[01:00:39] *** Quits: bwalker_ (~bwalker@ip70-190-226-244.ph.ph.cox.net) (Quit: Leaving) [01:13:08] *** Joins: tkulasek (~tkulasek@134.134.139.72) [01:16:12] mhae: could you create a github issue for that booting problem? [01:46:56] *** Joins: dlw1 (~Thunderbi@114.255.44.143) [01:46:57] *** Quits: dlw (~Thunderbi@114.255.44.143) (Read error: Connection reset by peer) [01:46:57] *** dlw1 is now known as dlw [04:03:49] *** Joins: tomzawadzki (~tomzawadz@134.134.139.76) [05:45:57] *** Quits: tkulasek (~tkulasek@134.134.139.72) (Ping timeout: 240 seconds) [06:33:06] *** Joins: mhae (0fd3c95d@gateway/web/freenode/ip.15.211.201.93) [08:07:33] *** Quits: lhodev (~lhodev@66-90-218-190.dyn.grandenetworks.net) (Quit: My MacBook has gone to sleep. ZZZzzz…) [08:31:05] *** Quits: tomzawadzki (~tomzawadz@134.134.139.76) (Ping timeout: 256 seconds) [08:45:02] *** Joins: felipef (~felipef@62.254.189.133) [08:47:42] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [09:05:07] *** Quits: felipef (~felipef@62.254.189.133) (Quit: Leaving...) [09:38:20] jimharris: we should get this QoS fix in for 18.04: https://review.gerrithub.io/#/c/spdk/spdk/+/409125/ [09:39:21] *** Joins: travis-ci (~travis-ci@ec2-23-20-7-207.compute-1.amazonaws.com) [09:39:22] (spdk/master) scripts/pkgdep: add suse packages (John Kariuki) [09:39:22] Diff URL: https://github.com/spdk/spdk/compare/e6acab993064...abeba4553345 [09:39:22] *** Parts: travis-ci (~travis-ci@ec2-23-20-7-207.compute-1.amazonaws.com) () [09:41:06] drv: done [09:44:40] *** Joins: travis-ci (~travis-ci@ec2-54-234-125-172.compute-1.amazonaws.com) [09:44:41] (spdk/master) blobstore: do not produce error when requesting number of clones (Tomasz Zawadzki) [09:44:41] Diff URL: https://github.com/spdk/spdk/compare/abeba4553345...882558f7e896 [09:44:41] *** Parts: travis-ci (~travis-ci@ec2-54-234-125-172.compute-1.amazonaws.com) () [09:44:53] *** Joins: travis-ci (~travis-ci@ec2-54-234-125-172.compute-1.amazonaws.com) [09:44:54] (spdk/master) bdev/qos: free the QoS channel when disabling (GangCao) [09:44:54] Diff URL: https://github.com/spdk/spdk/compare/882558f7e896...2f94b4d4e360 [09:44:54] *** Parts: travis-ci (~travis-ci@ec2-54-234-125-172.compute-1.amazonaws.com) () [09:45:29] I'm here for a little while if anything needs my immediate attention [09:46:01] bwalker: can you take a look at the last couple of QoS fixes? I think they're OK, but it would be good to have your review [09:46:09] I saw them all - look good [09:46:16] are they all merged already? [09:46:17] bwalker: and also ziye's last comment on https://review.gerrithub.io/#/c/spdk/spdk/+/407743/ [09:46:23] yes, https://github.com/spdk/spdk/commit/804ebf9985e93697809c7e09cc732b0bfcf0c19f and https://github.com/spdk/spdk/commit/2f94b4d4e3607400419068eb74b64221df93ea15 [09:46:40] yeah those look fine to me [09:50:33] ziye's patch looks like it will work - I think using spdk_for_each_channel is going to be more robust in general [09:50:40] but what he has there looks functional to me [10:07:59] *** Joins: travis-ci (~travis-ci@ec2-54-87-71-4.compute-1.amazonaws.com) [10:08:00] (spdk/master) nvmf: disconnect other connections if admin qpair disconnects. (Ziye Yang) [10:08:00] Diff URL: https://github.com/spdk/spdk/compare/2f94b4d4e360...adfa9f6dc75d [10:08:00] *** Parts: travis-ci (~travis-ci@ec2-54-87-71-4.compute-1.amazonaws.com) () [10:10:27] do I remember some chatter about putting a human readable name in the io_device structure? Sure would be handy I think in debug - in my fairly basic setup I have over 50 io_devices on the global list after init [10:13:40] *** Joins: travis-ci (~travis-ci@ec2-23-20-7-207.compute-1.amazonaws.com) [10:13:41] (spdk/master) test/iscsi: utilize network namespaces in iSCSI tests (Tomasz Zawadzki) [10:13:42] Diff URL: https://github.com/spdk/spdk/compare/adfa9f6dc75d...5bafc240e5ae [10:13:42] *** Parts: travis-ci (~travis-ci@ec2-23-20-7-207.compute-1.amazonaws.com) () [10:22:44] peluse: sounds like a reasonable idea; it would probably mean an API change, though [10:23:55] yeah, probably not worth doing that. I may add some debug functions that can be called from gdb to help traverse all the io_devices and channels, etc. Will see how it goes, doing it all manually right now and may not prove to be worth anything other than learning... will see [10:26:43] i think that API change might be worth it though [10:27:20] yeah [10:28:58] ahh, OK. cool. I'll finish my current manual debug session which may take all freaking day and then I'll look at doing that as a separate patch [10:40:06] the question is whether we want the name to be unique or not - i would say it does not have to be unique but could be convinced otherwise [10:40:30] for example, if you have 8 NVMe SSDs, you can just use "nvme_ctrlr" for all of them [10:40:52] yes, I think it should just be a debugging aid, not checked for uniqueness or used for any other purpose [10:41:06] instead of forcing the caller to come up with their own unique name - it's likely whatever they come up with won't really help, they'll want "nvme_ctrlr" and then the pointer value for the io_device [10:41:13] if you're debugging [10:45:09] *** Quits: mhae (0fd3c95d@gateway/web/freenode/ip.15.211.201.93) (Ping timeout: 260 seconds) [10:54:58] same page w/both of you. My only intended use was for debug so whoever is using it can jam whatever they want in there [11:03:52] bwalker, jimharris: this blobstore splitting fix looks important: https://review.gerrithub.io/#/c/spdk/spdk/+/409201/ [11:04:10] also looks like a great situation for a unit test [11:26:02] agreed - can you send Tomek an e-mail asking for a unit test? [11:26:08] ok [11:28:31] do we want that patch for 18.04? It brings contiguous memory requirement from ~5MB down to ~3MB https://review.gerrithub.io/c/spdk/spdk/+/408925 (event/reactor: take into account 1 extra slot in event mempools) [11:31:09] (just asking, no pressure) [11:31:20] seems OK to me [11:31:43] we already had the magic 262144 number, I guess [11:37:21] was it calculated somehow? [11:37:32] git blame shows some code movement and i didn't dig deeper [12:06:54] no idea where it came from originally [12:07:17] it was introduced with the original import of the event library (modified later to divide by socket_count) [13:18:07] *** Quits: sethhowe (~sethhowe@192.55.54.42) (Remote host closed the connection) [13:21:04] *** Joins: sethhowe (sethhowe@nat/intel/x-otendhloasalpjox) [13:24:24] *** Quits: sethhowe (sethhowe@nat/intel/x-otendhloasalpjox) (Remote host closed the connection) [13:25:43] some changelog updates for 18.04: https://review.gerrithub.io/#/c/spdk/spdk/+/409223/ [13:25:53] if there's any other major feature additions, please let me know [13:25:58] *** Joins: sethhowe (~sethhowe@192.55.54.42) [13:29:56] fixed my completion after channel destroy... my bug, all of 5 characters to fix it :( Was crossing channels in the write path, using the wrong one in a parm list [13:31:36] *** Joins: travis-ci (~travis-ci@ec2-23-20-7-207.compute-1.amazonaws.com) [13:31:37] (spdk/master) blobstore: fix payload offset incrementation on operation split (Tomasz Kulasek) [13:31:37] Diff URL: https://github.com/spdk/spdk/compare/5bafc240e5ae...5557a572f5f4 [13:31:37] *** Parts: travis-ci (~travis-ci@ec2-23-20-7-207.compute-1.amazonaws.com) () [14:13:25] *** Quits: lhodev (~lhodev@inet-hqmc07-o.oracle.com) (Remote host closed the connection) [15:58:50] *** Joins: lhodev (~lhodev@66-90-218-190.dyn.grandenetworks.net) [16:43:30] *** Joins: travis-ci (~travis-ci@ec2-54-234-125-172.compute-1.amazonaws.com) [16:43:31] (spdk/master) bdev/qos: No operation for zero QoS setting on not enabled bdev (GangCao) [16:43:31] Diff URL: https://github.com/spdk/spdk/compare/5557a572f5f4...409353a23e0d [16:43:31] *** Parts: travis-ci (~travis-ci@ec2-54-234-125-172.compute-1.amazonaws.com) () [16:55:26] bwalker, ziyeyang: this nvmf-related failure just happened on master: https://ci.spdk.io/spdk/builds/release/master/5042/fedora-06/build.log [16:55:47] pretty sure this is related to "nvmf: disconnect other connections if admin qpair disconnects." - nvmf_qpair_fini is in the call stack [16:58:25] I think we should revert that for the release, unless there is an obvious bugfix [16:59:41] *** Joins: travis-ci (~travis-ci@ec2-23-20-7-207.compute-1.amazonaws.com) [16:59:42] (spdk/master) changelog: add bdev QoS and iSCSI VPP support (Daniel Verkamp) [16:59:43] Diff URL: https://github.com/spdk/spdk/compare/409353a23e0d...145b2f7bd323 [16:59:43] *** Parts: travis-ci (~travis-ci@ec2-23-20-7-207.compute-1.amazonaws.com) ()