| *** annegentle has joined #openstack-swift | 00:01 | |
| *** annegentle has quit IRC | 00:06 | |
| *** yuan has quit IRC | 00:07 | |
| clayg | torgomatic: so this ring builder debug - it makes my test output sorta scrolly | 00:09 |
|---|---|---|
| *** tsg has quit IRC | 00:17 | |
| *** david-lyle has quit IRC | 00:24 | |
| *** david-lyle has joined #openstack-swift | 00:27 | |
| *** tsg has joined #openstack-swift | 00:29 | |
| torgomatic | clayg: when a test fails, you mean? | 00:31 |
| clayg | torgomatic: yeah :D | 00:31 |
| torgomatic | clayg: yeah, I noticed that, but I didn't end up doing anything about it :p | 00:32 |
| torgomatic | I'm not sure why everything's at debug-level, but I'm guessing it's the testrunner doing something to the root logger | 00:33 |
| clayg | torgomatic: yeah - we could maybe try to squelch something in the test setup? | 00:33 |
| torgomatic | clayg: yeah, that could work... lemme see | 00:34 |
| *** kota_ has quit IRC | 00:36 | |
| *** lcurtis has joined #openstack-swift | 00:47 | |
| *** annegentle has joined #openstack-swift | 01:03 | |
| *** annegentle has quit IRC | 01:08 | |
| mattoliverau | FYI, my boss is actually in my city today (rare cause he lives and works in a different city) so will be in and out most the day in meetings etc. | 01:10 |
| *** gyee has quit IRC | 01:12 | |
| *** zhill has quit IRC | 01:45 | |
| *** lcurtis has quit IRC | 01:46 | |
| *** bill_az has quit IRC | 01:54 | |
| *** tsg has quit IRC | 01:59 | |
| *** annegentle has joined #openstack-swift | 02:00 | |
| openstackgerrit | Samuel Merritt proposed openstack/swift: Bulk upload: treat user xattrs as object metadata https://review.openstack.org/176161 | 02:05 |
| *** imkarrer_ has joined #openstack-swift | 02:07 | |
| *** annegentle has quit IRC | 02:09 | |
| *** annegentle has joined #openstack-swift | 02:27 | |
| *** esker has joined #openstack-swift | 02:38 | |
| *** jamielennox is now known as jamielennox|away | 02:39 | |
| *** Vinsh has joined #openstack-swift | 02:57 | |
| *** Vinsh has quit IRC | 03:01 | |
| *** annegentle has quit IRC | 03:10 | |
| *** annegentle has joined #openstack-swift | 03:13 | |
| *** annegentle has quit IRC | 03:20 | |
| *** annegentle has joined #openstack-swift | 03:21 | |
| *** annegentle has quit IRC | 03:26 | |
| *** MVenesio has quit IRC | 03:51 | |
| *** km_ has joined #openstack-swift | 03:57 | |
| *** km has quit IRC | 03:59 | |
| *** kei_yama has quit IRC | 04:01 | |
| *** kei_yama has joined #openstack-swift | 04:02 | |
| *** imkarrer__ has joined #openstack-swift | 04:04 | |
| *** imkarrer_ has quit IRC | 04:06 | |
| *** jrichli has joined #openstack-swift | 04:10 | |
| *** evanjfraser has quit IRC | 04:11 | |
| *** jrichli has quit IRC | 04:17 | |
| *** evanjfraser has joined #openstack-swift | 04:21 | |
| *** annegentle has joined #openstack-swift | 04:22 | |
| *** ppai has joined #openstack-swift | 04:27 | |
| *** annegentle has quit IRC | 04:28 | |
| *** ppai has quit IRC | 04:55 | |
| *** km_ has quit IRC | 05:22 | |
| *** kei_yama has quit IRC | 05:23 | |
| *** annegentle has joined #openstack-swift | 05:24 | |
| *** SkyRocknRoll has joined #openstack-swift | 05:24 | |
| *** SkyRocknRoll has joined #openstack-swift | 05:24 | |
| *** kei_yama has joined #openstack-swift | 05:25 | |
| *** annegentle has quit IRC | 05:29 | |
| *** yuan has joined #openstack-swift | 05:33 | |
| *** bkopilov has quit IRC | 05:35 | |
| *** bkopilov has joined #openstack-swift | 05:37 | |
| *** km has joined #openstack-swift | 05:38 | |
| *** silor has joined #openstack-swift | 05:44 | |
| *** haigang has joined #openstack-swift | 05:44 | |
| *** tsg has joined #openstack-swift | 05:45 | |
| *** zaitcev has quit IRC | 05:53 | |
| *** tsg has quit IRC | 05:58 | |
| *** tsg has joined #openstack-swift | 05:59 | |
| *** tsg has quit IRC | 06:14 | |
| *** annegentle has joined #openstack-swift | 06:25 | |
| *** annegentle has quit IRC | 06:29 | |
| *** proteusguy has joined #openstack-swift | 06:54 | |
| *** kota_ has joined #openstack-swift | 06:59 | |
| kota_ | wow, all rc2paches got -1 verified at granade gate...why... | 07:09 |
| *** imkarrer__ has quit IRC | 07:16 | |
| *** annegentle has joined #openstack-swift | 07:25 | |
| *** annegentle has quit IRC | 07:31 | |
| *** acoles_away is now known as acoles | 07:37 | |
| *** slo_ has quit IRC | 07:43 | |
| *** jistr has joined #openstack-swift | 07:50 | |
| *** proteusguy has quit IRC | 07:55 | |
| acoles | kota_: hmmm, i just put recheck on them all, looks like bug 1446847 had a fix committed so maybe they will pass | 07:56 |
| openstack | bug 1446847 in OpenStack-Gate "stable/juno gate is busted on Oslo releases on 4/21" [Undecided,New] https://launchpad.net/bugs/1446847 | 07:56 |
| *** jordanP has joined #openstack-swift | 08:09 | |
| kota_ | acoles: Thanks to make the recheck :) I didn't know the command "reverify". | 08:16 |
| kota_ | acoles: FYI, I'm joining mid-cycle ops meetup report in Japan :P | 08:17 |
| kota_ | leaving until irc meeting time for dinner and sleep | 08:18 |
| *** kota_ has quit IRC | 08:18 | |
| *** sandywalsh has quit IRC | 08:24 | |
| *** sandywalsh has joined #openstack-swift | 08:24 | |
| *** annegentle has joined #openstack-swift | 08:26 | |
| *** breitz has quit IRC | 08:27 | |
| *** breitz has joined #openstack-swift | 08:29 | |
| *** ndk has quit IRC | 08:30 | |
| *** ndk has joined #openstack-swift | 08:31 | |
| *** annegentle has quit IRC | 08:31 | |
| *** haigang has quit IRC | 08:32 | |
| *** ppai has joined #openstack-swift | 08:35 | |
| *** haigang has joined #openstack-swift | 08:41 | |
| *** proteusguy has joined #openstack-swift | 08:43 | |
| *** haigang has quit IRC | 08:46 | |
| *** haigang has joined #openstack-swift | 08:47 | |
| *** haigang has quit IRC | 08:52 | |
| *** haigang has joined #openstack-swift | 08:58 | |
| *** kei_yama has quit IRC | 09:00 | |
| *** km has quit IRC | 09:02 | |
| *** ppai has quit IRC | 09:19 | |
| *** annegentle has joined #openstack-swift | 09:27 | |
| *** annegentle has quit IRC | 09:31 | |
| *** ppai has joined #openstack-swift | 09:33 | |
| *** silor has quit IRC | 09:53 | |
| *** aix has joined #openstack-swift | 10:00 | |
| *** slo_ has joined #openstack-swift | 10:04 | |
| slo_ | At the Swift install point - having multi storage server (let's say 5 of them). I am interested is it OK to call swift-ring-builder CREATE command (for setting replica count at acc/cont/obj) just at one storage node and than copy all al *.ring.gz files to all storage nodes within /etc/swift directory or do I have to call the same create command also on other storage nodes? | 10:08 |
| acoles | slo_: yes, its ok to create ring files on one node and then copy to all others, see http://docs.openstack.org/juno/install-guide/install/apt/content/swift-initial-rings.html | 10:12 |
| slo_ | acoles: thx for confirmation | 10:18 |
| *** annegentle has joined #openstack-swift | 10:28 | |
| *** annegentle has quit IRC | 10:33 | |
| *** proteusguy has quit IRC | 10:34 | |
| *** geaaru has joined #openstack-swift | 10:42 | |
| *** haigang has quit IRC | 10:58 | |
| *** erlon has joined #openstack-swift | 10:59 | |
| *** aix has quit IRC | 11:07 | |
| *** haigang has joined #openstack-swift | 11:08 | |
| *** proteusguy has joined #openstack-swift | 11:09 | |
| *** annegentle has joined #openstack-swift | 11:28 | |
| *** annegentle has quit IRC | 11:34 | |
| *** esker has quit IRC | 11:38 | |
| *** aix has joined #openstack-swift | 11:58 | |
| *** haigang has quit IRC | 12:02 | |
| *** haigang has joined #openstack-swift | 12:03 | |
| *** ppai has quit IRC | 12:06 | |
| *** haigang has quit IRC | 12:12 | |
| *** tsg has joined #openstack-swift | 12:18 | |
| *** ppai has joined #openstack-swift | 12:18 | |
| *** annegentle has joined #openstack-swift | 12:29 | |
| slo_ | acoles: are you there | 12:38 |
| *** SkyRocknRoll has quit IRC | 12:40 | |
| slo_ | i am interested about *.builder files, should I also copy them to other servers, just as i did this *.gz? | 12:40 |
| slo_ | i guess verifiying the ring content is possible just on node, where builder files also exist at the same time, for example i am calling swift-ring-builder object.builder command (http://docs.openstack.org/juno/install-guide/install/yum/content/swift-initial-rings-object.html) to list the content on the node to which i just copied *.ring.gz files and I get no results. It works just on the node, where *.builder files are prese | 12:44 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/129154 | 12:47 |
| acoles | slo_: you don't need the *.builder files on all the nodes, but you do need them on the node(s) where you run swift-ring-builder. so, yes, you guessed right. | 12:52 |
| slo_ | acoles: ok thx | 12:57 |
| acoles | slo_: welcome | 12:57 |
| *** ppai has quit IRC | 13:01 | |
| *** mahatic has joined #openstack-swift | 13:18 | |
| *** annegentle has quit IRC | 13:20 | |
| *** dencaval has joined #openstack-swift | 13:20 | |
| jordanP | test_tempurl_keys_hidden_from_acl_readonly fail on a fresh devstack+swift install | 13:20 |
| jordanP | ResponseError: 403: 'Forbidden' | 13:20 |
| *** pberis has joined #openstack-swift | 13:21 | |
| *** Gu_______ has joined #openstack-swift | 13:21 | |
| *** esker has joined #openstack-swift | 13:24 | |
| *** annegentle has joined #openstack-swift | 13:29 | |
| *** Vinsh has joined #openstack-swift | 13:30 | |
| jordanP | probably a devstack bug though | 13:31 |
| *** annegentle has quit IRC | 13:32 | |
| *** annegentle has joined #openstack-swift | 13:32 | |
| *** Vinsh has quit IRC | 13:39 | |
| *** annegentle has quit IRC | 13:57 | |
| *** annegentle has joined #openstack-swift | 13:59 | |
| *** Gu_______ has quit IRC | 14:02 | |
| *** jrichli has joined #openstack-swift | 14:08 | |
| *** lcurtis has joined #openstack-swift | 14:09 | |
| *** petertr7 has joined #openstack-swift | 14:12 | |
| *** haomaiwang has quit IRC | 14:12 | |
| *** haomaiwang has joined #openstack-swift | 14:13 | |
| *** wbhuber has joined #openstack-swift | 14:18 | |
| *** Vinsh has joined #openstack-swift | 14:23 | |
| *** imkarrer_ has joined #openstack-swift | 14:30 | |
| *** pberis has quit IRC | 14:31 | |
| *** pberis has joined #openstack-swift | 14:39 | |
| *** Vinsh_ has joined #openstack-swift | 14:42 | |
| *** tsg has quit IRC | 14:45 | |
| *** Vinsh has quit IRC | 14:45 | |
| openstackgerrit | Janie Richling proposed openstack/swift: FakeFooters middleware https://review.openstack.org/165517 | 14:49 |
| openstackgerrit | Janie Richling proposed openstack/swift: Allow middleware to override metadata header checking https://review.openstack.org/163806 | 14:49 |
| openstackgerrit | Janie Richling proposed openstack/swift: WIP - working on the encryption feature. https://review.openstack.org/157907 | 14:49 |
| openstackgerrit | Janie Richling proposed openstack/swift: Enable middleware to set metadata on object POST https://review.openstack.org/158401 | 14:49 |
| *** bsdkurt has joined #openstack-swift | 14:49 | |
| *** pberis has quit IRC | 14:52 | |
| *** Guest_ has joined #openstack-swift | 14:53 | |
| *** bill_az has joined #openstack-swift | 14:55 | |
| *** Guest_ has quit IRC | 14:56 | |
| notmyname | clayg: on https://review.openstack.org/#/q/I34747094a645201e52da7081d9215d5b2cb5d9ff,n,z is that not tested with probe tests? | 15:05 |
| bill_az | tdasilva: ping | 15:05 |
| notmyname | bill_az: FYI tdasilva probably won't be around for a while. he's on paternity leave | 15:06 |
| openstackgerrit | Merged openstack/swift: Don't apply the wrong Etag validation to rebuilt fragments https://review.openstack.org/174552 | 15:06 |
| *** annegentle has quit IRC | 15:06 | |
| *** annegentle has joined #openstack-swift | 15:06 | |
| bill_az | notmyname: ok - the right priority! | 15:06 |
| notmyname | bill_az: anything I can help with? | 15:07 |
| bill_az | notmyname: I have a question about change https://github.com/openstack/swift/commit/fa89064933fefa33702520b40734c11f08b2c569 | 15:07 |
| bill_az | this adds a check that device directory exists - | 15:08 |
| bill_az | where as previously if device directory did not exist it would be created | 15:08 |
| bill_az | For our deployment on clustered filesystem backend, we did not create the "virtual devices", and would like object server create these on first access | 15:09 |
| bill_az | Is this the intended behavior? | 15:09 |
| bill_az | notmyname: line 646-654 in the commit reference | 15:11 |
| notmyname | bill_az: in get_hashes? | 15:13 |
| *** annegentle has quit IRC | 15:13 | |
| notmyname | I'm just looking for references to mkdirs() in that github diff | 15:14 |
| bill_az | get_dev_path | 15:14 |
| bill_az | notmyname: the mount_check logic adds check_dir | 15:14 |
| bill_az | notmyname: we have in the past set mount_check=False, and directories would be created; | 15:15 |
| bill_az | but now the check_dir will fail because device directory is not created initially | 15:16 |
| bill_az | if the first test (line 647) were based on sef.mount_test rather than mount_test that is passed to the function, we would be ok, | 15:19 |
| swifterdarrell | bill_az: if the device dirs are mount points (like most swift deployments), it can be pretty bad to let them get auto-vivified and start jamming data into them when the data disk is accidentally not mounted; I'd bet preventing that was the reason for the change | 15:21 |
| swifterdarrell | bill_az: I think the right thing to do is to change to ensure those dirs exist in your deployment; you must have referenced them in the ring, right? So they must be known | 15:21 |
| bill_az | notmyname: yep - we can do that, I wanted to first understand the intent of the change. Got it now - thanks! | 15:22 |
| notmyname | swifterdarrell: thanks | 15:22 |
| swifterdarrell | bill_az: no prob :) | 15:23 |
| *** imkarrer_ has quit IRC | 15:24 | |
| bill_az | swifterdarrell: I think the issue still exists for account & container data, though it's a lot smaller problem there | 15:25 |
| *** minwoob has joined #openstack-swift | 15:32 | |
| *** annegentle has joined #openstack-swift | 15:33 | |
| *** silor has joined #openstack-swift | 15:34 | |
| *** zaitcev has joined #openstack-swift | 15:37 | |
| *** ChanServ sets mode: +v zaitcev | 15:37 | |
| notmyname | clayg: FWIW, with that patch you put a -2 on, probe tests pass. hmm...but I don't have different replication IPs, so it might be passing by accident | 15:38 |
| *** annegentle has quit IRC | 15:39 | |
| *** annegentle has joined #openstack-swift | 15:40 | |
| *** gyee has joined #openstack-swift | 15:44 | |
| *** zhill has joined #openstack-swift | 15:46 | |
| acoles | notmyname: pardon me eavesdropping..., i think clayg is right (he always is!) GET is not decorated to be a replication method so an object server configured to be a replication server should not allow to GETs | 15:48 |
| notmyname | acoles: right. so therefore that actually will fail in real clusters but work in my saio where it's all kindof fakey | 15:49 |
| notmyname | also, it means that (1) it shoudl stay as the main IP for now and (2) later we should actually change it so that the background traffic stays on the "replication" network | 15:50 |
| acoles | notmyname: yes, i'm not sure where, or if at all, probe tests use separate replication servers | 15:50 |
| notmyname | do you agree? | 15:50 |
| acoles | yes, i think i agree | 15:51 |
| notmyname | we should probably update the saio to use more than one network and some separate replication-only storage servers | 15:52 |
| notmyname | ok, but the point is that for today, that patch should NOT be in the RC2 | 15:52 |
| acoles | agree | 15:53 |
| *** patchbot has quit IRC | 15:55 | |
| *** patchbot has joined #openstack-swift | 15:57 | |
| clayg | acoles: if I was *always* right I would have seen that peluse had done it right the first time | 15:59 |
| acoles | heh | 16:00 |
| clayg | swifterdarrell was all like "and they all use replication_ip - *right*" - and I was like "of course!" /me goes to check | 16:00 |
| clayg | notmyname: I was going to try to add some second replication servers to my saio today on a different port - tought it would be easier than setting up a seperate network | 16:01 |
| notmyname | cool | 16:01 |
| swifterdarrell | heh | 16:01 |
| clayg | notmyname: i believe that would have caused probetests to fail | 16:01 |
| notmyname | clayg: you're just giving me less and less reasons to not use vSAIO | 16:01 |
| notmyname | anyone know about http://www.dotscale.io? looks interesting | 16:03 |
| *** proteusguy has quit IRC | 16:03 | |
| * notmyname wonders if I could convince people to let me go and then do field trips to visit acoles and cschwede | 16:03 | |
| notmyname | clayg: acoles: also, I guess we can't call it a "replication network" any more | 16:07 |
| acoles | notmyname: we may need a whole session on 'naming' in vancouver ;) | 16:08 |
| notmyname | ok, looks like our RC2 patches are all in the gate and moving forward. we should have the tag soon | 16:08 |
| notmyname | acoles: definitely a fishbowl session. more people to contribute naming ideas will make it easier ;-) | 16:09 |
| cschwede | notmyname: field trip? sounds awesome! :) | 16:09 |
| * cschwede orders sunny weather for June in Hamburg | 16:09 | |
| acoles | notmyname: well if you go to paris i think *i* might need to make the trip to meet you | 16:10 |
| notmyname | cschwede: or we could all meet in paris for that conference | 16:10 |
| acoles | ^^ idea there | 16:10 |
| *** jistr has quit IRC | 16:12 | |
| notmyname | if I understand the gate issues from yesterday, the capping of requirements on stable/* versions caused the problem. a dependency pulled in it's own dependencies (with different versions) and that's what conflicted. | 16:13 |
| notmyname | eg foo-server pulls A<=v1 and B, but B also pulls in B>v1 | 16:14 |
| notmyname | something like that | 16:14 |
| *** Guest___ has joined #openstack-swift | 16:15 | |
| notmyname | and recently all of the stable releases have had their versions capped. so if we release kilo with Foo.dependency >= v1.0.1, it will be capped to also have <v2.0.0 | 16:16 |
| notmyname | (which I find slightly ironic that openstack, which doesn't use semver for releases, is trusting that all the dependencies do follow semver properly) | 16:16 |
| *** Guest___ has quit IRC | 16:23 | |
| *** Guest___ has joined #openstack-swift | 16:23 | |
| *** haomaiwang has quit IRC | 16:28 | |
| *** Guest___ has quit IRC | 16:28 | |
| *** haomaiwang has joined #openstack-swift | 16:28 | |
| *** thumpba has joined #openstack-swift | 17:01 | |
| *** thumpba_ has joined #openstack-swift | 17:02 | |
| *** Vinsh_ has quit IRC | 17:04 | |
| *** thumpba has quit IRC | 17:05 | |
| *** peluse has quit IRC | 17:07 | |
| *** annegentle has quit IRC | 17:13 | |
| *** mahatic has quit IRC | 17:15 | |
| *** jordanP has quit IRC | 17:15 | |
| hurricanerix | morning | 17:20 |
| clayg | hurricanerix: morning! | 17:22 |
| clayg | hump day! | 17:22 |
| clayg | we might even have a swift meeting later | 17:23 |
| clayg | ... wonder what we'll talk about | 17:23 |
| hurricanerix | They are opening an in-n-out across the street from us tomorrow. I expect to be wfh the rest of the week... =) | 17:23 |
| clayg | hurricanerix: oh goodness - people go nuts for the in-and-out's | 17:25 |
| hurricanerix | clayg: yeah, i mean, it's ok, but i think when they opened the one in Austin, they had like a 45 min wait at the drive through and it backed up traffic on the highway. | 17:26 |
| hurricanerix | or maybe that was Dallas or something. It was somewhere in Texas. | 17:26 |
| clayg | notmyname: we go to the one down 19th from us sometimes (on the way to target) - always busy | 17:26 |
| clayg | hurricanerix: yeah that sounds about right | 17:27 |
| hurricanerix | clayg: I think the only thing I don't like is the lack of bacon. | 17:27 |
| clayg | you know i never noticed that - but I now have a big issue with in-and-out | 17:27 |
| clayg | it's basically not 'merican to offer me the option of adding bacon to my double cheeseburger | 17:28 |
| hurricanerix | guess it's byob | 17:28 |
| clayg | lol | 17:28 |
| clayg | notmyname: torgomatic: acoles: so the one part of that ill fated replication_ip change I liked was the fix in FakeRing | 17:28 |
| hurricanerix | V2!tnD%l6cEswaj%v0wecJiO*XaV@b1cT2Hk3uLF | 17:29 |
| hurricanerix | whoops that was the wrong window | 17:29 |
| clayg | notmyname: torgomatic: acoles: I'm also not 100% sure that at somepoint we may want to move the fragment rebuilding to the replication network | 17:29 |
| clayg | not sure if that means we decorate GET @replication - or turn off the repliation-server option or ... something else? | 17:30 |
| clayg | er.... consistency network? | 17:30 |
| clayg | ick | 17:30 |
| acoles | fixer-upper-network ;) | 17:31 |
| clayg | even-more-not-for-customers-network | 17:31 |
| acoles | clayg: so gerrit didn't announce this here https://review.openstack.org/#/c/176403/ WIP Revert to another handoff when primary is unavailable | 17:32 |
| acoles | clayg: i ripped off your probe test and started playing with ssync changes | 17:32 |
| acoles | clayg: just first baby steps though | 17:32 |
| -openstackstatus- NOTICE: gerrit is restarting to clear hung stream-events tasks. any review events between 16:48 and 17:32 utc will need to be rechecked or have their approval votes reapplied to trigger testing in zuul | 17:33 | |
| clayg | ok, I think I finally only now understand the "on the next pass" part | 17:34 |
| clayg | acoles: i'm sorta worried that on the 409 the sender still wants to send the object and the reciever may not be prepared to drain the object body? | 17:34 |
| *** zhill_ has joined #openstack-swift | 17:35 | |
| acoles | yeah, more tests would be good and thats a good example ^^ | 17:36 |
| *** gyee is now known as chinese_gyee | 17:37 | |
| *** chinese_gyee is now known as gyee | 17:38 | |
| acoles | clayg: i also managed to rebase my ssync+POST patch which *may* be a basis for the rx sending back hints during missing_checks, but I need to add some support to ECDiskFileManager still, so thats WIP | 17:38 |
| clayg | acoles: was your update to patch 175076 just a rebase? oh i see patch 176403 was dependent | 17:38 |
| patchbot | clayg: https://review.openstack.org/#/c/175076/ | 17:38 |
| *** mahatic has joined #openstack-swift | 17:38 | |
| acoles | yes, my +2 survived so only rebase | 17:39 |
| clayg | acoles: awesome - you know crunch time is over - you can stop showing us up now if you want to | 17:39 |
| clayg | acoles: did you see what torgomatic is working on? deepcopy rings and extracing object metadata from tarball uploads? | 17:40 |
| clayg | these are things you piddle with post EC | 17:40 |
| clayg | :) | 17:40 |
| acoles | clayg: lol. saw them fly by but not looked close yet. | 17:41 |
| clayg | heh | 17:41 |
| acoles | away for dinner, back for meeting | 17:41 |
| acoles | clayg: you see , i *am* taking it easier ^^ | 17:41 |
| clayg | acoles: even notmyname is slipping - we've got "merged" changes on the priority reviews dashboard | 17:41 |
| clayg | heh | 17:41 |
| clayg | good | 17:42 |
| *** david-lyle has quit IRC | 17:42 | |
| openstackgerrit | Merged openstack/swift: Bump PyECLib version from 1.0.3 to 1.0.7 https://review.openstack.org/175525 | 17:50 |
| notmyname | clayg: I'm just trying to give you an excuse to not put in so many late hours :-) | 17:51 |
| notmyname | (also, they merged while I was on my bike) ;-) | 17:51 |
| clayg | notmyname: uhhhhhh huhhhhhh | 17:52 |
| notmyname | looks like we're about ~1hour from having RC2, if the zuul status is to be believed | 17:55 |
| *** annegentle has joined #openstack-swift | 17:58 | |
| *** annegentle has quit IRC | 18:01 | |
| *** geaaru has quit IRC | 18:01 | |
| *** annegentle has joined #openstack-swift | 18:02 | |
| *** aix has quit IRC | 18:21 | |
| *** Vinsh has joined #openstack-swift | 18:22 | |
| *** Guest_ has joined #openstack-swift | 18:32 | |
| *** zhill_ has quit IRC | 18:40 | |
| torgomatic | clayg: exactly. working on little self-contained features is relaxing as heck compared to EC :) | 18:41 |
| clayg | torgomatic: :D | 18:43 |
| notmyname | clayg: FYI, I think keystone changed the default token expiry time to 1 hour | 18:44 |
| openstackgerrit | Merged openstack/swift: Use reconstruct insetad of decode/encode https://review.openstack.org/175741 | 18:45 |
| openstackgerrit | Merged openstack/swift: Release Import of Translations from Transifex https://review.openstack.org/175324 | 18:45 |
| notmyname | cool. I think we have the RC2 patches landed now | 18:46 |
| clayg | whoot! | 18:47 |
| notmyname | clayg: SHA f8dee761bd36f857aa1288c27e095907032fad68 look good for RC2? | 18:48 |
| notmyname | if so, then I'll pass it on to ttx | 18:48 |
| notmyname | torgomatic: ^ | 18:48 |
| notmyname | anyone ^ | 18:48 |
| torgomatic | notmyname: confirmed; that looks like a SHA | 18:48 |
| torgomatic | ;) | 18:48 |
| notmyname | torgomatic: you're learning too much from oslo.uuid.is_uud() methods | 18:49 |
| clayg | bill_az: swifterdarrell: I have an outstanding change that fixes the mount check to include a dir check for account and container - somewhere - I think | 18:49 |
| torgomatic | notmyname: also, http://www.bzarg.com/p/how-to-pronounce-hexadecimal/ | 18:49 |
| clayg | bill_az: swifterdarrell: I do think we *justified* the change because of how terrible things behave with mount_check = false - but *honestly* the reason was to support probetests making disks 507 when mount_check = False on swift-all-in-ones | 18:50 |
| *** cutforth has joined #openstack-swift | 18:50 | |
| clayg | bill_az: swifterdarrell: internally swift tests picked up some churn in the change but I think everyone assumed real systems always ruan mount_check = True and the False option was only for development environments | 18:51 |
| clayg | bill_az: swifterdarrell: so... good to know! | 18:51 |
| *** ho has joined #openstack-swift | 18:52 | |
| ho | good morning! | 18:53 |
| notmyname | torgomatic: wonderful | 18:54 |
| notmyname | meetign in #openstack-meeting in 3 minutes | 18:57 |
| *** kota_ has joined #openstack-swift | 18:58 | |
| kota_ | morning | 18:58 |
| clayg | https://wiki.openstack.org/wiki/Meetings/Swift | 18:59 |
| clayg | swift-bench!? | 18:59 |
| mattoliverau | morning | 18:59 |
| clayg | this is going to be crazy - i better plug into a power source and get moar coffee | 18:59 |
| notmyname | clayg: yeah, rememeber that one? | 18:59 |
| clayg | notmyname: of course, it comes installed with a default bench.conf on a vagrant-swift-all-in-one | 19:00 |
| notmyname | lol | 19:00 |
| notmyname | meeting starting | 19:01 |
| cschwede | wow, good participation in the meeting! | 19:01 |
| *** Guest_ has quit IRC | 19:03 | |
| *** ryshah has joined #openstack-swift | 19:03 | |
| *** Guest_ has joined #openstack-swift | 19:06 | |
| *** ryshah has quit IRC | 19:06 | |
| *** ryshah has joined #openstack-swift | 19:07 | |
| *** dmorita has joined #openstack-swift | 19:08 | |
| *** annegentle has quit IRC | 19:14 | |
| *** Guest_ has quit IRC | 19:17 | |
| *** annegentle has joined #openstack-swift | 19:18 | |
| clayg | i'm not really sure I want to make seperate replication servers the default in vagrant-swift-all-in-one - but making it optional adds it's own complexity | 19:18 |
| *** dencaval has quit IRC | 19:21 | |
| acoles | cschwede: hi! did you publish your durability calculations or calculator tool anywhere? | 19:21 |
| *** dencaval has joined #openstack-swift | 19:22 | |
| cschwede | acoles: Hi Alistair! Indeed: http://enovance.github.io/swift-durability-calculator/ | 19:22 |
| acoles | cschwede: wow! thank you | 19:23 |
| *** Guest_ has joined #openstack-swift | 19:24 | |
| cschwede | acoles: you’re welcome! let me know if you have any questions | 19:24 |
| *** sandywalsh has quit IRC | 19:30 | |
| *** sandywalsh has joined #openstack-swift | 19:31 | |
| *** ryshah has quit IRC | 19:35 | |
| *** Guest_ has quit IRC | 19:35 | |
| *** Guest_ has joined #openstack-swift | 19:39 | |
| *** shakamunyi has joined #openstack-swift | 19:40 | |
| *** Fin1te has joined #openstack-swift | 19:42 | |
| *** proteusguy has joined #openstack-swift | 19:50 | |
| *** Guest_ has quit IRC | 19:51 | |
| *** annegentle has quit IRC | 19:52 | |
| *** dmorita has quit IRC | 19:54 | |
| *** openstackgerrit has quit IRC | 19:54 | |
| *** openstackgerrit has joined #openstack-swift | 19:55 | |
| *** cutforth has quit IRC | 19:55 | |
| mattoliverau | cools, thatnks all, I'm going back to bed for a while. | 19:55 |
| clayg | that was fun! | 19:55 |
| clayg | mattoliverau: g'night | 19:55 |
| kota_ | me, too, good night. | 19:56 |
| acoles | mattoliverau: me too | 19:56 |
| clayg | kota_: mattoliverau: acoles: thanks - sleep tight! | 19:56 |
| *** kota_ has quit IRC | 19:56 | |
| *** mahatic has quit IRC | 19:57 | |
| cschwede | also leaving - have a good night / day everyone! bye | 19:57 |
| *** acoles is now known as acoles_away | 19:57 | |
| jrichli | cschwede: you too! | 20:02 |
| *** dencaval has quit IRC | 20:05 | |
| *** Fin1te has quit IRC | 20:07 | |
| *** annegentle has joined #openstack-swift | 20:14 | |
| *** silor has quit IRC | 20:16 | |
| *** annegentle has quit IRC | 20:28 | |
| *** annegentle has joined #openstack-swift | 20:29 | |
| *** sandywalsh has quit IRC | 20:33 | |
| *** sandywalsh has joined #openstack-swift | 20:34 | |
| *** Gu_______ has joined #openstack-swift | 20:34 | |
| *** Gu_______ has quit IRC | 20:55 | |
| *** Fin1te has joined #openstack-swift | 21:06 | |
| clayg | man seperate replication servers is getting sorta hairy - i mean it'd probably be pretty managable if I didn't make it optional - but you need a fair bit more RAM in the box to turn them on by default and there's a config *explosion* | 21:08 |
| clayg | ok, it may be some sort of sin against the chef gods - but I'm going to go with a dumber recipe smarter template approach | 21:14 |
| swifterdarrell | clayg: there are only chef devils | 21:16 |
| zaitcev | Like Rachael Ray | 21:18 |
| *** gyee has quit IRC | 21:20 | |
| clayg | zaitcev: i didn't know who that was - but after a quick search I've decided that's funny | 21:24 |
| zaitcev | clayg: Only now I realized that it was not a typo for "chief gods" and you meant a deployment tool called "Chef". | 21:25 |
| zaitcev | We're a Puppet show here. | 21:26 |
| *** gyee has joined #openstack-swift | 21:26 | |
| *** openstackgerrit has quit IRC | 21:29 | |
| *** openstackgerrit has joined #openstack-swift | 21:30 | |
| *** Fin1te has quit IRC | 21:31 | |
| *** ho has quit IRC | 21:32 | |
| openstackgerrit | Samuel Merritt proposed openstack/swift: Make RingBuilders deep-copy-able https://review.openstack.org/175567 | 21:35 |
| *** wbhuber has quit IRC | 21:41 | |
| *** annegentle has quit IRC | 21:45 | |
| bill_az | swifterdarrell: clayg: thanks for explanations | 21:53 |
| swifterdarrell | bill_az: yw | 21:54 |
| swifterdarrell | clayg: torgomatic: what are the relative merits of type(self) vs. self.__class__ ? | 21:56 |
| swifterdarrell | clayg: torgomatic: (thinking of the builder __deepcopy__, of course) | 21:57 |
| *** jrichli has quit IRC | 22:06 | |
| *** esker has quit IRC | 22:07 | |
| mattoliverau | Morning | 22:09 |
| notmyname | hello again | 22:09 |
| *** openstackgerrit has quit IRC | 22:11 | |
| *** openstackgerrit has joined #openstack-swift | 22:11 | |
| clayg | swifterdarrell: self.__class__ is how you do it if you're *old-skewl* | 22:11 |
| clayg | swifterdarrell: that is old style classes won't give you the class back when you type them (I think they're? like type - type?) | 22:12 |
| clayg | swifterdarrell: someone dinged me on it the other day - and somewhere on stack overflow it agreed the canonical way to get the class of an instances is type(thing) - which does read fairly nicely | 22:13 |
| clayg | so I'm trying it out! | 22:13 |
| swifterdarrell | clayg: wacky | 22:16 |
| clayg | swifterdarrell: ok well - i mis-spoke, type of instance of old-style class is just 'instance' - which I guess is some primative | 22:18 |
| swifterdarrell | clayg: so does that mean self.__class__ works in more cases than type(self)? | 22:19 |
| clayg | swifterdarrell: yeah | 22:19 |
| clayg | like I said - *old-skewl* | 22:19 |
| swifterdarrell | clayg: k, we should probably use that, then | 22:19 |
| clayg | swifterdarrell: meh, old style classes are balls | 22:20 |
| clayg | you can't even have them in python3 | 22:20 |
| swifterdarrell | clayg: does self.__class__ still work in python3? | 22:20 |
| *** erlon has quit IRC | 22:21 | |
| clayg | swifterdarrell: yeah it appears so | 22:21 |
| swifterdarrell | clayg: (I don't even care *that* much, unlike the location of deep copying, which I think is more important) | 22:22 |
| *** annegentle has joined #openstack-swift | 22:22 | |
| swifterdarrell | torgomatic: ^^^^^^^^^^^^ | 22:22 |
| clayg | swifterdarrell: i don't know the relevant stackoverflows with commentary from the likes of Alex Martelli seem to prefer the use of type with new style classes - althrough perhaps mostly for stylistic reasoning | 22:27 |
| clayg | swifterdarrell: i'm sure it's not as important as the deepcopy thing - but I honestly don't understand what that memo bit is all about :\ | 22:27 |
| swifterdarrell | clayg: it's a mechanism to detect and handle cycles | 22:28 |
| clayg | swifterdarrell: this much i gathered from the docs :P | 22:28 |
| swifterdarrell | clayg: oh | 22:28 |
| swifterdarrell | clayg: I mean, maybe we know we don't have cycles, but we should probably still code the __deepcopy__ like it wants and use deepcopy(sub_thing, memo) inside of it? | 22:29 |
| clayg | swifterdarrell: calling deepcopy(to_dict, memo) seems pretty sane | 22:30 |
| clayg | but I'd honestly have to setup some sort of scenario with a custom class that needs to implement deepcopy and has the cycle scenario so that I could a) implement it in correctly and undestand the failure then b) implement it correctly and verify the failure mode is averted | 22:31 |
| *** fbo has quit IRC | 22:31 | |
| clayg | seems like a bunch of effort just to strip off the logging attribute? | 22:31 |
| swifterdarrell | clayg: sure... I also wasn't sure, but I think you'd ideally want to have an instance of the new thing, put its id(...) into memo, *then* call your deepcopy(self.xyz, memo) | 22:32 |
| swifterdarrell | clayg: but maybe only relevant if a sub-thign could have a reference to the thing being copied (a big cycle) | 22:32 |
| * swifterdarrell shrugs | 22:32 | |
| clayg | my guess is whatever our ring builder code is doing would be happier with deepcopy(builder.to_dict()) than with deepcopy(builder) and this whacked out __deepcopy__ thing? | 22:32 |
| clayg | swifterdarrell: do we even *do* deepcopy in to_dict? | 22:33 |
| swifterdarrell | clayg: agreed; I was planning on doing that anyway if this patch doesn't land in time (i.e. remove our call to deepcopy on the instance that has the logger thing) | 22:33 |
| clayg | no of course not - that would have been insane - yeah I think as long as to_dict returns only python primatives it's just better | 22:33 |
| swifterdarrell | clayg: we don't currently, it's just shallow assignments of {'replicas': self.replicas, ... } | 22:34 |
| torgomatic | I'm pretty sure the memo dict is just there to stop deepcopy from recursing forever, so the only time it'd really matter is if you had a data structure with two RingBuilders that shared internal arrays | 22:34 |
| clayg | yup *shurg* | 22:34 |
| torgomatic | if you deepcopy with the code we have now, the copy will have two RingBuilders that don't share arrays because we don't pass along the memoizer dict | 22:34 |
| swifterdarrell | torgomatic: ya, i'm just saying that it's slightly better (less confusing to a new reader, if nothing else) to do the plumbing like the docs say you should | 22:35 |
| torgomatic | swifterdarrell: sure | 22:35 |
| torgomatic | one sec | 22:35 |
| torgomatic | fix made, running tests now | 22:35 |
| swifterdarrell | torgomatic: ya, i don't think it's a correctness issue at all | 22:35 |
| swifterdarrell | torgomatic: clayg: and I can totally live w/o pushing deepcopy crap into to_dict() | 22:36 |
| torgomatic | swifterdarrell: sounds good to me | 22:36 |
| torgomatic | I just passed the memo dict to deepcopy() like the docs say | 22:36 |
| torgomatic | and that way we don't need to copy all this stuff just to serialize it to disk | 22:37 |
| torgomatic | which is nice | 22:37 |
| *** fbo has joined #openstack-swift | 22:37 | |
| clayg | for sure +1 there | 22:37 |
| openstackgerrit | Samuel Merritt proposed openstack/swift: Make RingBuilders deep-copy-able https://review.openstack.org/175567 | 22:39 |
| torgomatic | swifterdarrell: clayg: there you go ^^ | 22:39 |
| clayg | swifterdarrell: torgomatic: how do you know you have the placement of memo update in the right place? | 22:45 |
| clayg | swifterdarrell: torgomatic: if I have a list of ring builders and I deepcopy it - will I get a fresh copy of each builder even if the same builder appears in the original list twice?! | 22:45 |
| *** annegentle has quit IRC | 22:45 | |
| clayg | these are things I start to think about as soon as you start poking at doing it "right" | 22:46 |
| swifterdarrell | clayg: torgomatic: I think for cycle detection you sort of want new_thing = some_class(); mem[id(old_thing)] = new_thing; <deepcopy-old_thing-into-new_thing> | 22:46 |
| *** shakamunyi has quit IRC | 22:46 | |
| clayg | swifterdarrell: you stepped in it now bro - better go ahead and shoot torgomatic a diff with a test :P | 22:46 |
| openstackgerrit | Denis Cavalcante proposed openstack/swift: Add storage policy support for sorting method https://review.openstack.org/160877 | 22:47 |
| swifterdarrell | clayg: ...wondered what that smell was | 22:47 |
| openstackgerrit | Denis Cavalcante proposed openstack/swift: Add storage policy support for sorting method https://review.openstack.org/160877 | 22:59 |
| *** jd__ has quit IRC | 23:08 | |
| openstackgerrit | Denis Cavalcante proposed openstack/swift: Add storage policy support for sorting method https://review.openstack.org/160877 | 23:08 |
| *** jd__ has joined #openstack-swift | 23:08 | |
| *** david-lyle has joined #openstack-swift | 23:24 | |
| *** petertr7 has quit IRC | 23:31 | |
| *** minwoob has quit IRC | 23:58 | |
| *** thumpba_ has quit IRC | 23:59 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!