[00:19:01] *** Joins: gila (~gila@5ED74129.cm-7-8b.dynamic.ziggo.nl) [02:16:31] *** Joins: travis-ci (~travis-ci@ec2-54-198-133-189.compute-1.amazonaws.com) [02:16:32] (spdk/v18.10.x) nvmf: check the qpair->ctrlr (Ziye Yang) [02:16:32] Diff URL: https://github.com/spdk/spdk/compare/2f1ac3644a4d...86dbefdda6d8 [02:16:32] *** Parts: travis-ci (~travis-ci@ec2-54-198-133-189.compute-1.amazonaws.com) () [02:21:16] *** Joins: travis-ci (~travis-ci@ec2-35-175-246-169.compute-1.amazonaws.com) [02:21:17] (spdk/v18.10.x) SPDK 18.10.1 (Tomasz Zawadzki) [02:21:17] Diff URL: https://github.com/spdk/spdk/compare/86dbefdda6d8...df70a66ff40a [02:21:17] *** Parts: travis-ci (~travis-ci@ec2-35-175-246-169.compute-1.amazonaws.com) () [02:25:19] *** Joins: travis-ci (~travis-ci@ec2-54-205-80-185.compute-1.amazonaws.com) [02:25:20] (spdk/v18.10.1) SPDK 18.10.1 (Tomasz Zawadzki) [02:25:20] Diff URL: https://github.com/spdk/spdk/compare/v18.10.1 [02:25:20] *** Parts: travis-ci (~travis-ci@ec2-54-205-80-185.compute-1.amazonaws.com) () [03:21:36] *** Joins: travis-ci (~travis-ci@ec2-54-163-139-160.compute-1.amazonaws.com) [03:21:37] (spdk/v18.10.x) version: 18.10.2 pre (Tomasz Zawadzki) [03:21:37] Diff URL: https://github.com/spdk/spdk/compare/df70a66ff40a...a516dce2afd4 [03:21:37] *** Parts: travis-ci (~travis-ci@ec2-54-163-139-160.compute-1.amazonaws.com) () [07:46:41] darsto, please take a look at https://review.gerrithub.io/c/spdk/dpdk/+/437886 when you can. There are many patches waiting for this - not urgently but it would be great to get the compression deps on master sooner than later [08:07:54] Hi peluse: Saw my addition of reviewer to https://review.gerrithub.io/437232. Also noted emails closing 18.10.1 merge window announcing release of 18.10.1. Nonetheless, is this aforementioned change (on master) something you're thinking may get backported to v18.10.x? Perhaps a 18.10.2 release? [08:08:38] Also, a pull did not reveal the presence (yet) of a v18.10.1 tag. [08:11:25] Actually, the v18.10.1 tag *does* exist when pulling from github. It doesn't appear on a pull back'd by gerrithub. [08:15:14] lhodev, not implying anything about release just curious if this patch affects any of your attempts to repro - are you still in that mode where it happened easily but now sorta stopped? [08:56:48] *** Joins: travis-ci (~travis-ci@ec2-54-160-155-226.compute-1.amazonaws.com) [08:56:49] (spdk/master) test: Use ut_multithread framework in blobfs async test (Ben Walker) [08:56:50] Diff URL: https://github.com/spdk/spdk/compare/c9c4a3346d7a...b035e2b84852 [08:56:50] *** Parts: travis-ci (~travis-ci@ec2-54-160-155-226.compute-1.amazonaws.com) () [09:04:01] peluse: should we update our pkgdep.sh to install ipsec unconditionally then? [09:29:39] peluse: I'm experimenting more this morning without that added patch. Too early to tell at this point. [09:31:14] darsto, I don't think we want to do that. With the proposed soln in that patch we simply always build the compressdev and cryptodev APIs which have no external deps so there's no reason to make everyone install ipsec and isal just for fun :) [09:31:46] we'll keep the dpdk makefile conditional CONFIG* switches for the PMD for compress and crypto as those are the ones that have the ext deps [09:32:19] and the reason we have to split it out like this is because of a build order issue on the DPDK side, if you try to build both at the same time it pukes unless the APIs are built first which we can't control from the makefile.... [09:33:03] lhodev, OK cool. Keep in mind that patch is still just some experiments, lots of comments still ongoing there so don't spend a huge amount of time on it as if it were the soln.... [09:33:25] gotcha [09:36:40] peluse: so i guess if it doesn't increase the build time significantly then it's good to go [09:40:51] darsto, yeah it doesn't. I didn't stopwatch it but those 2 modules are pretty small [09:41:32] +1 [09:41:36] +2, I mean [10:20:36] LOL [10:34:18] bwalker, or jimharris if one of you could take a look that's be great. Once it lands I'll push a patch to move the submodule.... https://review.gerrithub.io/c/spdk/dpdk/+/437886 [10:37:48] *** Joins: travis-ci (~travis-ci@ec2-35-175-246-169.compute-1.amazonaws.com) [10:37:49] (spdk/master) rpc: reconcile options for construct_nvme_bdev (Seth Howell) [10:37:50] Diff URL: https://github.com/spdk/spdk/compare/b035e2b84852...cc7a099c0c47 [10:37:50] *** Parts: travis-ci (~travis-ci@ec2-35-175-246-169.compute-1.amazonaws.com) () [10:41:22] done and merged [10:59:33] *** Joins: travis-ci (~travis-ci@ec2-54-156-114-1.compute-1.amazonaws.com) [10:59:34] (spdk/master) util/crc16: Add spdk_crc16_t10dif_copy to use in read strip and write insert (Shuhei Matsumoto) [10:59:34] Diff URL: https://github.com/spdk/spdk/compare/cc7a099c0c47...3947bc2492c7 [10:59:34] *** Parts: travis-ci (~travis-ci@ec2-54-156-114-1.compute-1.amazonaws.com) () [12:35:19] OK, this small series has the compression deps as well as resolves the DPDK issues with building one, the other or both. The 2nd won't pass on all Jenkins TP machines yet due to missing deps. [13:02:24] *** Joins: travis-ci (~travis-ci@ec2-54-160-155-226.compute-1.amazonaws.com) [13:02:25] (spdk/master) nvmf: add the transport shared buffer num configuration option. (Ziye Yang) [13:02:25] Diff URL: https://github.com/spdk/spdk/compare/3947bc2492c7...58f1624497c6 [13:02:25] *** Parts: travis-ci (~travis-ci@ec2-54-160-155-226.compute-1.amazonaws.com) () [14:08:53] *** Quits: darsto (~darsto@89-78-174-111.dynamic.chello.pl) (Remote host closed the connection) [14:15:12] sethhowe: While continuing to test with test/nvmf/fio/nvmf_fio.py, I cannot deny I remained bothered by the 'fio: verification read phase will never start because write phase uses all of runtime' messages. [14:15:12] I did some digging. [14:16:13] This page I think suggests to me that verification isn't (likely?) being done: https://github.com/axboe/fio/issues/322 [14:16:15] oh - that's just because it's using a nonsensical fio configuration file [14:16:54] Reading through that, I altered the 'verify_template' variable in nvmf_fio.py by adding: verify_backlog=8 [14:17:19] Now, upon launch, I'm not seeing that warning. [14:21:25] If I understand correctly, that directs fio to perform reads to verify after every 8 blocks have been written. [14:21:50] yeah [14:22:05] if you want to submit a patch for that, it'd go right in [14:24:59] I'll play with this a while. I want to see if by varying that whether or not I see any I/O errors of the sort: fio: io_u error on file /dev/nvme0n4: Input/output error: write offset=8126464, buflen=262144. [14:25:19] I suspect we won't, but nonetheless I just want to play this out for some iterations. [14:27:02] bwalker: Do you think the error message cited above occurred as a result of the RDMA connection somehow getting dropped? [14:28:50] yeah that's exactly what happened [14:29:03] the target also sets the qpair to an error state, which occurred because the RDMA connection dropped [14:29:21] the challenge is figuring out why the RDMA connection dropped [16:53:31] *** Quits: gila (~gila@5ED74129.cm-7-8b.dynamic.ziggo.nl) (Quit: My Mac Pro has gone to sleep. ZZZzzz…) [17:10:31] *** Joins: Shuhei (caf6fc61@gateway/web/freenode/ip.202.246.252.97) [20:09:35] *** Quits: Shuhei (caf6fc61@gateway/web/freenode/ip.202.246.252.97) (Ping timeout: 256 seconds) [21:00:24] *** Joins: ziyeyang_ (~ziyeyang@192.55.46.46) [21:02:26] *** Quits: ziyeyang_ (~ziyeyang@192.55.46.46) (Client Quit) [21:03:02] *** Joins: ziyeyang_ (ziyeyang@nat/intel/x-mzyfmrmkayjorwkr) [23:07:38] *** Quits: ziyeyang_ (ziyeyang@nat/intel/x-mzyfmrmkayjorwkr) (Quit: Leaving) [23:33:55] *** Joins: gila (~gila@5ED74129.cm-7-8b.dynamic.ziggo.nl) [23:39:33] *** Quits: gila (~gila@5ED74129.cm-7-8b.dynamic.ziggo.nl) (Quit: Textual IRC Client: www.textualapp.com)