:37] | <ergonlog1c> | "drush @sitename uli" would generally only work on the master server, iirc |
| [06:53:14] | <ergonlog1c> | I know there was some work last year to make stand-alone remotes the canonical source, and so rsync back to the master after initial deployment |
| [06:53:29] | <ergonlog1c> | but I don't recall ever using that functionality myself |
| [06:54:19] | <ergonlog1c> | I usually only ever run remotes as clusters for high-availability and performance |
| [07:02:58] | <ergonlog1c> | colan: could it be a utf8mb4 issue? maybe mysql default config changed or something |
| [07:04:05] | <helmo> | colan: still a hanging upgrade? |
| [07:04:19] | <colan> | helmo: yeah :( |
| [07:04:41] | <colan> | ergonlog1c: well, it worked on staging... |
| [07:05:04] | <helmo> | when you look at the task queue ... is there anything in there? |
| [07:06:07] | <helmo> | I'v had it before on a box where it turned out that the queued hadwas not running ... so a few tasks were pening and the upgrade was waiting for them to finish |
| [07:07:51] | <colan> | helmo: there was a task running, but deleting it didn't help. |
| [07:08:04] | <colan> | also restarted hosting-queued |
| [07:10:03] | * hestenet has quit () |
| [07:13:01] | <helmo> | colan: what does `drush @hostmaster php-eval "echo (hosting_task_count() + hosting_task_count_running())"` return? |
| [07:13:33] | <colan> | helmo: 1 |
| [07:14:50] | <colan> | so running() is returning 1. |
| [07:15:16] | <helmo> | colan: then your waiting for drush_hosting_pause_validate() in pause.hosting.inc |
| [07:15:45] | <colan> | helmo so how to flush? |
| [07:16:28] | <helmo> | find the task and delete it .... status HOSTING_TASK_PROCESSING (-1) or HOSTING_TASK_QUEUED (0) |
| [07:23:39] | <colan> | found a -1 and deleted it. re-running full-upgrade.... seems to have passed that spot.... |
| [07:25:46] | <colan> | helmo: thanks!! the upgrade finished, and there don't seem to be any problems. now i want to document this somewhere. |
| [07:27:34] | <colan> | Can we add a log message at that spot, something like, "Waiting for running and queued tasks to terminate..."? |
| [07:27:50] | <colan> | So if it hangs there, it'll be obvious what's going on. |
| [07:29:01] | <helmo42[m]> | It must be the level... There is such a message. |
| [07:34:14] | * ergonlog1c wonders where the "troubleshooting" section of our docs went... |
| [07:34:36] | <ergonlog1c> | colan: that's definitely a good 'troubleshooting' entry |
| [07:35:01] | <colan> | ergonlog1c: yeah, but it would go away if the log entry actually showed up. ;) |
| [07:35:59] | <colan> | how to turn drush_log() into something that's always present? |
| [07:39:14] | <colan> | hmm. 'message' doesn't shows up, but 'success' does. |
| [07:41:10] | <colan> | ergonlog1c: i think the problem is that the message shows up with debugging on, but debugging mesgs can't be turned on, at least in my case. |
| [07:57:42] | * v20th has quit (Quit: Leaving) |
| [07:58:33] | <colan> | https://www.drupal.org/node/2861696 |
| [07:58:34] | <hefring> | https://www.drupal.org/node/2861696 => Extra output via debugging messages cannot be enabled on Debian package upgrades [#2861696] => 0 comments, 1 IRC mention |
| [08:44:42] | * theMusician has quit (Quit: theMusician) |
