[04:58:51] klateck, can you look into why the execution of pkgdep.sh in https://review.gerrithub.io/c/spdk/spdk/+/429523 failed on freebsd and ioat machines in the ITP TP? It wan OK on all the CH TP. [04:59:19] Note that the patch above does a bunch of other stuff for compress but the latest push was just a test to execute pkgdep in prep for running the complete patch. [04:59:26] I'll be offline for a few hours... [05:01:04] peluse, we do have euro community meeting today, right? (probably tomorrow for you) [05:01:21] There is no trello card for today's meeting [05:01:48] I'll create one if you don't mind [05:57:10] *** Joins: vmysak (vmysak@nat/intel/x-qdftyyjrupsssczn) [06:52:20] peluse I took a look into these fails. So for Linux based systems it seems it'd be best to add --allowerasing yum option for the failing line. Looks like a dependency issue with openssl [06:52:41] For freebsd - not 100% sure but looks like a proxy issue. [06:54:23] BTW - I discourage installing dependencies that way. Let me (or sethhowe, or pniedzwx) to install the deps on systems. Reason is - your patch most possibly did not run on all of the systems in the pool. [06:55:16] Let me/us know*, typo [06:57:30] klateck, yup, sethhowe and I talked about it and did it this way for the CH systems and figured we just hit the TP systems manually after talking to you [06:57:40] fore the reason you mention [06:57:58] klateck: do you have an easy way to run "some script" on all of the systems in the Jenkins TP? [06:58:16] however, the proxy and --allowerasing issues seem to need fixing in both the freebsd tp machine in in the pgkdep.sh don't you think? [06:58:33] yum isn't a freebsd thing [06:58:44] Well, the most important reason is that a patch set will get run only on a limited set of machines, and new deps will be installed only on them:) [06:59:00] jimharris yeah, but only as my local utility, nothing official [06:59:05] just a really simple script [06:59:38] might be good to set up a Jenkins maintenance job for installing/unintalling deps from all systems [06:59:53] klateck: i think that's fine - it means if we have the ability to broadly install packages, etc. on all of the systems easily for that day when we have 100 systems :) [07:00:39] who wouldn't want to manually install stuff on 100 systems? ;) [07:02:50] yeah, totally understand the reason why using a patch to install dep isn't the correct way to do it on ITP machines and that yum isn't a freebsd thing. That still leaves a few things (a) the freebsd ITP machine seems to have a proxy issue (b) the pgkdep.sh needs the yum switch (c) eventually we need a decent way to manage TP systems but it would be good if you can manually fix (a,b) and manually run an updated pkgdep soon so the compressdev dependencies [07:02:50] patch can land. Make sense? [07:36:13] Yup. Do you want me to install these deps today on all systems? [07:37:15] are there any implictations to applying the -allowerasing flag? [07:38:13] I don't think so, yum cannot solve some issues with already installed version of openssl so we have to remove it and install new. [07:38:20] Bumped into this a couple of times, never caused issues [07:46:28] klateck, yes please [07:48:22] and then we need to talk about a system mgmt plan for the TP systems, not just for updating dependencies but overall taking care of them. bwalker mentioned use of popular stuff like puppet and chef, not saying either of those makes more sense than using Jenkins but the idea is to look at this as a bigger thing that also encompasses updating dependencies on the TP systems [08:55:47] *** Quits: vmysak (vmysak@nat/intel/x-qdftyyjrupsssczn) (Ping timeout: 240 seconds) [09:43:09] *** Joins: travis-ci (~travis-ci@ec2-54-158-227-104.compute-1.amazonaws.com) [09:43:10] (spdk/master) iscsi: Propagate SCSI error status to split SCSI read/write commands (Shuhei Matsumoto) [09:43:10] Diff URL: https://github.com/spdk/spdk/compare/7d9ed0cf4239...37af0cc930b0 [09:43:10] *** Parts: travis-ci (~travis-ci@ec2-54-158-227-104.compute-1.amazonaws.com) () [09:59:47] jimharris: I'm in lab now. I think reactors will go to sleep only if the user has set that "maximum additional latency" parameter in the future, which is only useful if they expect to be really idle [10:15:14] klateck, peluse: Something like this: https://blog.trifork.com/2013/04/02/ansible-example-playbook-to-setup-jenkins-slave/ [10:44:36] *** Joins: travis-ci (~travis-ci@ec2-54-162-142-30.compute-1.amazonaws.com) [10:44:37] (spdk/master) nvme: Eliminate identify errors to Discovery ctrlr (Lance Hartmann) [10:44:37] Diff URL: https://github.com/spdk/spdk/compare/37af0cc930b0...e865a52415d0 [10:44:37] *** Parts: travis-ci (~travis-ci@ec2-54-162-142-30.compute-1.amazonaws.com) () [10:50:41] *** Joins: vmysak (vmysak@nat/intel/x-gtpybysjjqnfptwp) [11:12:46] *** Joins: travis-ci (~travis-ci@ec2-54-157-37-139.compute-1.amazonaws.com) [11:12:47] (spdk/master) build: remove duplicate spdk.app.mk includes (Jim Harris) [11:12:47] Diff URL: https://github.com/spdk/spdk/compare/e865a52415d0...9c2aea2ad581 [11:12:47] *** Parts: travis-ci (~travis-ci@ec2-54-157-37-139.compute-1.amazonaws.com) () [11:28:37] *** Quits: vmysak (vmysak@nat/intel/x-gtpybysjjqnfptwp) (Ping timeout: 246 seconds) [11:49:22] klateck, still around? [12:20:56] *** Joins: vmysak (vmysak@nat/intel/x-jdllnbyvqaetvtmd) [13:06:47] *** Quits: vmysak (vmysak@nat/intel/x-jdllnbyvqaetvtmd) (Ping timeout: 240 seconds) [13:48:59] In 18.10, I ran some traffic via nvme/perf to nvmf_tgt (a single malloc bdev) without complaint. I've pulled master and rebuilt, and now I see on the target: rdma.c: 530:spdk_nvmf_rdma_set_ibv_state: *NOTICE*: IBV QP#0 changed to: IBV_QPS_ERR [13:49:25] that's not an error - just printed when you disconnect [13:49:37] or when the discovery controller disconnects [13:50:57] I noticed it both initially and at end, so that's consistent with the two cases you described. The appearance of '_ERR' always grabs my attention. [13:51:14] yeah I need to change that message [13:51:43] Feel free to add me as a review on that change ;-) [17:29:20] *** Joins: Shuhei (caf6fc61@gateway/web/freenode/ip.202.246.252.97) [20:03:41] *** Quits: Shuhei (caf6fc61@gateway/web/freenode/ip.202.246.252.97) (Ping timeout: 256 seconds) [22:19:26] *** Quits: bwalker (~bwalker@134.134.139.72) (ZNC - http://znc.in) [22:19:50] *** Joins: bwalker (bwalker@nat/intel/x-rysvgjayomcmstuv) [22:19:50] *** ChanServ sets mode: +o bwalker [22:19:50] *** Server sets mode: +cnrt [22:19:51] *** Server sets mode: +cnrt