[00:00:38] bwalker: sure I'll take a look [02:02:02] *** Joins: travis-ci (~travis-ci@ec2-54-236-165-224.compute-1.amazonaws.com) [02:02:03] (spdk/master) lib/ftl: check for null (Claire J. In) [02:02:03] Diff URL: https://github.com/spdk/spdk/compare/643af88f132b...3be9febdf9ec [02:02:03] *** Parts: travis-ci (~travis-ci@ec2-54-236-165-224.compute-1.amazonaws.com) () [02:55:02] *** Quits: gila (~gila@94-212-217-121.cable.dynamic.v4.ziggo.nl) (Ping timeout: 245 seconds) [03:06:50] *** Joins: gila (~gila@94-212-217-121.cable.dynamic.v4.ziggo.nl) [05:58:24] *** Quits: gila (~gila@94-212-217-121.cable.dynamic.v4.ziggo.nl) () [06:00:32] *** Joins: gila (~gila@94-212-217-121.cable.dynamic.v4.ziggo.nl) [06:14:42] *** Joins: travis-ci (~travis-ci@ec2-18-205-20-197.compute-1.amazonaws.com) [06:14:43] (spdk/master) nvmf: put \0 at end of default serial number (Jim Harris) [06:14:43] Diff URL: https://github.com/spdk/spdk/compare/3be9febdf9ec...ca44fd69552f [06:14:43] *** Parts: travis-ci (~travis-ci@ec2-18-205-20-197.compute-1.amazonaws.com) () [06:25:08] *** Joins: felipef (~felipef@62.254.189.133) [07:10:00] Is anyone else getting 404 on GitHub? [07:10:02] eg. https://github.com/spdk/spdk/tree/refs/heads/master/lib [08:33:18] peluse!! [10:00:41] *** Quits: felipef (~felipef@62.254.189.133) (Remote host closed the connection) [10:41:47] felipef: how did you get to that URL? [10:43:04] i do get a 404 on that specific URL [10:43:30] but I can't find any problem when clicking around various links through https://github.com/spdk/spdk [10:51:19] anyone seeing problems fetching from gerrithub? [10:58:02] seems to be working now [11:17:32] *** Joins: travis-ci (~travis-ci@ec2-100-24-14-167.compute-1.amazonaws.com) [11:17:33] (spdk/master) jsonrpc.md: add cache param to construct_ftl_bdev (Darek Stojaczyk) [11:17:33] Diff URL: https://github.com/spdk/spdk/compare/ca44fd69552f...06ec27a54d50 [11:17:33] *** Parts: travis-ci (~travis-ci@ec2-100-24-14-167.compute-1.amazonaws.com) () [12:29:54] *** Joins: travis-ci (~travis-ci@ec2-3-88-179-163.compute-1.amazonaws.com) [12:29:55] (spdk/master) iscsi: fix up segment fault due to failed initialization (Changpeng Liu) [12:29:56] Diff URL: https://github.com/spdk/spdk/compare/e87e8263dfc0...790c47d479bf [12:29:56] *** Parts: travis-ci (~travis-ci@ec2-3-88-179-163.compute-1.amazonaws.com) () [12:32:15] *** Joins: travis-ci (~travis-ci@ec2-100-24-14-167.compute-1.amazonaws.com) [12:32:16] (spdk/master) bdev/compress: add unit test mocks and empty functions (paul luse) [12:32:16] Diff URL: https://github.com/spdk/spdk/compare/790c47d479bf...b95d25975dde [12:32:16] *** Parts: travis-ci (~travis-ci@ec2-100-24-14-167.compute-1.amazonaws.com) () [12:58:47] darsto!! :) [14:03:42] bwalker, sethhowe: can you take a look at my comments on https://review.gerrithub.io/c/spdk/spdk/+/449253 [14:03:52] not sure if i'm way off base here - you guys know this much better than i do [14:10:42] hmm [14:11:33] the accounting on those lists could stay there, but in the error path we are going to need code that "completes" all unposted wrs [14:22:08] The reason I am updating the count when I queue the work requests is that if we allow ourselves to queue more WRs than we are allowed to post at once, then we would need to traverse the list and split it when we get to the last element we can legally post. [14:23:12] We don't relly have a good way of recovering from a failed ibv_post_send call right now. Most of the error handling for that case is preemptive. We do our best to ensure we are never overshooting the requirements the send queue puts on us. [14:26:36] bwalker: I see what you mean now. I do need to jump back and free the requests in the error case, otherwise they get lost forever. [14:26:37] if i'm reading bwalker's response correctly, when a qpair gets into an error state, it needs to clean up this pending list and remove itself from the list of qpairs? [14:28:48] when it enters the error state, it waits to receive a completion for each of the wrs [14:28:51] and decrements these counts [14:28:56] before it destroys itself [14:29:04] but if the wrs are sitting in this list, they won't ever get a completion [14:29:17] so we just need to loop through them and reduce the counts/release the resources when we enter the error state [14:29:21] that was what i was worried about wrt the accounting [14:29:48] but what you're proposing makes more sense [14:30:02] than moving the accounting like i was asking about [14:30:37] yeah I think seth is right that we need to do the accounting up front so that we go down a queueing path if we have too many [14:35:42] Thanks Jim, that was a good catch. I'll have the patches updated in a second. Luckily the send and recv lists are per-qp so the accounting is pretty easy. [16:29:05] *** Joins: travis-ci (~travis-ci@ec2-18-207-150-239.compute-1.amazonaws.com) [16:29:06] (spdk/master) thread: Implement a thread scheduler (Ben Walker) [16:29:07] Diff URL: https://github.com/spdk/spdk/compare/1679104edb14...032920f25031 [16:29:07] *** Parts: travis-ci (~travis-ci@ec2-18-207-150-239.compute-1.amazonaws.com) () [16:30:38] *** Joins: travis-ci (~travis-ci@ec2-23-22-252-77.compute-1.amazonaws.com) [16:30:39] (spdk/master) bdev/compress: misc comment and log message cleanup (paul luse) [16:30:39] Diff URL: https://github.com/spdk/spdk/compare/032920f25031...1e6cdc190d4d [16:30:39] *** Parts: travis-ci (~travis-ci@ec2-23-22-252-77.compute-1.amazonaws.com) () [16:32:45] *** Joins: travis-ci (~travis-ci@ec2-54-197-164-221.compute-1.amazonaws.com) [16:32:46] (spdk/master) rdma: optimize and move buffers variable. (Seth Howell) [16:32:47] Diff URL: https://github.com/spdk/spdk/compare/1e6cdc190d4d...f4adbc79cef0 [16:32:47] *** Parts: travis-ci (~travis-ci@ec2-54-197-164-221.compute-1.amazonaws.com) () [16:43:19] *** Joins: travis-ci (~travis-ci@ec2-18-234-210-143.compute-1.amazonaws.com) [16:43:20] (spdk/master) ut/iscsi/portal_grp: fix license/copyright header (Jim Harris) [16:43:21] Diff URL: https://github.com/spdk/spdk/compare/f4adbc79cef0...fee323b2abb8 [16:43:21] *** Parts: travis-ci (~travis-ci@ec2-18-234-210-143.compute-1.amazonaws.com) () [16:45:19] *** Joins: Shuhei (caf6fc61@gateway/web/freenode/ip.202.246.252.97) [16:46:05] I was so lazy, I didn't know paste/nopaste mode of VIM. [16:48:12] I will not have you fix such a stupid bug again, thank you. [17:00:41] @Shuhei @jimharris the next spdk bug scrub meeting is tomorrow right? [17:04:15] I read your update to github. [17:05:40] *** Joins: travis-ci (~travis-ci@ec2-18-207-150-239.compute-1.amazonaws.com) [17:05:41] (spdk/master) ocf: finish OCF reqs after put_io_channel (Vitaliy Mysak) [17:05:42] Diff URL: https://github.com/spdk/spdk/compare/fee323b2abb8...4fad4b86dd6e [17:05:42] *** Parts: travis-ci (~travis-ci@ec2-18-207-150-239.compute-1.amazonaws.com) () [17:11:54] yea we're putting more people on the issue on our side to try to figure out how to repro this. I'm wondering it would make sense for me to join the meeting and be available to answer any questions you guys have when you're discussing the issue tomorrow [17:12:39] we're thinking that the issue might be with logout from the target [17:12:57] because the test that injects errors doesn't ever kick the drive from the kernel [17:15:02] it just deletes it from our io stack which includes logging out the intiiator [17:16:03] and i t hink when we kick a drive from the kernel when the management daemon is running it detects that we kicked the drive and tells our io stack to logout from the target [17:16:12] well, it does the same delete process [18:12:34] *** Quits: ppelplin (~ppelplin@134.134.139.75) (Read error: Connection reset by peer) [18:12:34] *** Quits: mszwed (mszwed@nat/intel/x-hbpscckgmhsyzizh) (Read error: Connection reset by peer) [18:12:34] *** Quits: pawelkax (pawelkax@nat/intel/x-bcvowlkgeixzsfxg) (Quit: ZNC 1.7.0 - https://znc.in) [18:13:56] *** Joins: ppelplin (~ppelplin@134.134.139.75) [20:17:25] *** Quits: Shuhei (caf6fc61@gateway/web/freenode/ip.202.246.252.97) (Ping timeout: 256 seconds) [22:21:02] Project autotest-nightly-failing build #317: STILL FAILING in 21 min. See https://ci.spdk.io/spdk-jenkins for results. [22:36:44] Project autotest-nightly build #449: FAILURE in 36 min. See https://ci.spdk.io/spdk-jenkins for results. [23:09:08] *** Joins: felipef (~felipef@cpc92310-cmbg19-2-0-cust421.5-4.cable.virginm.net) [23:16:01] *** Quits: felipef (~felipef@cpc92310-cmbg19-2-0-cust421.5-4.cable.virginm.net) () [23:16:16] *** Joins: mszwed (~mszwed@134.134.139.75)