[00:03:13] tom__: can you post full build log? [00:03:17] *** Joins: dlw1 (~Thunderbi@114.255.44.139) [00:04:36] *** Quits: dlw (~Thunderbi@114.255.44.143) (Ping timeout: 268 seconds) [00:04:37] *** dlw1 is now known as dlw [00:17:05] If i just copy paste it here it will be a mess [00:22:32] tom__: use a website like pastebin.com [00:25:51] darsto here is the link to pastebin https://pastebin.com/f6Lg6F29 [00:27:20] your cpu might be missing some instruction sets [00:27:24] what's your cpu? [00:28:57] its Intel® Core™ i5-3470S CPU but i am running this vm on qemu [00:31:57] this cpu doesn't seem to have ssse3 [00:32:31] dpdk won't compile on unsupported cpus since it always compiles with -march=native [00:34:33] Ok, i was able to compile dpdk 18.05 separately, i got these errors while compiling spdk. [00:34:49] is there any workaroud for this / [00:38:43] still, it's dpdk limitation. Errors come from inline code in header files, so you may not encounter them while compiling dpdk itself [00:42:27] if it's just for testing or development purposes, you could rewrite rte_memcpy.h to use simple memcpy instead of ssse3 calls [00:42:35] you'll probably sacrifice a lot of performance though [00:43:59] hey darsto i was able to compile spdk. All i had to do is add "-cpu qemu64,+ssse3,+sse4.1,+sse4.2,+x2apic" flag while starting my qemu [00:44:38] thank you so much for your help :) [00:44:54] ahh so it's a vm [00:44:59] i can't read as always [00:45:23] :) [00:50:06] :) The main part was pointing out the missing ssse3. Thanks !! [00:56:23] *** Joins: tkulasek (~tkulasek@192.55.54.44) [00:56:24] *** Joins: tkulasek_ (~tkulasek@192.55.54.44) [01:04:24] *** Quits: tom__ (b649cc4a@gateway/web/freenode/ip.182.73.204.74) (Quit: Page closed) [03:58:34] *** Quits: dlw (~Thunderbi@114.255.44.139) (Ping timeout: 264 seconds) [04:13:01] *** Quits: darsto (~darsto@89-68-12-72.dynamic.chello.pl) (Ping timeout: 256 seconds) [04:17:37] *** Joins: darsto (~darsto@89-68-12-72.dynamic.chello.pl) [06:02:59] *** Joins: johnmeneghini (~johnmeneg@216.240.30.5) [06:26:19] *** Joins: dlw (~Thunderbi@114.252.43.105) [07:23:40] *** Quits: dlw (~Thunderbi@114.252.43.105) (Quit: dlw) [07:36:50] *** Quits: lhodev (~lhodev@66-90-218-190.dyn.grandenetworks.net) (Quit: My MacBook has gone to sleep. ZZZzzz…) [08:49:45] *** Joins: travis-ci (~travis-ci@ec2-54-196-204-91.compute-1.amazonaws.com) [08:49:46] (spdk/master) nvme: factor out request timeout checking (Daniel Verkamp) [08:49:46] Diff URL: https://github.com/spdk/spdk/compare/3d9b3e972a09...8b158aaacdec [08:49:46] *** Parts: travis-ci (~travis-ci@ec2-54-196-204-91.compute-1.amazonaws.com) () [09:10:10] *** Joins: travis-ci (~travis-ci@ec2-54-196-204-91.compute-1.amazonaws.com) [09:10:11] (spdk/master) nvmf: Correctly unregister nvmf tgt as an I/O device (Ben Walker) [09:10:11] Diff URL: https://github.com/spdk/spdk/compare/5eb33f0a5fe1...8c9508f9b330 [09:10:11] *** Parts: travis-ci (~travis-ci@ec2-54-196-204-91.compute-1.amazonaws.com) () [09:51:08] *** Joins: JoeGruher (86868b4b@gateway/web/freenode/ip.134.134.139.75) [09:53:15] LD_PRELOAD=/examples/nvme/fio_plugin/fio_plugin fio [09:53:28] This is an environment variable? Or goes in the FIO config file? [09:53:58] environment variable [09:54:05] thanks [09:54:09] you can also possibly launch the SPDK fio plugin as a regular fio plugin [09:54:19] by specifying ioengine=/path/to/plugin.so [09:54:22] in the config file [09:54:32] we recommended LD_PRELOAD for awhile because fio had a bug when unloading [09:54:34] plugins [09:54:37] oh, and then I don't need the environment variable? that seems easier [09:54:47] not sure if it still does or not [09:54:56] haha ok i guess i'll find out [09:56:40] if i've already built SPDK but didn't include the --with-fio option at the time i'll have to build it again? [09:56:58] yes [09:57:05] K [09:57:07] otherwise it didn't build the plugin [09:57:19] yeah i see that [09:57:26] i thought perhaps i could just go to /spdk/examples/nvme/fio_plugin/ and run make [09:57:30] since there's a makefile in there [09:57:35] that might work actually [09:57:52] but easier to just rebuild from the top - it won't rebuild things that you've already built [09:58:18] and the build system parallelizes nicely - so run make -j [09:58:21] or slightly more [09:58:22] you'd also need to compile dpdk with -fPIC [09:58:26] I do make -j80 [09:58:33] K [09:58:46] build time isn't really a problem it's a fast system [09:58:48] doing -with-fio takes care of all that [09:58:53] note that you also need to run the plugin with the version of fio you built it against (not a system-wide version of fio installed by the distro, for example) [09:58:54] i just don't want to break what I already have working :) [09:59:07] (unless they happen to be the exact same version) [09:59:28] good tips [09:59:35] i've only got fio v3.6 from teh repo [10:36:59] *** Quits: JoeGruher (86868b4b@gateway/web/freenode/ip.134.134.139.75) (Ping timeout: 260 seconds) [11:05:13] *** Joins: JoeGruher (86868b4b@gateway/web/freenode/ip.134.134.139.75) [11:06:18] if i want to do ioengine=/path/to/plugin.so, I don't actually have a .so file in that directory after doing a build with fio, would i use ioengine=/path/spdk/examples/nvme/fio_plugin/fio_plugin ? [11:08:54] right, that's the shared object, just not named with .so extension [11:10:53] yeah sorry we don't stick .so on the end of the filename [11:11:11] it works! neat [11:11:42] thx for the help [11:21:00] *** Quits: JoeGruher (86868b4b@gateway/web/freenode/ip.134.134.139.75) (Quit: Page closed) [11:22:02] *** Joins: lhodev (~lhodev@inet-hqmc01-o.oracle.com) [11:24:04] *** Quits: lhodev (~lhodev@inet-hqmc01-o.oracle.com) (Remote host closed the connection) [11:24:35] *** Joins: lhodev (~lhodev@inet-hqmc01-o.oracle.com) [11:26:35] *** Quits: lhodev (~lhodev@inet-hqmc01-o.oracle.com) (Remote host closed the connection) [11:27:06] *** Joins: lhodev (~lhodev@inet-hqmc01-o.oracle.com) [11:29:07] *** Quits: lhodev (~lhodev@inet-hqmc01-o.oracle.com) (Remote host closed the connection) [11:29:37] *** Joins: lhodev (~lhodev@inet-hqmc01-o.oracle.com) [11:31:39] *** Quits: lhodev (~lhodev@inet-hqmc01-o.oracle.com) (Remote host closed the connection) [11:32:08] *** Joins: lhodev (~lhodev@inet-hqmc01-o.oracle.com) [11:34:10] *** Quits: lhodev (~lhodev@inet-hqmc01-o.oracle.com) (Remote host closed the connection) [11:34:41] *** Joins: lhodev (~lhodev@inet-hqmc01-o.oracle.com) [11:36:42] *** Quits: lhodev (~lhodev@inet-hqmc01-o.oracle.com) (Remote host closed the connection) [11:38:52] *** Joins: lhodev (~lhodev@inet-hqmc01-o.oracle.com) [11:40:54] *** Quits: lhodev (~lhodev@inet-hqmc01-o.oracle.com) (Remote host closed the connection) [11:41:24] *** Joins: lhodev (~lhodev@inet-hqmc01-o.oracle.com) [11:43:25] *** Quits: lhodev (~lhodev@inet-hqmc01-o.oracle.com) (Remote host closed the connection) [11:43:55] *** Joins: lhodev (~lhodev@inet-hqmc01-o.oracle.com) [11:45:57] *** Quits: lhodev (~lhodev@inet-hqmc01-o.oracle.com) (Remote host closed the connection) [11:46:28] *** Joins: lhodev (~lhodev@inet-hqmc01-o.oracle.com) [11:48:31] *** Quits: lhodev (~lhodev@inet-hqmc01-o.oracle.com) (Remote host closed the connection) [11:49:02] *** Joins: lhodev (~lhodev@inet-hqmc01-o.oracle.com) [11:51:04] *** Quits: lhodev (~lhodev@inet-hqmc01-o.oracle.com) (Remote host closed the connection) [11:51:35] *** Joins: lhodev (~lhodev@inet-hqmc01-o.oracle.com) [11:56:51] *** Quits: tkulasek_ (~tkulasek@192.55.54.44) (Ping timeout: 240 seconds) [11:56:51] *** Quits: tkulasek (~tkulasek@192.55.54.44) (Ping timeout: 240 seconds) [12:54:09] looking more at this intermittent rocksdb failure [12:55:16] in all of the failure cases, blobfs is reporting an sst file size as 73819750 bytes, but the MANIFEST indicates it should be smaller (usually something right around 64MB) [12:55:34] so i converted 73819750 to hex - 0x4666666 [12:56:20] blobfs is possessed! [13:07:38] *** Quits: pohly (~pohly@p54BD5098.dip0.t-ipconnect.de) (Quit: Leaving.) [13:26:21] yikes [13:27:01] did I ever tell you about the time when ADOT sent my wife a new license plate that read "DVL-WMN"? True story, I called to get it replaced and the lady just laughed, couldn't believe they sent that out [13:55:03] *** Quits: lhodev (~lhodev@inet-hqmc01-o.oracle.com) (Remote host closed the connection) [14:26:36] *** Quits: johnmeneghini (~johnmeneg@216.240.30.5) (Quit: Leaving.) [14:29:36] *** Joins: lhodev (~lhodev@66-90-218-190.dyn.grandenetworks.net) [15:06:23] *** Joins: johnmeneghini (~johnmeneg@pool-100-0-53-181.bstnma.fios.verizon.net) [15:21:40] *** Joins: JoeGruher (86868b48@gateway/web/freenode/ip.134.134.139.72) [15:22:31] is there a meaningful difference between testing with the spdk nvme fio plugin versus the spdk bdev fio plugin (for the case where there's just a 1:1 mapping from nvme devices to bdevs)? [15:22:51] or would it essentially be two ways of doing the same thing in that case? [15:25:24] the nvme fio_plugin is written against the NVMe driver API directly, while the bdev fio_plugin is written against the bdev API (bdev_nvme layered on top of nvme driver for this case) [15:25:32] so there is some extra overhead in using the bdev fio_plugin [16:16:51] thx [16:21:44] *** Quits: JoeGruher (86868b48@gateway/web/freenode/ip.134.134.139.72) (Ping timeout: 260 seconds) [17:28:49] *** Joins: travis-ci (~travis-ci@ec2-54-196-204-91.compute-1.amazonaws.com) [17:28:50] (spdk/master) nvme: factor out qpair destruction function (Changpeng Liu) [17:28:50] Diff URL: https://github.com/spdk/spdk/compare/1fd010702693...967339f3e533 [17:28:50] *** Parts: travis-ci (~travis-ci@ec2-54-196-204-91.compute-1.amazonaws.com) () [18:07:31] Madhu has posted his version 1 spdk_thread changes. See: https://review.gerrithub.io/#/c/spdk/spdk/+/414293/ [18:07:55] Please submit this patch for CIT verification. [18:33:22] *** Joins: Shuhei (caf6fc61@gateway/web/freenode/ip.202.246.252.97) [20:39:34] *** Quits: Shuhei (caf6fc61@gateway/web/freenode/ip.202.246.252.97) (Ping timeout: 260 seconds) [21:31:59] *** Quits: johnmeneghini (~johnmeneg@pool-100-0-53-181.bstnma.fios.verizon.net) (Quit: Leaving.) [22:48:49] *** Joins: pohly (~pohly@p54BD5098.dip0.t-ipconnect.de)