[00:30:26] *** Quits: felipefr (~felipef@219.93.16.226) (Remote host closed the connection) [01:58:28] *** Quits: nKumar (uid239884@gateway/web/irccloud.com/x-cbmoetcnkicgsaqb) (Ping timeout: 240 seconds) [02:07:01] *** Quits: Shuhei (caf6fc61@gateway/web/freenode/ip.202.246.252.97) (Quit: Page closed) [04:03:31] *** Joins: nKumar (uid239884@gateway/web/irccloud.com/x-szmizwohfgjvstgr) [08:16:21] jimharris yeah, that is definitely doable. [09:01:40] *** Joins: lhodev (~Adium@inet-hqmc07-o.oracle.com) [09:02:04] *** Parts: lhodev (~Adium@inet-hqmc07-o.oracle.com) () [09:34:44] jimharris, drv: during live migration development I hit issue that I can't use bdev for more that one SCSI LUN because bdev name is also a LUN name. [09:35:44] I know that there are bdev aliases incoming [09:36:47] can I abuse this to create two luns backed by one bdev? [09:37:17] or there will be some terrible disaster when I do that? :D [09:38:10] ppeplinski: sorry, the most recent failure of your blob patch was my fault. I'll rerun it now. [11:22:37] *** Joins: Samuel (49e726bd@gateway/web/freenode/ip.73.231.38.189) [11:23:04] I have a question on how bdev handles IO timeout, can anyone help? [11:24:36] https://github.com/spdk/spdk/blob/master/lib/bdev/nvme/bdev_nvme.c#L795 [11:26:04] After timeout_cb is called and abort command is sent to admin queue to abort the timed-out IO, who is going to remove that IO from it's qpair? I don't see anywhere handling this. [11:47:52] Samuel, there's nothing to remove, the command was submitted so if the controller found it and aborts it you'll get a completion. If it doesn't then it's kinda up to the controller what to do, this is from the 1.3 spec "It is implementation [11:47:53] specific when a controller chooses to complete the Abort command when the command to abort is not [11:47:53] found." [11:49:15] [11:50:28] I see, so the IO callback will get called, after controller found/abort it and send out a completion to that qpair. [11:51:06] that's what I believe should happen, yes [11:51:50] and, of course, the abort command itself should complete otherwise things are totally hosed and it will indicate if the cmd was actually found/aborted or not [11:57:08] yeah, and the status code for the aborted command on the completion code is in the SC field as: "07h Command Abort Requested: The command was aborted due to a Command Abort command [11:57:08] being received that specified the Submission Queue Identifier and Command Identifier of this [11:57:08] command." [12:13:52] Got it, thanks! [12:14:11] callback is https://github.com/spdk/spdk/blob/master/lib/bdev/bdev.c#L1739, and it checks SC field. [12:15:18] Thank you very much, Paul! [12:18:36] you bet! [12:47:49] *** Joins: darsto_ (~dstojacx@89-68-135-211.dynamic.chello.pl) [13:31:21] *** Quits: pniedzwx (~pniedzwx@192.55.54.40) (*.net *.split) [13:47:56] jimharris: the patch you requested to the build pool has been merged. you can now push a patch with [RFC] in the title and it won't run on the build pool. If you amend the commit and remove [RFC] it will then run the tests on the review. [13:47:58] *** Joins: pniedzwx (~pniedzwx@192.55.54.40) [13:48:35] excellent [14:04:48] so, meaning Request For Comment but don't burn time in test? [14:07:08] or... this one is Really F^&*king Cool so no need to test, just merge? [14:07:54] or... Run For Cover, this one might blow the whole system? [14:10:38] the first one, but maybe we need some tags for those other ones too :) [14:11:42] rock-n-roll :) [14:14:14] *** Quits: Samuel (49e726bd@gateway/web/freenode/ip.73.231.38.189) (Quit: Page closed) [14:38:58] *** Quits: jkkariu (~jkkariu@192.55.54.40) (*.net *.split) [14:39:56] nice [14:40:22] could we send out an email about this? [14:51:13] good idea - I nominate sethhowe :) [14:51:23] new gerrit has built-in functionality for running automated tests [14:51:27] https://chromium-review.googlesource.com/c/chromium/src/+/809706 [14:52:24] the tryjobs thing - seems cool as well [14:52:57] hmm, interesting - I wonder what GerritHub's update policy is [14:57:31] no idea. I can't find an article now, but I've read that chromium will drop the old UI by the end of this year [14:58:08] this would mean that the new UI would be feature-complete [14:59:21] drv: will do. [15:01:01] found it. first post, last sentence https://groups.google.com/a/chromium.org/forum/#!topic/chromium-dev/jqEllOsqeoU [15:01:58] *** Joins: jkkariu (~jkkariu@192.55.54.40) [15:37:09] pausing the build pool to update fedora-04 and fedora-08 [16:12:09] *** Joins: felipefr (~felipef@219.93.16.226) [16:47:02] *** Quits: felipefr (~felipef@219.93.16.226) (Remote host closed the connection) [17:20:04] RFC for proposed Blobstore API changes: https://review.gerrithub.io/#/c/390698/ [17:20:21] using sethhowe's nifty RFC hook so I don't waste time on the build pool! [19:03:55] RFC... makes us sound so "official" :) [19:04:48] "Mr. Harris, I approve of your current RFC with some minor revisions to the verbiage outlined in section 1.1 paragraph A, sentence 3"