notmyname | GET /time ===>> 5:07 Brain Full | 00:07 |
---|---|---|
peluse_ | clayg: I think we can find solutions for any scenario, just didn't seem like it was worth the effort out of the gate to cover those | 00:09 |
openstackgerrit | paul luse proposed a change to openstack/swift: Make quarantine and tmp dirs in DiskFile Storage Policy Aware https://review.openstack.org/96313 | 00:15 |
notmyname | peluse_: ah ok. so tmp..tmp-N and quarantined/objects..quarantined/objects-N | 00:20 |
*** ozialien has quit IRC | 00:20 | |
*** matsuhashi has joined #openstack-swift | 00:22 | |
*** bkopilov has joined #openstack-swift | 00:23 | |
clayg | oh no no no no | 00:23 |
clayg | well... hrmm.... | 00:23 |
clayg | so for container PUT w/o specifying the sp we were sending down the PUT to the containers bare, which when you're doing a rolling restart of your container servers to update swift.conf your nodes might have different ideas of what's the default :\ | 00:29 |
clayg | I'm gunna have the proxy select the default and push down the spi with the request - always | 00:30 |
notmyname | good thing you have that reconciler | 00:31 |
notmyname | ;-) | 00:31 |
openstackgerrit | A change was merged to openstack/swift: Fixes probe tests with non-zero default storage policy https://review.openstack.org/87492 | 00:34 |
yuan | creiht: clayg: swift-get-nodes accept the first parameter as the ring, so if the ring is specified, we could get the datadir from the ring name. we also plan to make swift-get-nodes to accept one policy name parameter, like 'swift-get-nodes -P nada a/c/o'. If you specified both the ring and the policy name and they're missmatched, it will print some alert message | 00:39 |
clayg | yuan: that all sounds *awesome* - good to have you in chan | 00:39 |
yuan | clag, thanks, let me work harder to address this today. It's 8:40am here now. | 00:41 |
notmyname | yuan: any questions or concerns on how the merge to master is happening? | 00:41 |
peluse_ | hey, yuan, cool! | 00:41 |
peluse_ | clayg: was your 'oh no no no no' thinking aloud wrt the prev discussion or related to the tmp/quarantine patch I just submitted? | 00:43 |
*** ozialien has joined #openstack-swift | 00:43 | |
*** foexle has quit IRC | 00:43 | |
clayg | the sp-headers thing | 00:45 |
clayg | i haven't looked at the tmp/q thing yet - i'm sure it's fine | 00:45 |
peluse_ | clayg: OK, just checking... off tun run some errands/buy some beer. Back in a bit :) | 00:45 |
* notmyname will likewise be in and out for the rest of the evening | 00:46 | |
*** thehybridtech has joined #openstack-swift | 00:48 | |
yuan | notmyname: no questions here but feel really excited about it! | 00:48 |
*** fifieldt has joined #openstack-swift | 01:00 | |
*** dmorita has joined #openstack-swift | 01:02 | |
*** shakamunyi has quit IRC | 01:16 | |
*** ozialien has quit IRC | 01:26 | |
zaitcev | "Downloading refs/changes/49/96049/1 from gerrit" | 01:26 |
*** Edward-Zhang has quit IRC | 01:26 | |
zaitcev | there's still time to issue cd(1) without arguments and drink myself senseless | 01:27 |
*** baojg has joined #openstack-swift | 01:35 | |
clayg | lol @ awkward unix reference to "go home" | 01:37 |
*** gyee has quit IRC | 01:41 | |
zaitcev | Actually I meant it literally. | 01:43 |
clayg | lol - do you HAVE to be in ~ to drink? | 01:43 |
*** nosnos has joined #openstack-swift | 01:45 | |
clayg | ... maybe safer | 01:50 |
*** haomaiwang has joined #openstack-swift | 01:51 | |
*** mwstorer has quit IRC | 01:52 | |
*** gyee has joined #openstack-swift | 01:54 | |
*** tusharsg has quit IRC | 02:00 | |
*** tusharsg has joined #openstack-swift | 02:06 | |
*** Edward-Zhang has joined #openstack-swift | 02:16 | |
*** zhiyan_ is now known as zhiyan | 02:34 | |
*** matsuhashi has quit IRC | 03:00 | |
*** judd7_ has quit IRC | 03:02 | |
*** matsuhashi has joined #openstack-swift | 03:04 | |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Allow to deprecate policy https://review.openstack.org/86827 | 03:28 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: make policies sortable https://review.openstack.org/96341 | 03:28 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: always send spi with container PUT https://review.openstack.org/96342 | 03:28 |
*** bkopilov has quit IRC | 03:28 | |
*** nosnos has quit IRC | 03:33 | |
*** matsuhashi has quit IRC | 03:34 | |
*** matsuhashi has joined #openstack-swift | 03:35 | |
*** matsuhashi has quit IRC | 03:39 | |
*** bkopilov has joined #openstack-swift | 03:41 | |
*** zhiyan is now known as zhiyan_ | 03:47 | |
*** shakamunyi has joined #openstack-swift | 03:48 | |
*** matsuhashi has joined #openstack-swift | 03:56 | |
*** Edward-Zhang has quit IRC | 04:03 | |
*** shakamunyi has quit IRC | 04:05 | |
*** baojg has quit IRC | 04:07 | |
*** baojg has joined #openstack-swift | 04:07 | |
clayg | yuan: you still online? | 04:09 |
*** ppai has joined #openstack-swift | 04:10 | |
*** zhiyan_ is now known as zhiyan | 04:10 | |
clayg | yuan: i took a stab at some cleanups for deprecating polices - another fix to container PUT fell out -> https://review.openstack.org/#/c/96342/1 | 04:11 |
*** psharma has joined #openstack-swift | 04:18 | |
*** Edward-Zhang has joined #openstack-swift | 04:33 | |
clayg | https://www.youtube.com/watch?v=o0YWRXJsMyM | 04:39 |
*** bkopilov has quit IRC | 04:40 | |
*** bkopilov has joined #openstack-swift | 04:41 | |
*** nosnos has joined #openstack-swift | 04:45 | |
*** zhiyan is now known as zhiyan_ | 04:52 | |
madhuri | clayg: need some help | 05:00 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Add functional tests for Storage Policy https://review.openstack.org/96042 | 05:00 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Update swift-container-info to be storage policy aware https://review.openstack.org/96043 | 05:00 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Merge container storage_policy_index https://review.openstack.org/96040 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Add Storage Policy Support to Accounts https://review.openstack.org/96041 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Add Storage Policy Support to list_endpoints https://review.openstack.org/96046 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Add Storage Policy Support to Container Sync https://review.openstack.org/96047 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Enqueue misplaced objects during container replication https://review.openstack.org/96044 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Add LRUCache to common.utils https://review.openstack.org/96045 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Add Storage Policy Support to ssync https://review.openstack.org/96034 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Put X-Timestamp in object 404 responses https://review.openstack.org/96035 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Add Storage Policy Support to the Auditor https://review.openstack.org/96032 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Add storage policy support for the Replicator https://review.openstack.org/96033 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Add container-reconciler daemon https://review.openstack.org/96038 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Add reconciler probetest outline https://review.openstack.org/96039 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Extend interface on InternalClient https://review.openstack.org/96036 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Extend direct_client https://review.openstack.org/96037 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Add Storage Policy Support https://review.openstack.org/96027 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Add Storage Policy Documentation https://review.openstack.org/96026 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Add Storage Policy support to Containers https://review.openstack.org/96029 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Update FakeRing and FakeLogger https://review.openstack.org/96028 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Add Storage Policy support to Object Updates https://review.openstack.org/96031 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Add Storage Policy support to Object Server https://review.openstack.org/96030 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Add Storage Policy support to the Account Reaper https://review.openstack.org/96049 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Add Storage Policy Support to Recon Middleware https://review.openstack.org/96048 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Update container storage_policy_index when auditing https://review.openstack.org/96352 | 05:01 |
openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Fixes probe tests with non-zero default storage policy https://review.openstack.org/96353 | 05:01 |
clayg | madhuri: nope | 05:02 |
*** shakamunyi has joined #openstack-swift | 05:02 | |
madhuri | clayg: ?? | 05:03 |
clayg | madhuri: oh heh, i thought you were asking a question | 05:04 |
clayg | madhuri: what's up | 05:04 |
clayg | ? | 05:04 |
*** bkopilov has quit IRC | 05:04 | |
*** bkopilov has joined #openstack-swift | 05:05 | |
yuan | clayg: thanks a lot, very cool | 05:05 |
clayg | yuan: ! yeah np, thanks for everything | 05:08 |
clayg | yuan: how you feeling on the bin scripts? | 05:09 |
*** Edward-Zhang has quit IRC | 05:09 | |
*** Edward-Zhang has joined #openstack-swift | 05:10 | |
yuan | clayg: the tests are bit tough than the scripts | 05:10 |
clayg | lol - yeah that sounds about right :\ | 05:11 |
*** shakamunyi has quit IRC | 05:14 | |
clayg | hrmm.... gerrits handling of rebases is very interesting | 05:22 |
*** haomaiwang has quit IRC | 05:24 | |
*** haomaiwa_ has joined #openstack-swift | 05:25 | |
*** zhiyan_ is now known as zhiyan | 05:29 | |
openstackgerrit | A change was merged to openstack/swift: Make FakeRing use more real code https://review.openstack.org/90933 | 05:29 |
clayg | madhuri: did you need something? | 05:36 |
clayg | madhuri: i'm about to head out | 05:36 |
clayg | yuan: you good? | 05:36 |
*** thehybridtech has quit IRC | 05:37 | |
clayg | well g'night | 05:37 |
yuan | clayg: have a good sleep | 05:38 |
*** shakamunyi has joined #openstack-swift | 05:40 | |
*** haomaiwa_ has quit IRC | 05:42 | |
*** haomaiwang has joined #openstack-swift | 05:43 | |
*** haomaiw__ has joined #openstack-swift | 05:47 | |
*** nshaikh has joined #openstack-swift | 05:48 | |
*** haomaiwang has quit IRC | 05:50 | |
*** shakamunyi has quit IRC | 05:50 | |
hugokuo | how are you today | 06:10 |
*** tusharsg has quit IRC | 06:15 | |
*** zaitcev has quit IRC | 06:20 | |
*** tusharsg has joined #openstack-swift | 06:21 | |
*** shakamunyi has joined #openstack-swift | 06:47 | |
*** shakamunyi has quit IRC | 06:59 | |
*** Edward-Zhang has quit IRC | 07:01 | |
*** haomaiw__ has quit IRC | 07:05 | |
*** haomaiwang has joined #openstack-swift | 07:06 | |
*** haomaiw__ has joined #openstack-swift | 07:09 | |
*** gyee has quit IRC | 07:10 | |
*** haomaiwang has quit IRC | 07:12 | |
*** shakamunyi has joined #openstack-swift | 07:25 | |
*** Edward-Zhang has joined #openstack-swift | 07:26 | |
*** shakamunyi has quit IRC | 07:36 | |
*** Edward-Zhang_ has joined #openstack-swift | 07:38 | |
*** Edward-Zhang has quit IRC | 07:38 | |
*** Edward-Zhang_ is now known as Edward-Zhang | 07:38 | |
*** ppai has quit IRC | 07:39 | |
*** ppai has joined #openstack-swift | 07:56 | |
*** baojg has quit IRC | 07:58 | |
*** matsuhashi has quit IRC | 07:58 | |
*** matsuhashi has joined #openstack-swift | 07:59 | |
*** mkollaro has joined #openstack-swift | 08:00 | |
*** matsuhas_ has joined #openstack-swift | 08:02 | |
*** shakamunyi has joined #openstack-swift | 08:03 | |
*** matsuhashi has quit IRC | 08:04 | |
*** shakamunyi has quit IRC | 08:14 | |
*** ppai has quit IRC | 08:20 | |
*** Midnightmyth has joined #openstack-swift | 08:28 | |
*** baojg has joined #openstack-swift | 08:28 | |
*** ppai has joined #openstack-swift | 08:33 | |
*** shakamunyi has joined #openstack-swift | 08:40 | |
*** shakamunyi has quit IRC | 08:52 | |
*** matsuhas_ has quit IRC | 08:53 | |
*** matsuhashi has joined #openstack-swift | 08:58 | |
*** foexle has joined #openstack-swift | 09:09 | |
*** shakamunyi has joined #openstack-swift | 09:19 | |
*** foexle has quit IRC | 09:21 | |
*** ppai has quit IRC | 09:23 | |
openstackgerrit | Christian Berendt proposed a change to openstack/swift: replaced e.message with str(e) https://review.openstack.org/93570 | 09:25 |
*** zhiyan is now known as zhiyan_ | 09:30 | |
*** ppai has joined #openstack-swift | 09:31 | |
openstackgerrit | Christian Berendt proposed a change to openstack/swift: change assertEquals to assertEqual https://review.openstack.org/93581 | 09:31 |
*** shakamunyi has quit IRC | 09:31 | |
*** matsuhashi has quit IRC | 09:33 | |
*** matsuhashi has joined #openstack-swift | 09:33 | |
openstackgerrit | Joel Wright proposed a change to openstack/python-swiftclient: Add importable SwiftService incorporating shell.py logic https://review.openstack.org/85453 | 09:33 |
*** matsuhashi has quit IRC | 09:35 | |
*** matsuhashi has joined #openstack-swift | 09:36 | |
*** mlipchuk has joined #openstack-swift | 09:37 | |
*** ccorrigan has joined #openstack-swift | 09:37 | |
openstackgerrit | Christian Berendt proposed a change to openstack/swift: change assertEquals to assertEqual https://review.openstack.org/93581 | 09:39 |
openstackgerrit | Joel Wright proposed a change to openstack/python-swiftclient: Add importable SwiftService incorporating shell.py logic https://review.openstack.org/85453 | 09:46 |
*** ppai has quit IRC | 09:49 | |
*** shakamunyi has joined #openstack-swift | 09:58 | |
*** nosnos has quit IRC | 10:02 | |
*** nosnos has joined #openstack-swift | 10:05 | |
*** shakamunyi has quit IRC | 10:05 | |
*** shakamunyi has joined #openstack-swift | 10:05 | |
*** ppai has joined #openstack-swift | 10:06 | |
*** Edward-Zhang has quit IRC | 10:07 | |
*** omame has quit IRC | 10:10 | |
*** shakamunyi has quit IRC | 10:10 | |
*** shakamunyi has joined #openstack-swift | 10:10 | |
*** dmorita has quit IRC | 10:12 | |
*** foexle has joined #openstack-swift | 10:16 | |
*** ahale has joined #openstack-swift | 10:16 | |
*** omame has joined #openstack-swift | 10:17 | |
*** shakamunyi has quit IRC | 10:20 | |
*** foexle has quit IRC | 10:30 | |
*** ppai has quit IRC | 10:35 | |
*** matsuhashi has quit IRC | 10:41 | |
*** nosnos has quit IRC | 10:41 | |
*** nosnos has joined #openstack-swift | 10:45 | |
*** jamie_h has joined #openstack-swift | 10:46 | |
*** nosnos has quit IRC | 10:46 | |
*** shakamunyi has joined #openstack-swift | 10:47 | |
*** nosnos has joined #openstack-swift | 10:47 | |
*** nosnos has quit IRC | 10:48 | |
*** shakamunyi has quit IRC | 11:00 | |
*** Midnightmyth has quit IRC | 11:02 | |
*** MooingLemur has quit IRC | 11:16 | |
*** MooingLemur has joined #openstack-swift | 11:18 | |
*** shakamunyi has joined #openstack-swift | 11:27 | |
*** Midnightmyth has joined #openstack-swift | 11:34 | |
*** shakamunyi has quit IRC | 11:41 | |
*** ppai has joined #openstack-swift | 11:43 | |
*** mlipchuk has quit IRC | 11:43 | |
*** foexle has joined #openstack-swift | 11:44 | |
*** haomaiw__ has quit IRC | 12:04 | |
*** haomaiwang has joined #openstack-swift | 12:04 | |
*** Midnightmyth has quit IRC | 12:06 | |
*** shakamunyi has joined #openstack-swift | 12:07 | |
*** shakamunyi has quit IRC | 12:20 | |
*** haomai___ has joined #openstack-swift | 12:22 | |
*** haomaiwang has quit IRC | 12:24 | |
*** baojg has quit IRC | 12:27 | |
*** baojg has joined #openstack-swift | 12:28 | |
*** baojg has quit IRC | 12:32 | |
openstackgerrit | Joel Wright proposed a change to openstack/python-swiftclient: Add importable SwiftService incorporating shell.py logic https://review.openstack.org/85453 | 12:53 |
*** miqui has joined #openstack-swift | 12:58 | |
gholt | It's kind of amazing, that Jenkins. Fails all the time for no reason with no explanation and then just passed 4 unit tests somehow that should have had zero chance of passing (extra kwarg in __init__). | 13:05 |
gholt | Systems you trust to gate your code... | 13:07 |
*** zhiyan_ is now known as zhiyan | 13:11 | |
*** ppai has quit IRC | 13:12 | |
*** shakamunyi has joined #openstack-swift | 13:16 | |
*** shakayumi has joined #openstack-swift | 13:18 | |
*** shakamunyi has quit IRC | 13:21 | |
*** elambert has quit IRC | 13:27 | |
gholt | With the timing, I guess the hackathon is going to end up being the code-roast-a-thon, eh? :) | 13:27 |
*** shakayumi has quit IRC | 13:34 | |
*** jamie_h has quit IRC | 13:38 | |
*** kenhui has joined #openstack-swift | 13:40 | |
*** Midnightmyth has joined #openstack-swift | 13:41 | |
*** shakayumi has joined #openstack-swift | 13:48 | |
acoles | gholt: you mean the test_container failures? looks like they pass for 96029 but then fail for 96030 onwards which is where the arg is removed from the proxy init. | 13:49 |
gholt | Yeah, I'm just realizing the new commit "set" isn't. :/ | 13:50 |
gholt | The doc/saio/bin/remakerings is in one new patchset but not in the final either. And probably other stuff. | 13:51 |
gholt | Well, the file is there, but not the new additions I mean. | 13:51 |
acoles | gholt: took me a while to figure what was going on ;) | 13:51 |
gholt | clayg: Sorry, I'm sure this is a pain in the ass, but yeah, something wonky is going on. ;) | 13:52 |
acoles | gholt: hm, remakerings looks ok to me. i did fresh git review -d 96049 today | 13:54 |
gholt | I'll try again, I had done a delete of the old branch and redid the review -d, but... I also did a git -m on another patch set in the chain, so maybe that effed everything up. Gotta love the whole Gerrit rewriting git history crap. | 13:55 |
acoles | gholt: and once i figured it, jenkins is correct on the tests. phew, the world can rest easy ;) | 13:55 |
gholt | And you're right, a redelete and fresh review -d got remakerings back in line. Grrr. :) | 13:56 |
acoles | gholt: likewise, deleted existing branch, then review -d | 13:56 |
*** haomai___ has quit IRC | 13:56 | |
gholt | So, if you ever do a git -m ... be sure to redo any dependent branches you might have. | 13:57 |
*** haomaiwang has joined #openstack-swift | 13:57 | |
*** nshaikh has quit IRC | 13:57 | |
gholt | -m is pretty useful, so I guess it's worth the small pain | 13:57 |
gholt | So I guess just the proxy.server.Application missing kwarg is the only real issue so far. | 13:59 |
gholt | clayg: ^ | 13:59 |
*** bkopilov has quit IRC | 14:00 | |
*** haomai___ has joined #openstack-swift | 14:06 | |
*** haomai___ has quit IRC | 14:07 | |
*** haoma____ has joined #openstack-swift | 14:07 | |
*** haomaiwang has quit IRC | 14:09 | |
*** bkopilov has joined #openstack-swift | 14:13 | |
*** psharma has quit IRC | 14:14 | |
*** diegows has joined #openstack-swift | 14:22 | |
*** byeager has joined #openstack-swift | 14:26 | |
openstackgerrit | paul luse proposed a change to openstack/swift: Add Storage Policy support to Containers https://review.openstack.org/96029 | 14:28 |
openstackgerrit | paul luse proposed a change to openstack/swift: Add Storage Policy support to Containers https://review.openstack.org/96029 | 14:37 |
*** jamie_h has joined #openstack-swift | 14:38 | |
*** zhiyan is now known as zhiyan_ | 14:52 | |
*** baojg has joined #openstack-swift | 14:53 | |
openstackgerrit | Yuan Zhou proposed a change to openstack/swift: Update swift-object-info/swift-get-nodes to be storage policy aware https://review.openstack.org/82734 | 15:14 |
*** kashyapk has joined #openstack-swift | 15:14 | |
openstackgerrit | Christian Berendt proposed a change to openstack/swift: change assertEquals to assertEqual https://review.openstack.org/93581 | 15:18 |
*** mwstorer has joined #openstack-swift | 15:23 | |
*** jergerber has joined #openstack-swift | 15:23 | |
*** elambert has joined #openstack-swift | 15:25 | |
*** kenhui has quit IRC | 15:25 | |
*** tdasilva has joined #openstack-swift | 15:26 | |
openstackgerrit | paul luse proposed a change to openstack/swift: Add Storage Policy Documentation https://review.openstack.org/96026 | 15:28 |
*** kashyapk has quit IRC | 15:29 | |
*** gholt has quit IRC | 15:31 | |
*** gholt has joined #openstack-swift | 15:32 | |
*** ChanServ sets mode: +v gholt | 15:32 | |
openstackgerrit | paul luse proposed a change to openstack/swift: Add Storage Policy Documentation https://review.openstack.org/96026 | 15:33 |
creiht | clayg: I just did a fresh checkout with git review -d 96049 | 15:33 |
creiht | unit tests are giving me 4 failures all with this error: | 15:33 |
creiht | http://paste.ronin.io/1c4f45d7 | 15:33 |
gholt | creiht: Yeah, read the scrollback and comments on the patchsets. ;P | 15:34 |
gholt | Although, peluse_ probably clobbered my patchset comments, thanks paul. :) | 15:35 |
creiht | oh | 15:35 |
creiht | heh | 15:35 |
peluse_ | gholt: do my best | 15:35 |
peluse_ | gholt: so I'm not clear on how the split into the new patch set is causing that error.. I would think the dependencies would work themselves out such that this shouldn't happen? | 15:36 |
gholt | Maybe clayg will know what to do. | 15:37 |
peluse_ | gholt: I"m sure he will... I do know that it's not a bug though :) | 15:38 |
gholt | It could be he has a split on his end somehow, but who knows, heheh. | 15:38 |
*** baojg has quit IRC | 15:46 | |
*** zhiyan_ is now known as zhiyan | 15:47 | |
*** zaitcev has joined #openstack-swift | 15:52 | |
*** ChanServ sets mode: +v zaitcev | 15:52 | |
*** zhiyan is now known as zhiyan_ | 15:57 | |
*** shakayumi is now known as shakamunyi | 15:59 | |
*** byeager has quit IRC | 16:01 | |
*** byeager has joined #openstack-swift | 16:02 | |
*** mkollaro has quit IRC | 16:03 | |
notmyname | good morning | 16:04 |
peluse_ | good morning! | 16:05 |
briancline | holler | 16:11 |
*** mkollaro has joined #openstack-swift | 16:11 | |
*** zhiyan_ is now known as zhiyan | 16:12 | |
clayg | creiht: gholt: ok unittest thing was just a rebase snafu - i have it fixed | 16:17 |
gholt | Cool | 16:18 |
peluse_ | sweet! | 16:19 |
*** nshaikh has joined #openstack-swift | 16:19 | |
clayg | anyway, it was a process break down - i wont make that same exact mistake again | 16:19 |
clayg | can you just like... remove the object_ring kwarg from the ContainerController test's __init__ and carry on? Or do you want me to refresh the patches? | 16:20 |
clayg | I really don't care - it's your inboxes - it'll come up tonight with whatever other review comments regardless | 16:21 |
*** zhiyan is now known as zhiyan_ | 16:21 | |
gholt | Should be fine tbh, already been ignoring them. | 16:22 |
peluse_ | claygg: I tried that, see patch history | 16:22 |
clayg | see... what patch... history... | 16:22 |
peluse_ | clayg: ahh, never mind. will just confuse things that you already have understood :) | 16:23 |
clayg | well... did you push up a change to gerrit? | 16:24 |
openstackgerrit | paul luse proposed a change to openstack/swift: Make quarantine and tmp dirs in DiskFile Storage Policy Aware https://review.openstack.org/96313 | 16:24 |
notmyname | anyone from clouda.ca in here? | 16:24 |
peluse_ | clayg: I did but I reverted it once I realized that it was a dependency and/or rebase thing. https://review.openstack.org/#/c/96029/ | 16:24 |
clayg | "out dated" | 16:24 |
clayg | wtf? | 16:24 |
notmyname | because today they just announced a public storage cloud powered by swift https://www.clouda.ca/technology/bulk-storage/ (which is awesome!) | 16:24 |
peluse_ | clayg: I also pushed some doc updates based on early morning comments as well... | 16:25 |
gholt | clayg: Hahah, peluse_ was trying to help, honest! | 16:25 |
clayg | peluse_: ok... so there's fixes in gerrit that I need to get back into my changes on my local changes which has already been rebased | 16:26 |
*** bgmccollum has joined #openstack-swift | 16:26 | |
clayg | you guys know how people say "never rebase if you're collaborting on branches" - and gerrit says "you must always rebase" - that's fun | 16:26 |
gholt | Heheh | 16:26 |
openstackgerrit | paul luse proposed a change to openstack/swift: Make quarantine and tmp dirs in DiskFile Storage Policy Aware https://review.openstack.org/96313 | 16:27 |
peluse_ | ^ man, I need a comment grammar/spellchecker | 16:27 |
clayg | ok, i need to run will to school... i'll try and clean this up after that | 16:27 |
peluse_ | clayg: so I'm thinking I'll just leave things alone there unless asked in the future :) | 16:28 |
peluse_ | notmyname: cool, thanks for passing that on. I like the website | 16:30 |
notmyname | peluse_: check out their home page. they burn hats | 16:30 |
peluse_ | notmyname: Love this one: https://www.clouda.ca/technology/bulk-storage/ "Reliable as F***" | 16:32 |
notmyname | :-) | 16:32 |
notmyname | peluse_: in the office here: "reliable as a fort?" | 16:35 |
zaitcev | or fish | 16:35 |
briancline | fsck? | 16:35 |
peluse_ | heh | 16:37 |
peluse_ | I have a t-shirt that I think came from an australia conf that says "what the f**k is openstack" and the two ** are little cloud icons. Maybe that's it :) | 16:38 |
notmyname | heh | 16:38 |
notmyname | peluse_: that was originally from the piston dopenstack think | 16:38 |
notmyname | but ya, the aussies were passing out a bunch of them | 16:40 |
*** Thurloat has joined #openstack-swift | 16:40 | |
*** swat30 has joined #openstack-swift | 16:41 | |
swat30 | Thurloat: yo | 16:41 |
Thurloat | Hey notmyname! swat30 and thurloat here from Cloud-A | 16:41 |
notmyname | Thurloat: swat30: hi! | 16:41 |
swat30 | notmyname: howdy! | 16:41 |
notmyname | cool to see your announcement this morning. we were just chatting about it in here | 16:41 |
Thurloat | all bad things, I presume. | 16:42 |
briancline | watched a linuxconf video discussing zuul the other night. kinda cleared up the specifics of its purpose | 16:42 |
notmyname | Thurloat: swat30: this is the IRC channel where swift contributors (ie devs and ops running prod clusters) do most of our collaboration | 16:43 |
notmyname | Thurloat: swat30: all of the voiced people (with a "+" by their nick) are core reviewers | 16:44 |
pandemicsyn | ...its where we gather to be with others who's spirit has also been crushed by running Swift in prod ;) | 16:44 |
tdasilva | briancline: do you have a link of that video? | 16:44 |
pandemicsyn | ..kidding......mostly | 16:44 |
notmyname | lol | 16:44 |
*** nshaikh has quit IRC | 16:45 | |
swat30 | haha | 16:45 |
notmyname | briancline: zuul is a pretty interesting system, I think. I think it has the potential to be one of the things in opnestack that has lasting significance | 16:45 |
swat30 | I have to say that getting it setup wasn't too bad | 16:45 |
notmyname | great to hear! | 16:45 |
*** NM has joined #openstack-swift | 16:45 | |
notmyname | swat30: and I'm told you're using keystone for auth? | 16:45 |
swat30 | notmyname: you got it | 16:46 |
notmyname | are you using other openstack projects? (eg are you using nova for the VMs?) | 16:46 |
swat30 | notmyname: pretty much using it exclusively for all of our stuff. Nova, Cinder, Glance, Neutron, and now Swift | 16:46 |
swat30 | and keystone of course | 16:47 |
notmyname | ah ok | 16:47 |
*** byeager has quit IRC | 16:47 | |
notmyname | looks like my 9:30 meeting is starting now. I'll be back later | 16:48 |
swat30 | notmyname: cool! great meeting you | 16:48 |
notmyname | swat30: Thurloat: please stay, ask questions, answer questions. jump it. great to have you here! | 16:48 |
swat30 | we'll hang around | 16:48 |
briancline | tdasilva: it was this one: https://www.youtube.com/watch?v=_xkcAnWnlrY | 16:48 |
Thurloat | thanks for the welcome notmyname :) | 16:48 |
tdasilva | briancline: thanks! | 16:48 |
briancline | tdasilva: however, it appears there's a newer one now: https://www.youtube.com/watch?v=xTPILCPRGeg | 16:48 |
tdasilva | briancline: ok, thx, will check it out... | 16:49 |
briancline | I think I need a business excuse to attend an AU linuxconf | 16:50 |
briancline | notmyname: yeah, I agree. I was pretty fascinated once I heard more about which pieces it was doing behind the scenes | 16:50 |
briancline | prior to that the boundaries between it, nodepool, et al. were unclear | 16:50 |
*** haoma____ has quit IRC | 16:58 | |
Thurloat | Is the swift horizon dashbaord a swift thing, or a horizon thing? | 17:09 |
*** kashyapk has joined #openstack-swift | 17:09 | |
swat30 | Thurloat: pretty sure anything horizon is done separately in the horizon project | 17:09 |
swat30 | separate team etc. | 17:10 |
*** zhiyan_ is now known as zhiyan | 17:12 | |
*** kashyapk has quit IRC | 17:15 | |
clayg | peluse_: no worries, i'm sure I can get it squared - but yeah comments on the changes are the best way for me to manage things moving forward | 17:17 |
*** lpabon has joined #openstack-swift | 17:19 | |
clayg | peluse_: do you think it matters that the docs reference a bunch of list endpoints stuff that doesn't work until later in the patch chain when the list endpoints gets updated to work with policies | 17:20 |
clayg | creiht: gholt: ^ maybe doesn't matter to you guys since you're mostly reviewing the end-of-the-line | 17:21 |
*** zhiyan is now known as zhiyan_ | 17:22 | |
gholt | I don't think it matters, except I wonder how Jenkins will handle merges to master later. | 17:22 |
gholt | And sorry for all the splng grmr stuff. I probably started everyone off on the wrong foot. ;) | 17:23 |
*** jergerber has quit IRC | 17:23 | |
gholt | Man, assertEquals really offends that guy; to the tune of way over 4,000 lines. I'm pretty if Python dropped assertEquals we could just add quick support for it calling assertEqual for us. | 17:26 |
*** pconstantine has left #openstack-swift | 17:27 | |
*** pconstantine_ has joined #openstack-swift | 17:27 | |
gholt | Cases like that call for a BDFL that says, "Please direct all that energy to something more useful." | 17:28 |
peluse_ | clayg: yeah, I don't think it matters, everyone knows its a patch chain | 17:28 |
*** bena has joined #openstack-swift | 17:28 | |
peluse_ | just curious, whats wrong with assertEuals anyway? | 17:29 |
gholt | They deprecated the plural form. | 17:29 |
clayg | who "deprecated" it? i don't think it raises a warning? | 17:30 |
peluse_ | gholt: the spelling/grammar stuff is all my fault. I feel bad it amde it this far with that many stupid things left, honestly it was reviwed like 12 times before you saw it. Imagine how bad my first cut was :) | 17:30 |
gholt | Heheheh! That's going to be true for any of us. | 17:30 |
clayg | does anyone know how to get ahold of Thiago da Silva? | 17:31 |
peluse_ | I keep trying to blame my keyboard but every time I buy a new one it just doesn't gett any better, can't figure it out | 17:31 |
peluse_ | clayg: what would you like done to him? | 17:31 |
clayg | he made a comment regarding policy "types" - and I agree that support was sorta half-baked, but it made a *little* since in the replicators so they could skip policy datadirs that weren't of type replication | 17:31 |
gholt | All I have is the email from Gerrit that says he works at RedHat. | 17:32 |
openstackgerrit | Donagh McCabe proposed a change to openstack/swift: Allow swift-recon to easily integate with Nagios https://review.openstack.org/96529 | 17:32 |
clayg | anyway, i don't really know how to address his comments - I guess he wants more doc - but it's really on there as a stub, since there's only one supported type and it's the default if not specified - it doesn't really need to be documented | 17:32 |
peluse_ | clayg: yeah, I commented on the comment because he also made it on the feature/ec version. We could add a list of currently supported types in that section and that would probably make it clear enough | 17:33 |
peluse_ | clayg: I have his email, will send it yo you | 17:34 |
bena | HI! just a comment. I am performing a test. I would like use s3curl.pl to connect to openstack swift. I found an example but I have doubt related to swift account:user (example: ./s3curl.pl --id '<account>:<user>' --key <password> --get -- -k -v -s http://<my IP>:8080). Have you any idea about what account,user I should use? | 17:34 |
clayg | peluse_: you marked some of acoles comments "done" in patchset 2, i think i got all of your changes merged into my working set - so I guess those should all be addressed. | 17:36 |
peluse_ | clayg: yup. also, check private conversation for that email adddress | 17:37 |
*** gyee has joined #openstack-swift | 17:37 | |
gholt | bena: Have you ever used Swift with Swift tools before? If so, you should have an account and user from that. | 17:39 |
* peluse_ stepping out for a few hrs.... | 17:41 | |
bena | +gholt I only download and install devstack with swift but I did not use Swift with swift tools. | 17:42 |
bena | +gholt also I added swift3 | 17:42 |
clayg | peluse_: ok so I think Thiago's point is that if you're doing swift-on-file it's silly to have the policy type described as "replication" in the /info call? | 17:42 |
clayg | notmyname: i'm just removing the Copyright line from the source files that attribute it to OpenStack Foundation | 17:45 |
*** byeager has joined #openstack-swift | 17:46 | |
*** byeager has quit IRC | 17:46 | |
*** byeager has joined #openstack-swift | 17:50 | |
*** ozialien has joined #openstack-swift | 17:51 | |
*** byeager has quit IRC | 17:51 | |
*** byeager has joined #openstack-swift | 17:52 | |
*** wasmum has joined #openstack-swift | 17:53 | |
*** byeager has quit IRC | 17:53 | |
wasmum | any swiftstack owners/operators have time for a quickie swift_undelete? | 17:54 |
*** byeager has joined #openstack-swift | 17:54 | |
ctennis | what's up wasmum? | 17:55 |
clayg | gholt: heh, it's going to have to be iter(self.by_index.values()) | 17:56 |
*** openstackgerrit has quit IRC | 17:57 | |
clayg | peluse_: so do you have a better name for Policy_0 - i'm fixing acoles bug | 17:58 |
wasmum | ctennis: thanks | 17:58 |
wasmum | ctennis: is the swift_undelete porject still active? Doing a python setup.py install results in the package being installed as /usr/local/lib/python2.7/dist-packages/swift_undelete-_VERSION_-py2.7.egg | 17:59 |
wasmum | ctennis: i add just 'undelete' delete in the pipeline, tried 'swift_undelete" as well | 17:59 |
wasmum | ctennis: not much docs on usage, so just trying to get this in and tested | 17:59 |
ctennis | yeah, it's still active, lemme look | 18:00 |
wasmum | ctennis: thank you, sir. | 18:00 |
ctennis | wasmum: It's part of our commercial package, I think it was open sourced but without much detail and probably wasn't tested standalone. | 18:01 |
wasmum | ctennis: thank you, Sir. I'll see if I can do something with it. do you all want a pull request if we can get it working in a standalone env? | 18:02 |
alpha_ori | wasmum: Yeah, in its current state it's probably only of casual interest as long as its restrictions / performance characteristics fit your use case. | 18:02 |
ctennis | wasmum: absolutely | 18:03 |
alpha_ori | wasmum: yeah, definitely. | 18:03 |
alpha_ori | wasmum: We intend to do some more focused development on it this summer. | 18:03 |
acoles | clayg: i wonder if you want to just make Policy_0 reserved? | 18:03 |
wasmum | alpha_ori: great! ctennis: I'll see what I can come up with thanks to both of you! | 18:04 |
ctennis | wasmum: you bet | 18:04 |
*** bena has quit IRC | 18:04 | |
*** zhiyan_ is now known as zhiyan | 18:13 | |
clayg | acoles: yeah that's a bug that I guess could happen - so we'll prevent it. I really want to have a better name for the existing policy exposed in /info when you upgrade your cluster w/o updating swift.conf | 18:14 |
gholt | I guess bena left. tbh, I probably wouldnt' be much help to him anyway :/ | 18:16 |
*** ozialien has quit IRC | 18:18 | |
Dieterbe | seeing a weird issue. I'm using https://github.com/ncw/swift which returns a list of all objects in a container, for a given object it claims the size is 0, though when i run "swift stat...." the file is 470MB, which i can confirm by downloading the file | 18:21 |
Dieterbe | when i sniff the traffic between the library and swift, i see it queries /v1/AUTH_vimeo/uploads?format=json&limit=10000 | 18:21 |
Dieterbe | but interestingly, the file doesn't even seem to appear anywhere in the network traffic. | 18:21 |
Dieterbe | does any of this make sense somehow? perhaps because of segments? | 18:21 |
*** zhiyan is now known as zhiyan_ | 18:23 | |
*** byeager has quit IRC | 18:31 | |
notmyname | /back | 18:31 |
*** byeager has joined #openstack-swift | 18:31 | |
notmyname | clayg: you're looking for tdasilva | 18:34 |
*** byeager has quit IRC | 18:36 | |
Thurloat | wb notmyname | 18:36 |
Thurloat | thanks for the shoutout on the twitter earlier. | 18:37 |
clayg | notmyname: I'm probably looking for peluse_ - I want to figure out if we just drop types - or if we're going to have to bite the bullet and make it extensible now. | 18:37 |
clayg | notmyname: peluse_: maybe if we just we just pull it out of /info we can punt on it for now? | 18:38 |
notmyname | Thurloat: I'm your local swift social media maven | 18:38 |
Thurloat | I'll solicit you for an RT in a couple weeks when we roll out more of the swift features, haha | 18:38 |
clayg | but if someone were to configure a storage policy with object server's using swift-on-file DiskFile it would be silly for the policy with the name GlusterFS to say it's type is 'replication' - and since those nodes won't even been running the object-replicator there's no need for the type field to be there to "skip" those directories | 18:39 |
wasmum | ctennis: for what it's worth, this is the issue i get statically specifying a version in setup.py and then adding it to the pipeline: | 18:39 |
wasmum | LookupError: Entry point 'main' not found in egg 'swift_undelete' (dir: /usr/local/lib/python2.7/dist-packages/swift_undelete-1.1-py2.7.egg; protocols: paste.filter_factory, paste.filter_app_factory; entry_points: ) | 18:39 |
wasmum | ctenni: I'm not even sure if it likes my stanza in the pipeline. well, off to the races i guess. | 18:40 |
notmyname | clayg: what are we pulling out of info? I think I missed the context. I think you're talking abut storage policy types maybe? | 18:40 |
clayg | oh shit, gholts last comments were on the update fake's chagne - he must be on the container change now :\ | 18:41 |
clayg | notmyname: yeah from tdasilva from redhat asking about why types are there if there's only one type? | 18:41 |
clayg | peluse_: did you push some changes onto 96029 as well? I have no idea how I'm going to track down if i'm stomping on anything you did from underneath me man :\ | 18:42 |
gholt | clayg: Actually, I haven't done much reviewing in the past couple hours. | 18:43 |
notmyname | clayg: peluse_: ah. yeah I had a similar question, but I can see why it would be more important to tdasilva :-) | 18:43 |
clayg | ok, well the stupid kwarg thing on the one change that i had to add in when i moved a test to an earlier commit but didn't pull out again in the later commit made last nights rebases pretty messy - and I think i'm all caught up on this mornings comments/changes (mostly docs, yay!) - do you guys want a clean push or just wait till tonight? | 18:46 |
clayg | peluse_: and nm, regarding the changes on 96029 - i see what you were trying to do - it's fine. | 18:47 |
clayg | peluse_: did you push anything else - or just that and docs? | 18:47 |
*** byeager has joined #openstack-swift | 18:49 | |
tdasilva | clayg: yes, I agree that in the case of swiftonfile it seems a little strange to define the type as 'replication' | 18:55 |
gholt | From this office, I think it's fine if you want to push right now. I think everyone is off doing other things atm. | 18:55 |
tdasilva | clayg: this would also be the case for swift-ceph-backend and other backends that don't rely on swift replication | 18:57 |
dmsimard | Hi, does anyone know if it's possible to have the ceilometer swift middleware talk to the ceilometer-api instead of the bus/database backend ? | 18:59 |
notmyname | dmsimard: the ceilometer middleware is managed by the ceilometer team, so you should ask there | 19:00 |
dmsimard | notmyname: I'm trying my luck here, haven't had much luck in #openstack-ceilometer :) | 19:00 |
clayg | tdasilva: yeah ok, point taken, when peluse_ comes back we'll figure out what we'll do there for now - thanks for the reviews - keep going! | 19:01 |
tdasilva | clayg: great work by you guys! will try to do some more reviews later today... | 19:03 |
*** pberis1 has joined #openstack-swift | 19:03 | |
wasmum | ctennis: got it | 19:03 |
wasmum | ctennis: this will work on a syncstack aio | 19:04 |
wasmum | ctennis: i just had my middleware filter dorked | 19:04 |
wasmum | ctennis: it should look like: [filter:swift_undelete] paste.filter_factory = swift_undelete:filter_factory | 19:05 |
wasmum | with the swift_undelete loaded in the pipeline before the proxy server | 19:05 |
wasmum | in the setup.py, you can optionall set a version | 19:05 |
peluse_ | clayg: I'm back | 19:05 |
peluse_ | clayg: wrt types maybe only add it to /info if its not replication? | 19:06 |
*** openstackgerrit has joined #openstack-swift | 19:06 | |
*** pberis has quit IRC | 19:07 | |
peluse_ | clayg: wrt other changes I pushed, I believe just the doc | 19:07 |
ctennis | cool wasmum | 19:07 |
peluse_ | clayg: others I limited to just comments so you don't get all sorts of crap to sort out | 19:08 |
peluse_ | clayg: wrt Policy_0 I don't have a better name but do agree for now a minimum would be not to allow that as a user selected name | 19:08 |
peluse_ | tdasilva: BTW, yes, thanks for the reviews - thanks to everyone for that matter -- easy for me to say since clayg is cleaning up after all of the comments :) | 19:09 |
*** openstackgerrit has quit IRC | 19:13 | |
*** zhiyan_ is now known as zhiyan | 19:14 | |
*** kashyapk has joined #openstack-swift | 19:15 | |
*** kashyapk has quit IRC | 19:22 | |
clayg | peluse_: maybe LEGACY, DEFAULT sounds better but seems confusing if someone defines a policy-1, makes that the default, and doesn't give a name/section for storage-policy:0 | 19:22 |
clayg | peluse_: adding the PolicyError for when name == default and index != 0 was trivial | 19:23 |
*** zhiyan is now known as zhiyan_ | 19:24 | |
clayg | peluse_: re: type - i'm more keen on just ripping it out for now, and i'm not sure I see the value of having it in /info at all - we expect the interface to be the same from the clients persepctive | 19:27 |
peluse_ | clayg: yeah, I've done down the legacy and default ideas before and legacy sounds so, I dunno wrong, and default I agree is confusing with our other defn so that's how I ended up at Policy_0 | 19:27 |
peluse_ | clayg: I think when we add EC though we will want something descriptive in /info to indicate that its EC | 19:27 |
notmyname | peluse_: maybe. maybe not | 19:28 |
clayg | maybe we name it LEGACY or something and if it's hte only policy defined don't show anything in /info? | 19:28 |
peluse_ | clayg: yeah, I'm down with that | 19:28 |
notmyname | what about a non-required field that is something like "notes" where the deployer can say whatever they want up to 500 chars or something? | 19:28 |
clayg | peluse_: well you have the *name* field which can be anything - if you don't want to call it "achive-large-data" you could always stick the "erasure-encoded" in the name? | 19:28 |
clayg | notmyname: yeah, i guess - but then you have product managers telling swift deployers to update their marking docs in the swift.conf! | 19:29 |
peluse_ | or "description"... | 19:29 |
notmyname | ya, description is good | 19:29 |
clayg | meh | 19:30 |
notmyname | clayg: is that a big concern? | 19:30 |
clayg | notmyname: yeah? swift.conf is not the right place for that kind of crap - it's like really important not to f' it up | 19:30 |
clayg | I'm going to do a push - everyone go clean out your inboxes and ignore the scrollback | 19:31 |
* peluse_ is holding on... | 19:31 | |
clayg | hrmm... | 19:32 |
peluse_ | only 27 or 28 emails :) | 19:33 |
clayg | dunno why the channel didn't gripe at us | 19:33 |
*** openstackgerrit has joined #openstack-swift | 19:34 | |
notmyname | openstackgerrit: oh hai | 19:35 |
peluse_ | sympathy maybe? | 19:35 |
*** foexle has quit IRC | 19:37 | |
clayg | ok, i'm going to pull down the policy_string on q/tmp change and see about adding that | 19:41 |
clayg | i really don't know about the description feild | 19:41 |
notmyname | clayg: don't worry. swifterdarrell just explained how wrong I am :-) | 19:42 |
clayg | for 1) the swift client support for formatting tall that nested json junk is kinda weak | 19:42 |
clayg | I think the name has to be there because it's part of the api - and probably the default - same reason | 19:42 |
peluse_ | clayg: we can always punt for now and just rip type from /info and deal with when it will actually be needed, when we add EC | 19:42 |
clayg | everything else you put somewhere and reference by name | 19:42 |
clayg | peluse_: even then I don't know if the "type" enum needs to be in /info - and more to the point I'm not sure how far we can get away with it as an enum | 19:43 |
*** kutija has joined #openstack-swift | 19:43 | |
clayg | i think there's an idea that proxy code will just grow "if policy.type == 'ec': a bunch of code elif type == 'replication': a bunch of code | 19:44 |
notmyname | so pull it out for now and add it back when needed | 19:44 |
clayg | but I think it's really going to just have to get something_the_proxy_needs = policy.get_something() | 19:44 |
notmyname | and then we can determine (orthogonally) if there is a need for an annotation on /info | 19:44 |
clayg | notmyname: so leave it in there half-baked and pull it out of /info - or just rip out types until we know what f we're doing? | 19:45 |
notmyname | clayg: the 2nd | 19:45 |
clayg | yay! right answer ;) | 19:46 |
clayg | peluse_: sorry | 19:46 |
notmyname | lol | 19:46 |
clayg | i'll work on it on the train - heading into the office - looking forward to more comments on those reviews guys - i feel like you're just about to dip into the good stuff :D | 19:46 |
peluse_ | hey, I'd rather get this on to master now and worry about EC as a separate discussion | 19:46 |
clayg | peluse_: i could also maybe use some help with the doc updates for depercation support - which I just rolled into the laste change | 19:47 |
peluse_ | clayg: can do, just let me know which patches are safe to push on w/o screwing your system up | 19:47 |
peluse_ | clayg: I have time right now if that works | 19:48 |
clayg | peluse_: probably just a diff would be best? | 19:48 |
*** Thurloat has quit IRC | 19:49 | |
clayg | you could also do a PR against github.com/clayg/sp-base (i don't think i ever turned on collaborator for you, but I should...) | 19:49 |
creiht | is it safe to check the code out now? | 19:49 |
glange | never | 19:49 |
clayg | creiht: yeah thanks - sorry for this morning - i should have peeked at gerrit last night before i went to bed | 19:50 |
clayg | i was up watching into darkness (which is free on netflix btw, and not terrible) | 19:50 |
peluse_ | clayg: not totally following you, maybe I took too many allergy pills this morning... | 19:50 |
clayg | peluse_: dont' push over any of the changes purposed to master PLEASE | 19:50 |
clayg | peluse_: any other way you can think of to get me code is fine | 19:50 |
peluse_ | clayg: gotcha | 19:50 |
clayg | email a diff, paste something, push somewhere else | 19:50 |
peluse_ | print out some stuff, fax it.. whatever it takes :) | 19:51 |
clayg | i know at the conference i was all about more collaboration on changes, but this patch chain stuff is already complicated enough thanks to gerrit | 19:51 |
clayg | notmyname: oh... one more PTL hammer | 19:52 |
peluse_ | clayg: I hear ya | 19:52 |
notmyname | where? | 19:52 |
clayg | notmyname: do you want POLICY_HEADER and BACKEND_POLICY_HEADER - or do you want to just use string literals 'X-Storage-Policy' and 'X-Backend-Storage-Policy-Index' everywhere? | 19:52 |
notmyname | the global constants is fine. seems better long-term probably | 19:53 |
notmyname | clayg: oh are you changing the name too? :-) | 19:53 |
gholt | X_BACKEND_STORAGE_POLICY_INDEX should work | 19:54 |
clayg | gholt: yup | 19:54 |
clayg | notmyname: the thing about the constants - although it's nice for flake8 - it's not like you can just change the constant and it doesn't matter - in an upgrade it's a terrible mess... like they can't really change | 19:55 |
notmyname | clayg: I don't feel strongly about it either way. I figured it would be easier to have them because it's already there | 19:55 |
gholt | So its use should just be to make typing easier and typos harder or at least detectable. | 19:55 |
clayg | plus in some places the code needs to say env['HTTP_X_BACKEND_STORAGE_POLICY_INDEX'] - shoudl i have a constant for that too? | 19:55 |
notmyname | but what gholt said :-) | 19:56 |
clayg | gholt: ok, so what's easier to type than X-Backend-Storage-Policy-Index | 19:56 |
peluse_ | yeah, I made them constants because very early on we kept changing the name and I was tired of search and replace everywhere, plus I was making typos left and right :) | 19:56 |
gholt | BACKEND_SPI :) | 19:56 |
clayg | right now it's POLICY_INDEX which is ok, but notmyname says he doesn't know it's a header | 19:56 |
gholt | That'll make everyone equally unhappy. | 19:56 |
peluse_ | hey notmyname, that's a header :) | 19:56 |
notmyname | peluse_: thanks! | 19:56 |
peluse_ | hehe | 19:57 |
clayg | yeah i don't love the acronyms, or silly short spellings like HDR | 19:57 |
gholt | Oh, gotcha, it is already a constant, I forgot already. | 19:57 |
clayg | notmyname: so anyway, PTL hammer | 19:57 |
gholt | Heheh | 19:57 |
peluse_ | clayg: yeah, I'm guilty of silly short abbreviations from past jobs - long history of doing that so will take me a while to break it | 19:57 |
gholt | I do comment how we need more of that BDFL attitude around here. | 19:57 |
clayg | gholt: go for it - no one would dare chanllenge the gholt | 19:58 |
peluse_ | I know how much you love idx means "index" too | 19:58 |
clayg | yah now that i have __init__ i'm all about the int(policy) | 19:58 |
clayg | er... __int__ | 19:58 |
clayg | wow muscule memory | 19:58 |
gholt | clayg: Hah, I'd have to be PTL to be BDFL, and we know that's not my path, heheh. | 19:59 |
creiht | lol | 19:59 |
clayg | gholt: i'm just saying in this cause you can throw the hammer just as easy | 19:59 |
gholt | Oh, gotcha. Thanks :) | 19:59 |
notmyname | clayg: you're the one actually typing in the code. :-) | 20:00 |
clayg | in this *case* - and lots of other cases too | 20:00 |
clayg | heh, i got used to POLICY and POLICY_INDEX - at least they're words | 20:00 |
clayg | at first I was much more in the camp of just type the strings | 20:00 |
peluse_ | it used to be POLICY_IDX :) | 20:00 |
clayg | oh god | 20:00 |
clayg | is it tx_id or tnx_id or txn_id? | 20:01 |
gholt | one's for your driver's license here | 20:01 |
clayg | lol | 20:01 |
clayg | fuck it; they stay as is for now | 20:01 |
notmyname | if you need a decision, then: use the constants | 20:01 |
clayg | i'm going to go fart with removing types | 20:02 |
gholt | Heheheh | 20:02 |
clayg | notmyname: ok, good, we *have* constants (but not for env['HTTP_X_BACKEND_STORAGE_POLICY_INDEX']) - and they're named POLICY and POLICY_INDEX - and that's confusing because they don't look like they're headers, but we don't have other constants for headers, so everyone can get used to it - DONE | 20:03 |
peluse_ | clayg: how many cups of coffee have you this today so far? | 20:04 |
notmyname | peluse_: I'm guessing none ;-) | 20:04 |
gholt | More muscle memory on the prior post, clayg? | 20:04 |
clayg | 3, but i just poor my mug | 20:04 |
notmyname | ya, I'd prefer that it would have _HEADER on there, but it doesn't seem like many others are concerned by that, so it's not a big deal to me | 20:05 |
clayg | gholt: no i don't think so... | 20:05 |
*** wasmum has quit IRC | 20:05 | |
gholt | Okay, sorry, I'm just 5y old and found it tremendously funny. You know me. | 20:05 |
clayg | i glad folks are having a good time - me too! | 20:06 |
peluse_ | heh, OK, I'm gonna go add some dececation info to the docs - err I mean deprecation info. clayg, I'll post a diff somewhere for you in a bit | 20:06 |
clayg | notmyname: ok, i'll tack a _HEADER on them | 20:06 |
clayg | i'm offline for a bit | 20:06 |
*** byeager has quit IRC | 20:11 | |
*** byeager has joined #openstack-swift | 20:11 | |
peluse_ | clayg: I'll also do the doc update for the pending tmp and quarantined dir updates | 20:13 |
*** zhiyan_ is now known as zhiyan | 20:15 | |
*** byeager has quit IRC | 20:16 | |
*** kashyapk has joined #openstack-swift | 20:19 | |
-openstackstatus- NOTICE: Gerrit is going offline to correct an issue with a recent project rename. ETA 20:45 UTC. | 20:21 | |
*** ChanServ changes topic to "Gerrit is going offline to correct an issue with a recent project rename. ETA 20:45 UTC." | 20:22 | |
*** jergerber has joined #openstack-swift | 20:22 | |
*** kashyapk has quit IRC | 20:23 | |
*** openstackgerrit has quit IRC | 20:25 | |
*** zhiyan is now known as zhiyan_ | 20:26 | |
peluse_ | sheesh | 20:31 |
notmyname | peluse_: you didn't want to review any code, did you? | 20:32 |
peluse_ | nah, there's nothing up there :) | 20:34 |
*** swat30 has quit IRC | 20:37 | |
*** ChanServ changes topic to "Storage Policies have been proposed to master. Soft freeze in effect until these patches land | First SP patch: https://review.openstack.org/#/c/96026/ | Last SP patch: https://review.openstack.org/#/c/96049/" | 20:40 | |
-openstackstatus- NOTICE: Gerrit is back online | 20:40 | |
*** byeager has joined #openstack-swift | 20:42 | |
zaitcev | portante: sorry it took me 3 years, but xattr is a problem no more https://bugzilla.redhat.com/show_bug.cgi?id=1102926 | 20:44 |
portante | great, thanks | 20:46 |
portante | will the Fedora community actually accept that, though? | 20:47 |
*** byeager_ has joined #openstack-swift | 20:49 | |
zaitcev | It certainly is a downstream patch to be carried forever, which we do not like in general. | 20:49 |
zaitcev | However, I think I considered various approaches. We could go ahead and package python-xattr. Russell had a review languishing somewhere. | 20:50 |
*** byeager has quit IRC | 20:50 | |
portante | Which is more likely to be accepted by Fedora? | 20:51 |
*** shakamunyi has quit IRC | 20:51 | |
*** shakamunyi has joined #openstack-swift | 20:52 | |
*** ozialien has joined #openstack-swift | 20:54 | |
zaitcev | portante: I sent e-mail to rbryant seeking his opinion. | 20:58 |
*** byeager_ has quit IRC | 20:59 | |
portante | great | 20:59 |
zaitcev | you're on cc | 20:59 |
*** byeager has joined #openstack-swift | 20:59 | |
zaitcev | I need to work on being more laconic. | 21:00 |
portante | thanks | 21:00 |
zaitcev | oh nice, I already have xattr tag https://slasti.zaitcev.us/zaitcev/xattr/ | 21:02 |
peluse_ | clayg: just emailed you some doc updates - let me know if you don't get them | 21:03 |
zaitcev | NOW | 21:03 |
*** fifieldt has quit IRC | 21:03 | |
zaitcev | With that 3-year-old monkey of xattr off my back, I may turn to Storage Policies and tear them like cat a coach. | 21:03 |
zaitcev | Unless Peter finds some other horror in Red Hat Distribution of OpenStack (RDO[tm]). | 21:04 |
peluse_ | well that sounds enthusiastic (I think)... | 21:04 |
dfg | clayg: you there? | 21:07 |
*** lpabon has quit IRC | 21:10 | |
*** shakamunyi has quit IRC | 21:14 | |
clayg | dfg: ohai | 21:17 |
clayg | peluse_: thanks | 21:17 |
*** zhiyan_ is now known as zhiyan | 21:18 | |
clayg | peluse_: fwiw the types extraction went over pretty well | 21:18 |
dfg | clayg: just added a comment on one of them. | 21:18 |
peluse_ | clayg: sweet | 21:19 |
clayg | the renaming of the POLICY constants is going to be more work (the sed introduce a bunch of flake8's) - so i'd just like everyone to be sure about POLICY_HEADER and POLICY_INDEX_HEADER | 21:19 |
dfg | clayg: going to my swift reviews page and seeing a gazillion patches all added by you in the same minute is giong to become a recurring nightmare for me :) | 21:19 |
creiht | heh | 21:20 |
*** fifieldt has joined #openstack-swift | 21:20 | |
creiht | clayg, notmyname: It might be useful to have a wiki page with a link to all of the policy reviews in order | 21:20 |
creiht | to make it easier to keep track of | 21:20 |
clayg | dfg: yes, i'm sorry for that - perhaps I can never ever ever do this again? | 21:20 |
notmyname | creiht: (like gholt) I've got a star on the first and the last one | 21:20 |
dfg | :) no- its just because of gerrit | 21:21 |
clayg | creiht: too bad stupid gerrit can't just cough up somethign like that for us | 21:21 |
*** NM has quit IRC | 21:21 | |
creiht | clayg: yeah | 21:21 |
creiht | notmyname: hrm | 21:21 |
* creiht goes to look | 21:21 | |
clayg | so this page is not *terrible* but it's not in order either - https://review.openstack.org/#/q/status:open+project:openstack/swift+branch:master+topic:bp/storage-policies,n,z | 21:22 |
notmyname | creiht: I cna get it. will only take a few minutes | 21:22 |
creiht | clayg: I think that is decent enough | 21:23 |
clayg | dfg: when you have a comment can you go ahead and stick on -1 on it | 21:24 |
clayg | dfg: a) it' helps me keep my moral beat down b) i can find it | 21:24 |
dfg | haha | 21:24 |
dfg | a) is enough :) | 21:24 |
glange | wow | 21:24 |
dfg | crap- now I don't know where i added it... | 21:25 |
clayg | torololoololololo | 21:25 |
clayg | glange: when are you going to start reviewing it? | 21:25 |
dfg | https://review.openstack.org/#/c/96027/3 | 21:25 |
peluse_ | dfg, I saw it... was hunting through my archives to remember the justification - that lazy loading of the ring dates back to early last summer | 21:25 |
glange | I read the docs :) | 21:25 |
clayg | glange: no grammer fix comments!? | 21:26 |
dfg | he had some but they were all wrong | 21:26 |
glange | my grammer is not 100%, I thought maybe you should have a tech writer go over it :) | 21:26 |
glange | grammar! <-- it has an "a" | 21:27 |
*** zhiyan is now known as zhiyan_ | 21:27 | |
glange | it had a very conversational style :) | 21:27 |
notmyname | clayg: http://b.z19r.com/post/did-you-just-tell-me-to-go-fuck-myself ? | 21:28 |
dfg | peluse_: clayg seems like it would be better to load it on proxy init. there aren't going to be too many of these things are there? and it would be slow to pull it in on a per request basis. also the 500 needs to get fixed. | 21:30 |
peluse_ | dfg... hmm I found the patch where I used to instantiate all the rings at init... looking for the input from torgomatic where we changed it | 21:30 |
peluse_ | it had to do with how the proxy test code ended up.... | 21:31 |
clayg | dfg: yeah that's weird, what happens on master if you don't have an object.ring.gz | 21:31 |
dfg | clayg: ya I just tried it out. maybe i'm full of it... | 21:32 |
dfg | clayg: oh- the same thing. proxy starts and 500s when you get an obj | 21:33 |
clayg | dfg: no no no, we can make all the things better | 21:33 |
dfg | well- maybe you can :p | 21:33 |
*** openstackgerrit has joined #openstack-swift | 21:34 | |
clayg | idk, werid, so does master do the same thing with container/account rings? or only objects? | 21:34 |
peluse_ | same thing if the ring is missing you mean? | 21:34 |
clayg | yeah | 21:34 |
clayg | i guess with the automatic reloading of the ring data maybe it's... idk - very strange | 21:35 |
*** NM has joined #openstack-swift | 21:36 | |
dfg | peluse_: ya- | 21:36 |
*** byeager has quit IRC | 21:37 | |
clayg | peluse_: dfg: i'm cool with a for policy in POLICIES: policy.load_ring(swift_dir) in proxy's __init__ | 21:37 |
peluse_ | so I can't find details on why we changed to lazy loading, just found the change. I'm certain it was because of unit test's ability to pass in a ring collection. So many things have changed in that area since that decision was made maybe its something to think about now, clayg? | 21:37 |
dfg | clayg: ya- if there's no container.ring proxy won't start | 21:38 |
peluse_ | clayg: that's basically how it was | 21:38 |
*** byeager has joined #openstack-swift | 21:38 | |
peluse_ | here's the original and the next patch is when we changed it https://review.openstack.org/#/c/39839/26/swift/proxy/server.py | 21:39 |
dfg | well. it seems better to me. but on the other hand I'm only on like the 4 patch of like 50 so what do I know. i guess just wait for what others say etc. | 21:40 |
peluse_ | its very possible that the reason for the lazy loading no longer exists. the policy module itself looks very different today than it did back then | 21:40 |
peluse_ | clayg: want me to mess with it? | 21:41 |
creiht | If there isn't a good reason for the lazy loading, I would prefer to load on start | 21:41 |
peluse_ | OK, I'll snadbox it for a few and see where it goes... | 21:41 |
dfg | clayg: peluse_ : i | 21:42 |
dfg | 'm an idiot. master does blow up if there's no obj server (i didn't actually switch to master before) sorry | 21:42 |
dfg | fuck | 21:42 |
dfg | master does blow up if there's no object.ring.gz | 21:43 |
peluse_ | peluse: sorry, could have mentioned that. That's the main reason why we didn't do anything about the fact that things blow up if a policy ring is not there | 21:43 |
dfg | so anyway- i like the "for policy in POLICIES: policy.load_ring(swift_dir) in proxy's __init__" idea. | 21:44 |
clayg | cool - i've already got it hacked up - i'll see what needs to happen to get some tests around it - you'll see in tonights changes | 21:44 |
clayg | dfg: thanks! | 21:44 |
clayg | anyone going to bite on the POLICY_HEADER and POLICY_INDEX_HEADER name changes? because of the flake8 word I only want to change 'em once | 21:45 |
notmyname | creiht: https://wiki.openstack.org/wiki/Swift/PriorityReviews | 21:46 |
creiht | notmyname: awesome, thx | 21:47 |
*** openstackgerrit has quit IRC | 21:47 | |
peluse_ | clayg: rock on | 21:47 |
*** ujjain has quit IRC | 21:50 | |
*** Thurloat has joined #openstack-swift | 21:52 | |
*** jamie_h has quit IRC | 21:53 | |
creiht | where is storage policy type documented? | 21:54 |
peluse_ | clayg: FYI I had already hacked in the 2 lines you mention and it exposes a few errors in test code that defines policies but not corresponding rings (because they weren't loaded on init before). I'll leave it alone unless you want me to tackle the 3 errors I see anyway | 21:54 |
clayg | creiht: trying to get rid of that - did you find a refrence to it in the latest change - or maybe you had it in an old config? | 21:55 |
*** ujjain has joined #openstack-swift | 21:55 | |
peluse_ | clayg: its still in the docs until you take the stuff I sent you (type that is) | 21:55 |
clayg | peluse_: i'll find 'em no worries | 21:55 |
*** tusharsg has quit IRC | 21:55 | |
peluse_ | cool, let me know if you want help | 21:55 |
clayg | peluse_: i pulled it out | 21:55 |
clayg | peluse_: or rather i guess i *tried* to pull it out | 21:55 |
peluse_ | :) | 21:55 |
clayg | oh, unless maybe that wasn't in the last push? | 21:56 |
clayg | yeah i guess I did that on the train | 21:56 |
creiht | clayg: ahh cool | 21:56 |
clayg | creiht: types are going away ;) | 21:56 |
clayg | creiht: it was pre-mature | 21:56 |
creiht | clayg: https://review.openstack.org/#/c/96027/3/etc/swift.conf-sample | 21:56 |
peluse_ | you pulled it out pre-maturely, is that what you're saying? | 21:56 |
peluse_ | and on a train no less :) | 21:57 |
clayg | peluse_: no... i mis-spoke is all, the changes in gerrit still have types - but tomorrow they won't | 21:57 |
creiht | k | 21:57 |
creiht | I'll ignore it then | 21:57 |
clayg | reviews indicated that they either need to not be there or they need to be a bunch different | 21:57 |
clayg | we choose they don't need to be there (or notmyname did anyway and I was happy with that) | 21:57 |
creiht | yeah that makes sense to me | 21:58 |
peluse_ | creiht: FYI we'll need something to address what they were there for when we add EC though... that's why it was there (but really as a stub for now) thus no real need at the moment | 21:58 |
clayg | peluse_: i think even that is TBD, well need *something* - it may not look anything like a static list of strings that was VALID_TYPES before, and it may not be exposed in /info directly - but since we'll need to argue about it - they're gone for now! | 22:00 |
*** jergerber has quit IRC | 22:02 | |
*** byeager has quit IRC | 22:06 | |
peluse_ | nope, I'm with ya, that why I said "something" and not "types".... | 22:09 |
*** swat30 has joined #openstack-swift | 22:09 | |
*** byeager has joined #openstack-swift | 22:09 | |
*** shakamunyi has joined #openstack-swift | 22:11 | |
mattoliverau | Morning | 22:11 |
bgmccollum | i guess i missed this, but why was `swift-object-expirer` separately packaged in Ubuntu? Is the idea that it could be run external to an object server, so it needed to be packaged separately? | 22:16 |
notmyname | bgmccollum: it's a separate daemon. I can't tell you why it's separately packaged, but it is independent of the object server | 22:17 |
bgmccollum | right...but it used to be installed with just object-server... | 22:17 |
bgmccollum | *shrug* | 22:17 |
*** zhiyan_ is now known as zhiyan | 22:18 | |
notmyname | bgmccollum: you should take that up with your cannonical rep ;-) | 22:18 |
bgmccollum | its easy enough to just add to the list of packages to install...i guess i just missed the heads up wherever that would have been. | 22:19 |
notmyname | bgmccollum: normally you should tell people to run the expirer process on an admin server in the cluster. and you can add more if you have a lot of expiring obejcts. but it's not 1:1 with the object servers | 22:19 |
bgmccollum | notmyname: which probably explains the separate package. thx. | 22:20 |
creiht | should we make the policy names non-case sensitive? | 22:20 |
clayg | creiht: yeah, the names are used in header keys sometimes (e.g. account response) | 22:21 |
clayg | creiht: can you ding something somewhere? | 22:21 |
clayg | creiht: good one | 22:21 |
creiht | clayg: will do | 22:21 |
*** shakayumi has joined #openstack-swift | 22:21 | |
*** shakamunyi has quit IRC | 22:22 | |
*** kashyapk has joined #openstack-swift | 22:23 | |
*** zhiyan is now known as zhiyan_ | 22:27 | |
*** kashyapk has quit IRC | 22:28 | |
creiht | so if a storage policy is marked as deprecated, users can read objects, but they can't put any more objects with that policy? | 22:35 |
*** Midnightmyth has quit IRC | 22:36 | |
clayg | creiht: the reconciler.conf stuff should be there later with the reconciler change - i wasn't really sure how to handle docs for that - the list endpoint stuff is in the docs even though that doesn't show up till later - does it matter if i move it or not? | 22:41 |
creiht | clayg: doesn't matter to me as long as it is in the pipe | 22:43 |
clayg | yeah it is | 22:44 |
creiht | cool | 22:44 |
peluse_ | creiht: a container can contiue to be used you just can't create new continaers | 22:44 |
creiht | peluse_: ahh ok | 22:44 |
mattoliverau | creiht: the way i read the code was when a policy is deprecated, new containers couldn't be created with it, but the existing one remained, so those uses could still add to the etc. However the deprecated policy wont show up in /info. | 22:45 |
mattoliverau | or simply what peluse_ said :P | 22:45 |
creiht | hehe | 22:45 |
creiht | yeah I realized I was misreading it | 22:45 |
creiht | and probably a good time for me to call it a night :) | 22:45 |
peluse_ | yup, I just added all that to the docs too. Well, I sent to clayg and he'll add it to the chain at the next opportunity | 22:45 |
mattoliverau | lol, some of us are only just waking up :P | 22:45 |
creiht | hehe | 22:46 |
* creiht hands off the baton :) | 22:46 | |
peluse_ | yup to time to call it a night but to what mattoliverau said :) | 22:46 |
peluse_ | sheesh, I think you know what I mean. I can't seem to get the words out correctly | 22:46 |
creiht | hehe | 22:47 |
mattoliverau | Lol, well you both better sleep then :) For us aussies, Friday morning has started, and the weekend isnt too far away! | 22:47 |
peluse_ | nice! | 22:52 |
mattoliverau | Speaking of which, I think its coffee time so I actaully wake up. brb | 22:52 |
*** byeager has quit IRC | 22:55 | |
*** Edward-Zhang has joined #openstack-swift | 22:57 | |
*** NM has quit IRC | 23:03 | |
*** GoJkOrS has quit IRC | 23:14 | |
*** sungju has joined #openstack-swift | 23:19 | |
*** zhiyan_ is now known as zhiyan | 23:19 | |
*** mkollaro has quit IRC | 23:28 | |
*** zhiyan is now known as zhiyan_ | 23:28 | |
*** shri has joined #openstack-swift | 23:40 | |
*** tdasilva has left #openstack-swift | 23:42 | |
*** elambert has quit IRC | 23:47 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!