[00:08:30] *** Joins: ziyeyang_ (~ziyeyang@134.134.139.83) [00:27:43] *** Joins: baruch_ (~baruch@31.210.186.250) [00:30:19] *** Quits: baruch (~baruch@31.210.182.58) (Ping timeout: 268 seconds) [00:31:22] *** Quits: ziyeyang_ (~ziyeyang@134.134.139.83) (Remote host closed the connection) [01:12:16] *** Joins: gila (~gila@5ED4D9C8.cm-7-5d.dynamic.ziggo.nl) [01:18:25] *** Quits: baruch_ (~baruch@31.210.186.250) (Read error: Connection reset by peer) [01:21:30] *** Joins: tomzawadzki (tomzawadzk@nat/intel/x-btdlvnyxoyngrarp) [01:41:59] *** Quits: gila (~gila@5ED4D9C8.cm-7-5d.dynamic.ziggo.nl) (Quit: My Mac Pro has gone to sleep. ZZZzzz…) [01:46:30] *** Joins: gila (~gila@5ED4D9C8.cm-7-5d.dynamic.ziggo.nl) [02:10:55] bwalker: why do we need exactly fio-2.21 for fio_plugin? [02:11:35] is there anything that blocks us from updating to fio-3.x? [02:12:19] there are some bus in 2.21 that have been fixed in 3.3 [02:12:31] could we consider to update? [02:12:43] bus -> bugs [02:15:12] *** Joins: baruch_ (~baruch@bzq-82-81-85-138.red.bezeqint.net) [02:31:43] *** Quits: gila (~gila@5ED4D9C8.cm-7-5d.dynamic.ziggo.nl) (Quit: My Mac Pro has gone to sleep. ZZZzzz…) [02:32:30] *** Joins: gila (~gila@5ED4D9C8.cm-7-5d.dynamic.ziggo.nl) [02:34:42] *** Quits: Shuhei (caf6fc61@gateway/web/freenode/ip.202.246.252.97) (Quit: Page closed) [02:42:59] *** Quits: gila (~gila@5ED4D9C8.cm-7-5d.dynamic.ziggo.nl) (Quit: My Mac Pro has gone to sleep. ZZZzzz…) [03:53:13] *** guerby_ is now known as guerby [03:53:24] *** Quits: guerby (~guerby@ip165.tetaneutral.net) (Changing host) [03:53:24] *** Joins: guerby (~guerby@april/board/guerby) [04:43:39] *** Joins: gila (~gila@5ED4D9C8.cm-7-5d.dynamic.ziggo.nl) [04:54:23] *** Quits: gila (~gila@5ED4D9C8.cm-7-5d.dynamic.ziggo.nl) (Quit: My Mac Pro has gone to sleep. ZZZzzz…) [05:24:18] *** Joins: gila (~gila@94.212.217.200) [06:35:56] *** Quits: baruch_ (~baruch@bzq-82-81-85-138.red.bezeqint.net) (Ping timeout: 252 seconds) [07:45:05] could somebody help me with https://ci.spdk.io/builds/review/c7fd0d9026a3323911b62861ddcb5aab2b33e0e3.1513864990/fedora-04/build.log ? [07:45:32] I added a new library and db_bench fails to compile [08:50:45] darsto: you need to add the new library to lib/rocksdb/spdk.rocksdb.mk [08:51:18] this makefile segment gets included as part of the rocksdb build (which is done from a separate repository) [08:59:00] jimharris: ahh, that's where it is. thanks! [08:59:29] np [09:08:46] darsto: +1 on upgrading fio - there should be no reason we can't [09:09:32] we will need to test it out, update the fio_plugin readme, and update the versions on the test pool machines [09:23:12] *** Quits: gila (~gila@94.212.217.200) (Quit: My Mac Pro has gone to sleep. ZZZzzz…) [09:31:32] darsto: fio changes the plugin API from version to version often. But I think it just works on fio 3.x from what I've tested [09:31:49] so basically just make sure it compiles, do some testing, etc. [09:31:57] then you can update the docs to 3.3 or whatever is the latest now [09:35:33] drv, bwalker: could one of you take a look at https://review.gerrithub.io/#/c/380459/ [09:35:42] going through klateck's merge requests [09:37:24] this takes lvol_integrity from 13 to 32 seconds on fedora-03 [09:37:36] probably ok, but we should be on the lookout for big increases like that [10:11:37] so I broke scan-build and can't figure out how to unconfuse it [10:11:45] *** Quits: tomzawadzki (tomzawadzk@nat/intel/x-btdlvnyxoyngrarp) (Ping timeout: 264 seconds) [10:11:47] https://ci.spdk.io/builds/review/bf5a233094e006114a06ef7e47b0f6bd4b4066a9.1513823893/fedora-01/scan-build/report-42be62.html#EndPath [10:12:17] it is following along with the code and assumes "A <= B", then assumes " A != B", then assumes "A >= B" [10:12:21] which can't all be true [10:12:58] it must think the variables can mutate in a function call somewhere [10:12:59] but they can't [10:14:29] *** Joins: JamesLiu (~James@c-73-70-222-173.hsd1.ca.comcast.net) [10:51:32] well I just rewrote it to do the same thing but in a less clever way and now it seems fine [10:51:50] probably for the best - it's easier to read. I wrote out all of the conditions in an if statement [11:13:07] maybe it doesn't totally grok realloc? [11:13:25] but yeah, i don't see any problems with it [11:31:16] drv: do we still need this patch? https://review.gerrithub.io/#/c/392547/ [11:31:31] i just committed your bdev_unregister on shutdown patch [11:36:24] *** Joins: lhodev (~Adium@inet-hqmc01-o.oracle.com) [11:37:24] *** Parts: lhodev (~Adium@inet-hqmc01-o.oracle.com) () [13:11:42] *** Quits: JamesLiu (~James@c-73-70-222-173.hsd1.ca.comcast.net) (Remote host closed the connection) [13:21:05] jimharris: no, that one can be dropped now [14:12:13] *** Joins: JamesLiu (~James@2601:640:8300:10f3:adff:5367:6b4:3ca5) [14:16:50] *** Quits: JamesLiu (~James@2601:640:8300:10f3:adff:5367:6b4:3ca5) (Ping timeout: 252 seconds) [14:22:21] *** Joins: gila (~gila@5ED4D9C8.cm-7-5d.dynamic.ziggo.nl) [14:58:48] i'm making some modifications to spdk_histogram_data - basically allowing it be sized dynamically, for cases where we don't need as high of precision and can live with a smaller set of buckets [14:59:18] i wrote a test app to verify the relative performance differences of the existing and new approach - any objections if i check it in? [14:59:25] sounds good to me [14:59:32] we probably won't need it indefinitely - we can always remove it later [14:59:43] the proof is in the pudding [14:59:45] or the test code :) [15:13:36] *** Joins: JamesLiu (~James@2601:640:8300:10f3:adff:5367:6b4:3ca5) [15:15:35] *** Quits: gila (~gila@5ED4D9C8.cm-7-5d.dynamic.ziggo.nl) (Quit: My Mac Pro has gone to sleep. ZZZzzz…) [15:18:11] *** Quits: JamesLiu (~James@2601:640:8300:10f3:adff:5367:6b4:3ca5) (Ping timeout: 265 seconds) [15:41:58] *** Joins: ming (4c679796@gateway/web/freenode/ip.76.103.151.150) [15:44:26] jimharris: your first patch needs a .gitignore [15:44:44] Hi, Ben, I just setup a vhost target and tried to use the bdev it exported, below is my conf file(initiator side) [15:45:04] [VirtioUser0] [15:45:11] Path /home/ming.lin/spdk/app/vhost/vhost.0 [15:45:16] Queues 1 [15:45:21] Name mybdev0 [15:46:04] Then I tried to make blobfs against "mybdev0" [15:46:29] Here is the error I got: [15:46:31] rte_virtio/virtio_user/vhost_user.c: 220:get_hugepage_file_info: *ERROR*: Exceed maximum of 8 [15:47:22] argh [15:47:28] does it mean the initiator side can only use 8 huge pages? [15:48:07] ming: yes - unfortunately it does, this is a limitation in how vhost communicates the shared memory to the target process [15:48:21] the best practice is to use 1GB huge pages - then you can at least have 8GB of memory shared [15:49:57] ok, let me try. Thanks. [16:15:09] *** Joins: JamesLiu (~James@2601:640:8300:10f3:adff:5367:6b4:3ca5) [16:20:03] *** Quits: JamesLiu (~James@2601:640:8300:10f3:adff:5367:6b4:3ca5) (Ping timeout: 265 seconds) [16:54:28] jimharris: is blow conf correct? [16:54:34] Target side: [16:54:40] [Nvme] [16:54:45] TransportID "trtype:PCIe traddr:0000:03:00.0" Nvme0 [16:54:50] [VhostScsi0] [16:54:54] Name vhost.0 [16:54:58] Dev 0 Nvme0n1 [16:55:06] Initiator side: [16:55:09] [VirtioUser0] [16:55:15] Path /home/ming.lin/spdk/app/vhost/vhost.0 [16:55:19] Queues 1 [16:55:24] Name mybdev0 [16:56:15] But make blobfs fail, it says: [16:56:21] mkfs.c: 77:spdk_mkfs_run: *ERROR*: bdev mybdev0 not found [17:13:50] *** Joins: lhodev (~Adium@66-90-218-190.dyn.grandenetworks.net) [17:16:40] *** Joins: JamesLiu (~James@2601:640:8300:10f3:adff:5367:6b4:3ca5) [17:16:49] *** Quits: lhodev (~Adium@66-90-218-190.dyn.grandenetworks.net) (Client Quit) [17:21:26] *** Quits: JamesLiu (~James@2601:640:8300:10f3:adff:5367:6b4:3ca5) (Ping timeout: 265 seconds) [17:26:23] *** Joins: JamesLiu (~James@2601:640:8300:10f3:adff:5367:6b4:3ca5) [17:34:38] *** Quits: JamesLiu (~James@2601:640:8300:10f3:adff:5367:6b4:3ca5) (Quit: Leaving...) [19:04:29] *** Quits: ming (4c679796@gateway/web/freenode/ip.76.103.151.150) (Ping timeout: 260 seconds) [19:07:26] *** Joins: David____ (6a0b2209@gateway/web/freenode/ip.106.11.34.9) [20:32:19] hello everyone, this is David Zhang from Alibaba. May I ask a question about using huge page memory. We have a core dump due to rte_hash_lookup_data, however because it is from huge page memory, we cannot use gdb to track back. The memory info are not dumped with the core. [21:44:54] *** Quits: David____ (6a0b2209@gateway/web/freenode/ip.106.11.34.9) (Ping timeout: 260 seconds) [22:19:24] *** Joins: gila (~gila@5ED4D9C8.cm-7-5d.dynamic.ziggo.nl) [22:22:42] *** Quits: gila (~gila@5ED4D9C8.cm-7-5d.dynamic.ziggo.nl) (Read error: Connection reset by peer) [22:22:55] *** Joins: gila (~gila@5ED4D9C8.cm-7-5d.dynamic.ziggo.nl) [23:48:34] *** Quits: gila (~gila@5ED4D9C8.cm-7-5d.dynamic.ziggo.nl) (Quit: My Mac Pro has gone to sleep. ZZZzzz…)