[15:56:01] test [15:57:21] turned on my ZNC log [15:57:28] now to figure out how to put it somewhere searchable [15:58:23] test [15:58:37] test [16:00:20] *** Quits: krk_ (~krk@50.224.189.114) (Ping timeout: 260 seconds) [16:00:49] https://ci.spdk.io/irclog/ [16:01:02] it's not automatic and it's not searchable [16:01:07] can fix automatic [16:01:11] not sure how to fix searchable [16:01:19] well, I can probably figure that out [16:06:45] *** Joins: krk_ (~krk@50.224.189.114) [16:17:12] nice! [16:17:21] ctrl-f or cmd-f works :) [16:20:01] test [16:24:14] blah blah this should get logged automatically [16:26:05] *** Quits: krk_ (~krk@50.224.189.114) (Ping timeout: 240 seconds) [16:26:31] cool, it worked [16:26:35] the irc log syncs every 2 minutes [16:27:00] *** bwalker changes topic to 'Storage Performance Development Kit - http://www.spdk.io/. This channel is logged at https://ci.spdk.io/irclog/' [16:27:11] drv: what was that linux/if.h v. net/if.h fix? [16:27:24] i'd search bwalker's log but it doesn't go back that far [16:27:47] testing out the vagrant set up with the new dpdk submodule and ran into this [16:27:49] jimharris: it is not needed now since we removed the DPDK file from rte_vhost that was including it [16:28:04] but we could work around it if you need that file again [16:28:05] oh - I need to rebase my patch then [16:28:08] no [16:28:38] we committed those patches after i started my submodule patch yesterday [16:29:39] *** Joins: krk_ (~krk@50.224.189.114) [16:31:46] *** Joins: ziyeyang_ (~ziyeyang@134.134.139.78) [16:44:13] *** Quits: jstern__ (~jstern@134.134.139.83) (Ping timeout: 260 seconds) [16:47:28] *** Quits: krk_ (~krk@50.224.189.114) (Ping timeout: 260 seconds) [16:50:01] *** Joins: krk_ (~krk@50.224.189.114) [16:56:41] drv: build systems need the proxy to get out to fetch dpdk from github [16:58:27] hmm, yeah, I'm not sure if we want to actually allow that [16:58:58] but I guess for now (until we expose the build pool to the public) we can set the variables in the test user's bash_profile or wherever [16:59:41] does that work with git? or do you need to do git config? [16:59:50] * jimharris goes to check [17:00:47] I think https_proxy env var should be enough [17:01:53] eventually, it would be good if just the pool machine pulled the source and then provided it to the agents without them needing public internet access [17:02:23] *** Quits: krk_ (~krk@50.224.189.114) (Ping timeout: 245 seconds) [17:03:00] how do the agents get the spdk source today? don't they git pull? [17:03:37] yeah, currently they all 'git fetch' for every build directly from Gerrit [17:03:42] *** Joins: krk_ (~krk@50.224.189.114) [17:03:52] but ideally, the pool machine should do that once, and then drop the source in the network share [17:04:10] ok so once that gets fixed for the spdk source, this dpdk submodule would get fixed too [17:04:13] yeah [17:04:24] the pool script would need to have submodule handling, but that shouldn't be hard [17:04:31] (paging sethhowe) [17:12:42] Ok, I will put this on the list of updates. I am getting through the last few changes before we can start running public reviews right now and will address this after. [17:13:18] *** Joins: tsuyoshi (b42b2067@gateway/web/freenode/ip.180.43.32.103) [17:17:28] *** Quits: krk_ (~krk@50.224.189.114) (Ping timeout: 260 seconds) [17:21:06] *** Joins: krk_ (~krk@50.224.189.114) [17:23:17] drv: bah - -fPIC with the fio_plugin [17:24:08] i guess i could add -fPIC dynamically in the dpdkbuild makefile if CONFIG_FIO_PLUGIN=y? [17:24:25] or just change our spdk-17.02 branch to always build dpdk with -fPIC [17:31:49] *** Quits: tsuyoshi (b42b2067@gateway/web/freenode/ip.180.43.32.103) (Ping timeout: 260 seconds) [17:32:55] jimharris: I think we should just always build DPDK with -FPIC [17:33:00] *-fPIC [17:33:24] set extra cflags, or CONFIG_RTE_BUILD_SHARED_LIB=y? [17:33:53] I would prefer to stick with static linking [17:50:00] did you do anything special when building the dpdk libraries on the freebsd box? that system failed too somewhere in bdevio [17:52:53] I don't think so [17:53:08] fedora-04 is hosed [17:53:25] i tried the new pdu but that didn't seem to help [17:53:35] wait, maybe we had some freebsd-specific patch, does that sound familiar? [17:54:01] I think Ben fixed it upstream but it's probably not in DPDK 17.02 [17:54:04] hmmm - yeah, zeroing out the buffer? [17:54:11] no, something to do with PCI device removal [17:54:30] i'll fish it out and add that to our new branch :) [17:54:47] one more item for you to ponder [17:54:48] commit 24a535796824049fab8570d0b29e658470375876 on DPDK master [17:54:51] autopackage.sh [17:55:15] we could just skip autopackage on systems that don't have a DPDK installed outside of the tree [17:55:33] yeah, alternatively we could figure out if git archive can add submodules somehow [17:55:40] but I'm OK either way [17:56:08] is autopackage still valid? [17:56:25] i mean it used to be when we had our internal waikiki stuff [17:56:32] well, we should definitely still build with CONFIG_DEBUG=n [17:56:39] but that doesn't necessarily have to be in autopackage [17:56:42] sure [17:56:54] also testing whether it builds out of a tarball is probably also still a good idea [17:57:02] since that is how github releases end up [17:57:19] fedora-04 seems ok now, not sure if you did something or it just took a while to reboot [17:57:37] i'm guessing the power cycle I did just took a while [17:58:02] the status page still shows fedora-04 as offline [17:58:20] ah, the pool script fell over [17:58:21] and my failed submodule test as incomplete [17:58:23] genhtml: ERROR: cannot read /home/sys_sgsw/pool-spdk/pool/repo/dpdk/build/include/rte_memcpy.h [17:58:40] since the coverage tries to check out the source code too, that probably needs some changes in pool.py to do submodule magic [17:59:41] (also the pool shouldn't fail if the coverage step fails) [17:59:52] ok - well a few more things to iron out tomorrow [18:00:16] freebsd patch and autopackage submodules [18:00:49] i also want a "make install-deps" target to remove the list of packages in our README [18:00:57] that's next after this submodule stuff is in [18:07:34] drv: can you point me to where the pool.py changes need to go? [18:08:22] jimharris: subprocess.check_call([genhtml], shell=True) probably needs a try/catch around it [18:08:37] also probably the lcov step before that [18:08:41] is that the final solution? [18:09:01] for now :) [18:09:06] it should fail the build if that step fails [18:09:14] but the pool.py script shouldn't exit [18:10:01] ok - i'll take a look at that tomorrow too [18:10:31] I'm going to verify -1 the submodule patch for now so it doesn't keep running [18:17:04] *** Quits: ziyeyang_ (~ziyeyang@134.134.139.78) (Remote host closed the connection) [18:49:12] *** Quits: krk_ (~krk@50.224.189.114) (Remote host closed the connection) [19:17:10] *** Joins: ziyeyang_ (~ziyeyang@192.55.54.45) [19:33:32] *** Quits: ziyeyang_ (~ziyeyang@192.55.54.45) (Remote host closed the connection) [19:34:15] *** Joins: ziyeyang_ (~ziyeyang@134.134.139.76) [20:12:38] *** Joins: changpeng (~changpeng@192.55.54.39) [20:52:09] *** Quits: changpeng (~changpeng@192.55.54.39) (Ping timeout: 272 seconds) [21:00:20] *** Joins: changpeng (~changpeng@134.134.139.78) [21:17:29] *** Quits: changpeng (~changpeng@134.134.139.78) (Remote host closed the connection) [21:34:16] *** Joins: krk_ (~krk@2601:644:8502:4820:a165:9576:ca8:1829) [21:59:48] *** Quits: ziyeyang_ (~ziyeyang@134.134.139.76) (Ping timeout: 260 seconds) [22:19:45] *** Quits: krk_ (~krk@2601:644:8502:4820:a165:9576:ca8:1829) (Remote host closed the connection)