[03:05:51] *** Joins: ziyeyang_ (ziyeyang@nat/intel/x-izvodvbstrommorx) [03:27:23] *** Quits: ziyeyang_ (ziyeyang@nat/intel/x-izvodvbstrommorx) (Ping timeout: 260 seconds) [03:42:54] *** Joins: ziyeyang_ (~ziyeyang@192.55.54.44) [04:03:11] *** Joins: tsuyoshi (b42b2067@gateway/web/freenode/ip.180.43.32.103) [04:04:08] *** Quits: ziyeyang_ (~ziyeyang@192.55.54.44) (Ping timeout: 240 seconds) [04:42:43] *** Joins: ziyeyang_ (~ziyeyang@192.55.54.44) [05:04:00] *** Quits: ziyeyang_ (~ziyeyang@192.55.54.44) (Quit: Leaving) [05:39:25] *** Quits: tsuyoshi (b42b2067@gateway/web/freenode/ip.180.43.32.103) (Quit: Page closed) [08:35:38] i haven't yet [08:36:31] peluse: i think this configuration topic that we just discussed in the arch call would be a perfect one for you to help drive :) [08:36:44] let's chat a bit when you're in the office [08:39:24] jimharris, sounds good [08:40:33] today I'm WFH - I was planning on taking tomorrow off but some things got shuffled so I might head in, if not Mon for sure [08:43:51] sounds good - doesn't need to be today [08:59:16] *** Joins: jstern_ (~jstern@192.55.54.40) [09:27:35] those arch meetings aren't actually on my calendar [09:52:31] jimharris: in vbdev_error, if we are going to fail an I/O, does it make sense to just not submit the I/O to the underlying bdev at all? [09:52:35] (rather than doing all the child I/O stuff) [10:07:27] bwalker, did you find a time for IRC office hours yet? [11:03:30] peluse: I haven't scheduled any yet [11:04:07] I'm thinking it should probably be first thing in the morning our time [11:04:30] that works for east coast US, Europe [11:04:38] not great for APAC [11:04:58] I make no promises that I will be able to form coherent sentences before 10 AM ;) [11:05:04] bwalker, yeah, totally up to you. We'll never find a time that works for the world, just pick a time when you know, for now, that most of us don't have a conflict [11:05:37] my calendar is way unpredictable - I don't have many regular meetings [11:05:46] but I'm often in all day things that occur just once [11:05:54] so I'll take requests for times [11:06:05] peluse: I pushed a review that tweaks the unit test coverage so it only shows stuff in lib - see if that makes sense to you [11:06:10] the office hrs thing isn't meant to be a required meeting (for anyone else out there reading this), its just a time when we're aiming to have at least one maintainer online for discussion... [11:06:23] drv, cool, will look [11:06:46] should make it easier to just say "pick anything in ut_coverage and start working on it" [11:07:39] https://ci.spdk.io/builds/review/097e8ae230107ee6d54cabbb82186adb0488cbf5.1495130147/wkb-fedora-03/ut_coverage/ [11:07:49] drv, yeah that makes sense. I assume -r is for remove or something like that? [11:07:52] yeah [11:08:06] unfortunately, I don't think there's a whitelist type of function, so we just have to remove everything that we don't want [11:08:07] ... or "rubbish" maybe [11:08:10] haha [11:08:16] I like it [11:08:54] but I think we should remove modules in alphabetical order :) [11:09:36] oops, I got test in the wrong order [11:10:32] ask jim for some help if you get stuck... [11:11:01] I'll get him to sing the song while I type [11:13:49] https://www.youtube.com/watch?v=36IBDpTRVNE [11:31:01] jimharris, thought I'd try you vagrant setup on a linux box. Installed vbox, ext pack, vagrant and then sourced the script and ran vagrant up and got a network error of some kind. See while you were messing around with it? https://gist.github.com/peluse/94161f7572d56698f85745c79a98e697 [11:32:34] i never tried this on linux - i think drv or bwalker tried it but my understanding was they got stuck before this point [11:33:04] OK, there's no way I got further then either of them but thanks for saying that :) [11:33:17] I'll try on the mac side [11:33:29] does it leave the vagrant vm in a state where you can still do a "vagrant ssh"? [11:34:40] wow, yeah it does [11:35:12] does ifconfig show eth1? [11:35:24] was about to say, no its called... [11:35:30] eno1 [11:35:52] I mean: enp0s3 [11:37:21] https://github.com/mitchellh/vagrant/issues/7155 [11:37:23] similar issue [11:38:16] https://github.com/mitchellh/vagrant/commit/0505771481222b4893daa5454ff6c9c16f471fa4 [11:40:17] thats pretty old... [11:40:20] could you try applying that patch by hand to the plugin on your system? [11:40:30] wait, this is 2017 right?? [11:41:11] that patch is more than a year old but I haven't checked yet if/when it was merged upstream [11:41:25] and whatever package you have could have been spun before it was included [11:41:53] yeah, mine is way old [11:42:07] I have 1.8.1 and it was fixed in 1.9.5 [11:42:35] cool [11:43:09] will udpate and try again [11:43:20] or update, whichever works [11:48:05] my mac vagrant version is 1.9.4 which explains why I never saw this [11:48:36] running 1.9.5 now and it got passed that, not sure what its doing now [11:48:51] oh - does it look like its hung? [11:48:58] here's my last few lines. now it looks like its hung [11:48:59] ==> default: Machine booted and ready! [11:48:59] ==> default: Checking for guest additions in VM... [11:48:59] ==> default: Configuring and enabling network interfaces... [11:48:59] ==> default: Rsyncing folder: /home/peluse/spdk/ => /spdk [11:48:59] ==> default: Mounting shared folders... [11:49:01] default: /vagrant => /home/peluse/spdk/scripts/vagrant [11:49:03] ==> default: Running provisioner: shell... [11:49:05] default: Running: /tmp/vagrant-shell20170518-13981-kb4zuf.sh [11:49:09] ==> default: vm.nr_hugepages = 1024 [11:49:20] you need to set http proxies so the vm can do package updates [11:49:31] i think i mentioned that in the readme [11:49:40] * jimharris goes to check... [11:50:46] yeah - it's mentioned in the readme but maybe we need to make it more clear [11:51:03] which readme, the one in the vagrant dir or the root? [11:51:28] root [11:51:40] but it should be mentioned in both if its not already [11:51:44] hmmm, yeah I'm not seeing anything jumping out at me [11:52:00] does it start with "Hey Paul..."? [11:52:16] 1) Download DPDK as a subdirectory in the SPDK repository as described [11:52:16] above. You do *not* need to build DPDK - the Vagrant scripts will [11:52:16] do this for you. [11:52:18] 2) vagrant up [11:52:20] 3) vagrant ssh [11:52:22] If you are behind a corporate firewall, set http_proxy and https_proxy in [11:52:24] your environment before running step 2. [11:52:31] oh yes, it just timed out so its obvious now [11:54:54] hmmm, but my proxy env vars are setup OK and work just fine from the linux box running vbox [11:55:46] I'll mess with it more later, thanks [11:55:55] http_proxy is set? not just HTTP_PROXY? [11:56:07] I have just the lower case set [11:56:11] hmmm [11:56:18] https_proxy too? [11:56:22] yeah [11:56:35] ==> default: Err:1 http://us.archive.ubuntu.com/ubuntu xenial InRelease [11:56:35] ==> default: Cannot initiate the connection to us.archive.ubuntu.com:80 (2001:67c:1562::19). - connect (101: Network is unreachable) [IP: 2001:67c:1562::19 80] [11:57:06] weird [11:58:06] that's not an apt command is it? [11:58:16] that's not setup for proxy in the VM [11:58:28] vagrant should take care of all of that for you [11:58:39] yay vagrant [11:59:09] it works on my mac at least - i'll try to find some time to try this on my linux box [11:59:34] yeah, I'll keep dorking with it too [11:59:42] drv, bwalker: what was the issue one of you had with trying the vagrant setup on linux? [12:00:02] sudo apt-get update [12:00:02] <--- does not work from within the VM [12:00:42] cat /etc/apt/apt.conf [12:01:05] yeah, I just fixed it and apt works now [12:01:30] so seems like for some reason maybe http proxies weren't set up in the environment you start vagrant? [12:02:08] looks that way [12:03:09] need to make sure Vagrant.has_plugin is true I guess (in Vagrantfile) [12:03:22] for ""vagrant-proxyconf" that is [12:03:57] hmm, don't think I have it [12:05:44] installed now, trying again... [12:06:52] much better! I'l add that little gem to the readme [12:09:44] I may tweak both readmes just a little based on my first time experience too if that's cool... [12:11:25] I wasn't trying to use your vagrant, but rather build my own vagrant images for qemu [12:11:40] I think drv tried to use your setup though and ran into a problem [12:11:46] I don't remember what it was [12:12:18] peluse: please do feel free to update the readme - good to have a fresh set of eyes on it [12:12:46] my bad on the proxyconf - I forgot about that dependency [12:13:11] will do. gotta run for a few hours, latest errors are here https://gist.github.com/peluse/7b07d560ec5208e681a0b49421a49ef8 happening during build [12:42:52] I couldn't get vagrant to work on Linux with virt [12:43:09] and fedora doesn't package VirtualBox, so I didn't want to mess with that [12:43:28] where's your sense of adventure [12:44:37] I left it at home [12:47:47] not sure what happened in peluse's build error there - is part of the source code missing? [12:48:47] yeah - also not sure why it's building from /home/peluse/spdk - the vagrant set up puts all of the code at /spdk by default [12:50:31] hmm, good question [13:08:01] *** Quits: jstern_ (~jstern@192.55.54.40) (Ping timeout: 246 seconds) [13:34:08] *** Joins: Darren (c6ba0002@gateway/web/freenode/ip.198.186.0.2) [14:32:01] jimharris, I am sending out an e-mail asking for everyone's github user accounts for the new whitelist. I am sending the e-mail to NPG SW AZ and NPG SW ITP, What are the other groups we would want to add to the whitelist? [15:04:09] *** Quits: Darren (c6ba0002@gateway/web/freenode/ip.198.186.0.2) (Ping timeout: 260 seconds) [15:11:32] jimharris, if i vagrant ssh on in there and make from /spdk same error but if I first make clean then make it builds and runs UT fine [15:12:05] so put that in your pipe and smoke it! [15:21:08] looking... [15:22:15] we probably need to add an explicit "make clean" in the vagrant/build.sh file [15:22:59] could you try that? [15:25:07] will do... let ya know in a few [15:26:27] https://ci.spdk.io/builds/review/f4a59d267baca5f7919b49f01765bf4146db440c.1495145616/ [15:27:12] pretty [15:27:21] me like! [15:27:34] the automated test system is hooked up to gerrithub now [15:27:43] we still have to write up an explanation of how to use it [15:28:27] awesome! [15:28:36] http://ci.spdk.io/status/ [15:28:58] these are all available links on the public internet [15:29:43] And here's the query for all open code reviews: https://review.gerrithub.io/#/q/project:spdk/spdk+status:open [15:30:08] i think this deserves a happy hour!!! [15:30:13] absolutely [15:30:16] two happy hours! [15:30:26] do I hear 3? [15:30:38] nice work gents!!! [15:30:46] yeah - this is outstanding [15:31:17] * peluse thinks he can ditch VPN again very soon.... [15:31:32] * jimharris has heard that before [15:32:17] jimharris, adding make clean in build.sh fixed it. I'll add that to the readme updates patch, this is cool beans man! [15:32:24] reviews submitted to gerrithub have to be explicitly approved [15:32:29] unless you're on the trusted list [15:32:43] does that cost extra? [15:32:46] the trusted list is currently just myself and drv [15:33:02] normally $500, but we can cut you a special deal ;) [15:33:09] so I take bribes in cash and beer [15:33:11] 2 for 1 maybe? [15:34:48] I failed to mention that there is no permission check on the approval button beyond it only being reachable on the Intel intranet [15:35:21] jimharris, so not completely there yet... see https://gist.github.com/peluse/7b07d560ec5208e681a0b49421a49ef8 [15:36:23] hmmm [15:36:35] I'll put that in the readme :) [15:36:58] so it didn't just segfault it basically panic'd the kernel in your VM? [15:37:33] not sure how to see what happened, I can do poke around some vagrant stuff unless you already know where there is some sort of log or something somewhere [15:37:39] what I put in the gist is all I can see [15:37:48] that's what it seemed like happened to me [15:37:57] could you try stepping through hello_world in gdb? [15:38:17] sure seems like it crashed in rte_eal_init() - which means you'll have fun stepping through DPDK initailization code [15:38:42] if you want to hold off - I can spin this up on my mac tomorrow to see if something has regressed [15:38:49] so far you had to add: [15:38:56] 1) proxyconf plugin [15:39:00] 2) make clean [15:39:04] anything I'm missing? [15:40:07] yeah so far that's all I did different. I can try on my mac as well and see if it works there [15:40:23] ill try the gdb route and see if anything obvious pops up [15:46:15] hmm, was the make clean required because you had existing .d files from a previous build on your host system maybe? [15:46:49] could be [15:47:25] or maybe a .e file or two [15:50:58] .d, .e, whatever it takes [16:11:32] jimharris, hmmm, stepping through: https://gist.github.com/peluse/7b07d560ec5208e681a0b49421a49ef8 [16:31:17] jimharris, so I assume I need to run 'sudo spdk/scripts/setup.sh' first right? I do that and it works like a champ! [16:31:29] vagrant@localhost:/spdk/examples/nvme/hello_world$ sudo ./hello_world [16:31:29] Starting DPDK 17.02.0 initialization... [16:31:29] [ DPDK EAL parameters: hello_world -c 0x1 --file-prefix=spdk_pid1996 ] [16:31:29] EAL: Detected 2 lcore(s) [16:31:30] EAL: Probing VFIO support... [16:31:31] Initializing NVMe Controllers [16:31:32] Initialization complete.