[07:20:09] Hi guys, could you look at spdk event notifications card? https://trello.com/c/ZTIHxp3w/28-spdk-event-notifications [07:21:16] I think that asynchronous stream API makes most sense. Client would just connect to spdk rpc socket, but would not disconnect from it as long as he wants to receive the events. [07:21:47] Maybe you have other opinions? [07:22:50] I'm just writing a patch that would provide such RPC functionality, then I can figure out how to send actual events [07:28:10] *** Joins: pi-r (uid326872@gateway/web/irccloud.com/x-nltutputbxsnwvtc) [07:28:43] *** pi-r is now known as ppelplin-mobile [07:33:49] *** ppelplin-mobile is now known as ppelplin2 [08:58:16] *** Joins: alekseymmm (050811aa@gateway/web/freenode/ip.5.8.17.170) [09:06:51] sethhowe: can we get CI test systems moved to nvme-cli 1.6 branch? [09:07:12] "we" meaning "you" of course :-) [09:08:52] jimharris: Yep, of course! we'll do that right now ;) [09:09:04] lol [09:09:23] thanks! let me know once it's done - we have a couple of patches to merge after it's complete [09:11:02] *** Joins: travis-ci (~travis-ci@ec2-54-80-78-188.compute-1.amazonaws.com) [09:11:03] (spdk/master) jsonrpc: simplify parsing client response (Pawel Wodkowski) [09:11:03] Diff URL: https://github.com/spdk/spdk/compare/83b1102f7b0f...961a286a2742 [09:11:03] *** Parts: travis-ci (~travis-ci@ec2-54-80-78-188.compute-1.amazonaws.com) () [09:22:06] *** Joins: travis-ci (~travis-ci@ec2-107-20-88-221.compute-1.amazonaws.com) [09:22:07] (spdk/master) thread: add the missing newline character (Liang Yan) [09:22:08] Diff URL: https://github.com/spdk/spdk/compare/961a286a2742...c9118cae5742 [09:22:08] *** Parts: travis-ci (~travis-ci@ec2-107-20-88-221.compute-1.amazonaws.com) () [09:31:33] jimharris: We should be all good. I just swapped the chandler machines, and it looks like the poland machines have already been updated. [09:33:23] thanks! i just rebased the patch that removes the CONFIG.local symlink so we'll get confirmation that everything is ready [09:55:56] PSA: Pausing the chandler test pool momentarily to swap in some hardware. appx 10 min delay after the current build finishes. [10:15:30] PSA: Build pool back online. [10:36:37] jimharris: I was looking into the whole "we should turn on tracing and capture the shared memory files for nvmf" thing. Looks like Karol did it last month and it works great! https://review.gerrithub.io/c/spdk/spdk/+/425078 [12:43:41] jimharris: For the lack of knowledge regarding protocol of using "[RFC]" in gerrit. Searched around for info on that, but found little. Manually added you and pwodkowx so hopefully it appears on your respective dashboards [13:07:55] lhodev: [RFC] usage is tribal knowledge - we should update the development.md page probably for it - the key point is that the build pools will skip any patch with that in the commit message [13:13:34] comments posted [13:14:20] jimharris: Ben just was explaining the same when you msg'd ;-). Thank you [13:15:03] oh - i didn't see that in my IRC log [13:20:54] oh phone [13:20:57] on phone [13:22:31] jimharris: bwalker was explaining via voice call -- not IRC. [13:22:50] ah - got it [13:23:41] jimharris: Can anyone update the development.md or is that just maintainers? That's part of another repo -- not spdk repo, right? [13:24:01] it's in spdk.github.io, and actually anyone can send a patch for the website using the same process as the rest of SPDK [13:24:17] github.com/spdk/spdk.github.io [13:24:40] I'm happy to do the update to it then. [13:24:49] excellent [13:25:03] cool - the website is the standard GitHub pages stack. [13:25:19] which means it is statically generated from templates and markdown files using a tool called Jekyll [13:25:52] the development page is development.md [14:43:10] darsto: you there? [14:43:40] jimharris: here [14:43:50] dpdk 18.08 [14:44:05] what's next? I think all of the patches are merged [14:45:09] is nvme-cli updated on TP as well? [14:45:12] yes [14:45:22] then we're good to go [14:45:44] what about 423053? [14:45:46] legacy mem works straight away [14:46:01] do we need to merge 423053 with the patch that moves the dpdk submodule? [14:47:36] yes, exactly [14:48:00] since there's no common_spdk in dpdk 18.08, the build would just fail without that patch [14:48:21] cool - do you want to do that now? [14:49:33] you mean i should push a patch with dpdk submodule update? [14:49:45] sure, i'll do that now [14:50:30] i think you need to add the dpdk submodule change to 423053 [14:50:35] it can't be two separate patches [14:51:38] bwalker, sethhowe: why do we need to start the nvmf target with waiting for RPCs to create the RDMA transport with special options? [14:52:08] we had to do that before, because set_nvmf_target_options could only run before subsystems were initialized [14:52:13] I think that's left over from when the options were on the transport [14:52:16] weren't [14:52:23] the restriction can likely be relaxed now [14:53:06] ok - i'll -1 this doc patch then; we should also probably clean up all of the tests that do this behavior too (I didn't really realize this when those patches went in) [14:54:20] the transport is implicitly created when you add the first listener? [14:54:35] it is, but only for backward compatibility [14:54:49] the preferred way going forward is to explicitly create the transport via rpc [14:55:02] but I don't think that needs to happen before the subsystems are initialized - it can happen at any time [14:55:57] at some point (19.01 probably) we'll require explicit creation of transports and fail rpcs that previously would have implicitly created them [15:01:01] bwalker: are you ok with +2'ing that iscsi initiator doc patch and then i'll make that fix in a follow up patch? [15:01:09] then we can clear out the 3 patches after that one [15:01:16] sure [15:01:21] After cloning github.com/spdk/spdk.github.io, I assume like the SPDK src repo there's some git-hook I need to download via curl/wget, right? Can I use the same one, or is there one specific/unique for the SPDK site repo? [15:01:37] it's the same one [15:05:01] https://review.gerrithub.io/#/c/spdk/spdk/+/429974/ [15:05:20] jimharris: bwalker: ack [15:07:02] *** Joins: travis-ci (~travis-ci@ec2-54-211-242-61.compute-1.amazonaws.com) [15:07:03] (spdk/master) bdev/virtio: deprecate remove_virtio_scsi_bdev RPC (Darek Stojaczyk) [15:07:03] Diff URL: https://github.com/spdk/spdk/compare/2975822ffe6d...b237c1b5c34e [15:07:03] *** Parts: travis-ci (~travis-ci@ec2-54-211-242-61.compute-1.amazonaws.com) () [15:35:50] A clone of http://github.com/spdk/spdk.github.io != a clone from https://review.gerrithub.io/spdk/spdk.github.io [15:37:26] lhodev: By that do you mean the tip of master is in a different place? [15:39:54] Correct [15:41:17] lhodev: Yeah, we have had some problems with the sync being broken between gerrithub and github. We have fixed several of the repos, but it looks like we haven't gotten to that one yet. I will send gerrithub a service request. [15:41:48] The latter (review.gerrithub.io) appears much more recent. As such, I guess one should always clone from review.gerrithub.io -- not github.com [15:43:17] they are the same for me [15:45:29] On github, my tip of master is d6311bd748 (commit date Oct 4). From gerrithub, my tip of master is 8312532e (commit date Oct 9). [15:46:12] I see my issue, ok they are 3 commits different [15:46:31] and now they are the same [15:48:30] bwalker: did you "effect" that, or did a sync just coincidentally occur ? [15:49:01] I did it [15:49:12] sethhowe is sending a support request to fix it - it should have been automatic [15:49:31] but I'm the administrator so I do what I want [15:51:02] lhodev: bwalker: I sent the request. They usually handle it within 2 business days. [16:05:24] *** Joins: darsto_ (~darsto@89-78-174-111.dynamic.chello.pl) [16:06:19] *** Quits: darsto (~darsto@89-78-174-111.dynamic.chello.pl) (Ping timeout: 246 seconds) [16:06:19] *** darsto_ is now known as darsto [17:03:36] turns out there are 2 additional patches required for our bdev/crypto module to work with dpdk 18.08 [17:04:17] peluse: assigned you to https://review.gerrithub.io/c/spdk/spdk/+/429988/1 . Could you take a look? [17:06:16] crypto is that one thing I didn't test before, uh [23:31:54] *** Quits: drv (daniel@oak.drv.nu) (Ping timeout: 250 seconds) [23:39:00] *** Joins: tomzawadzki (tomzawadzk@nat/intel/x-llifxrulkjvswqar) [23:40:31] *** Quits: tomzawadzki (tomzawadzk@nat/intel/x-llifxrulkjvswqar) (Client Quit) [23:48:02] *** Joins: tomzawadzki (uid327004@gateway/web/irccloud.com/x-seqpmhczysqgmjwt) [23:53:21] *** Joins: drv (daniel@oak.drv.nu) [23:53:21] *** ChanServ sets mode: +o drv