| *** janonymous has joined #openstack-swift | 00:00 | |
| notmyname | I've said it for a long time: so many problems would go away if we just prevented people from deleting data. ie "def DELETE(self, req): return HttpMethodNotImplemented(req)" | 00:04 |
|---|---|---|
| mattoliverau | notmyname: totally agree, deleting sharding containers is something I haven't quite come to terms with, and that would make it _so_ much easier ;P | 00:10 |
| notmyname | I mean, really?! who needs to delete data? this is a *storage* system. not a *not storage* system. if you want to delete your data, just store it on a ram-disk | 00:11 |
| notmyname | ;-) | 00:11 |
| mattoliverau | lol | 00:13 |
| zaitcev | I used to share an office with a senior engineer at Sun who worked on redundant arrays of ramdisks | 00:16 |
| zaitcev | Because, he said, "our customers don't lose power, and if they do, they have datacenters on all continents" | 00:16 |
| zaitcev | It was before the SSD revolution, so he may be forgiven, I think, for trying to find the benefits that we take for granted today. | 00:17 |
| *** tqtran has quit IRC | 00:25 | |
| *** catintheroof has quit IRC | 00:25 | |
| *** clu_ has quit IRC | 00:30 | |
| clayg | tdasilva: do you have the change that fixes GET to do better with 404 timestamps | 00:31 |
| clayg | tdasilva: nm, found it - and added a refernce to lp bug #1503161 - since it seems related | 00:33 |
| openstack | Launchpad bug 1503161 in OpenStack Object Storage (swift) "[Re-open in 2015 Oct] DELETE operation not write affinity aware" [Undecided,New] https://launchpad.net/bugs/1503161 | 00:33 |
| *** tqtran has joined #openstack-swift | 00:43 | |
| *** mingyu has joined #openstack-swift | 00:50 | |
| *** nikivi has quit IRC | 00:54 | |
| *** mingyu has quit IRC | 00:55 | |
| *** diogogmt has quit IRC | 00:55 | |
| *** sams-gleb has joined #openstack-swift | 00:58 | |
| kota_ | good morning | 01:00 |
| *** rcernin has quit IRC | 01:01 | |
| timburke | clayg: related to that bug... when do we send the 404 if the *container* doesn't exist? when get_container_info fails on the proxy? what happens when the object server accepts a write, but the container's been deleted by the time the updater runs? | 01:01 |
| *** sams-gleb has quit IRC | 01:03 | |
| clayg | timburke: container db's will accept object updates for deleted objects - if it's a tombstone ... that's fine of course | 01:14 |
| clayg | timburke: I *do* think somewhere the proxy might reject a PUT if a contaier "doesn't exist" based on get_container_info - not sure if that effects delete | 01:15 |
| clayg | timburke: i'd ahve to trace the code... waht are you thinking? | 01:15 |
| timburke | clayg: *somewhere* we return a 404 if you try to delete an object in a container that doesn't exist, and that needs to be treated differently than getting back a 404 because the object didn't exist, but the container does | 01:18 |
| *** a1|away has quit IRC | 01:18 | |
| *** a1|away has joined #openstack-swift | 01:18 | |
| kota_ | clayg: could you have a time to talk about patch 405450? | 01:19 |
| patchbot | https://review.openstack.org/#/c/405450/ - swift - Fixups for ghost listing fix | 01:19 |
| timburke | clayg: as far as i'm concerned, 204 + 404 where container actually exists + 503 = 204, timestamps be damned | 01:19 |
| timburke | whereas 204 + 404 where *container* doesn't exist + 503 depends on the timestamps | 01:19 |
| clayg | kota_: sure | 01:20 |
| clayg | timburke: I think a 204 that is tombstoneing a t1.data and 404ing a t2.tombstone can return 404 - unless the responses indicate some other t3.data is also being tombstoned? | 01:21 |
| timburke | but if the proxy's already decided "container exists" before sending object-server requests, that simplifies things considerably | 01:21 |
| clayg | timburke: yeah, IIRC in PUT that happens near the top of the PUT method in proxy.controller.obj | 01:21 |
| kota_ | clayg: exactly, the updater change for error line could make a huge lines and updater will continue to dump it for all requests. | 01:21 |
| kota_ | hmm... | 01:21 |
| kota_ | not conclude my thought yet. | 01:22 |
| clayg | kota_: I think debug is fine for both - we didn't have an issue like "this was broken and we didn't know it" | 01:22 |
| kota_ | OTOH, object-server handles it as an error line, https://github.com/openstack/swift/blob/master/swift/obj/server.py#L251-L256, yes, it's only once when the request incomming. | 01:22 |
| timburke | clayg: i think mainly i'm irked that https://review.openstack.org/#/c/368264/ had to be a thing and i'm looking for ways to sneak that behavior into swift :P | 01:23 |
| patchbot | patch 368264 - swift3 - Fix the deletion of non-existent keys | 01:23 |
| kota_ | clayg: since I reviewed that, I've been realizing why we don't make self.logger.increment('async_pending') if we keep the pending file when the request failed? | 01:24 |
| clayg | timburke: ok, and so currently swift3 just alwasy returns 204? | 01:24 |
| kota_ | so it might be helpful rather than log line 'we hvae something failed and we keep the async pending for that' | 01:24 |
| kota_ | the failure seems to be incremented thought. | 01:24 |
| kota_ | https://github.com/openstack/swift/blob/master/swift/obj/updater.py#L247 | 01:25 |
| clayg | timburke: maybe just add a new header to mark the request as "404'd because no container" to all of the places we do that like https://github.com/openstack/swift/blob/5084a63770673c588b3fe832a9c9e52be7d82173/swift/proxy/controllers/obj.py#L713 | 01:26 |
| timburke | clayg: that's not *merged*; currently we 404 like swift. but it's just bit multiple customers in different contexts | 01:26 |
| timburke | like https://bugs.launchpad.net/swift3/+bug/1638512 | 01:27 |
| openstack | Launchpad bug 1638512 in Swift3 "Swift3 is not write affinity aware, that causes multipart upload failed in `404 (NoSuchKey)`" [Undecided,New] | 01:27 |
| clayg | timburke: are you trying to get me to review that s3 change? I read the commit message and I can see how it would be difficult to translate that behavior to the swift api if it's important | 01:27 |
| timburke | clayg: not really. just pointing out that users may have a different mental model for what those status codes mean -- and i'm partial toward the "if we write a tombstone, that's successful" model | 01:29 |
| clayg | timburke: the failed upload bug looks terrible - not obvious why changing the response code of DELETE requests for objects would effect the behavior when creating a multi-part object via swift3 | 01:29 |
| clayg | timburke: the statement "I'm partial toward [an api that behaves different than swift's does]" - it's not an immediately actionable statement to me | 01:31 |
| clayg | kota_: one compromise that I liked was to make L247 be a warning and make the new one (the one at error) a debug | 01:31 |
| *** Guest36 has joined #openstack-swift | 01:32 | |
| clayg | kota_: I don't think it's *quite* as good as both being debug - but I think the new one being error is actively bad - so i'm mostly concerened with that being changed | 01:32 |
| timburke | clayg: we persist an upload identifier that gets cleaned up upon completion. i could just as easily have the handling around *just that* tolerate 404s explicitly, but my users expect it *all* the time, so... | 01:32 |
| timburke | (honestly, it was a bit of a drive-by; i didn't even realize i was fixing that case) | 01:32 |
| timburke | clayg: how's this for actionable: "swift should always 204 when writing a .ts" | 01:33 |
| clayg | timburke: ok so I answered your question by showing you where the object controller currently deals with the container missing case - if you file a bug that says and enduser (or middleware) should be able to differentiate I think adding a header is the most obvious backwards compatible path forward | 01:33 |
| timburke | only that'll never merge, because someone somewhere might be depending the old behavior | 01:34 |
| clayg | timburke: i'm not sure if you mean should abstractly - like if we had a time machine - or the API should change - because i don't know how to make that backwards compatible | 01:34 |
| *** m_kazuhiro has joined #openstack-swift | 01:34 | |
| clayg | timburke: ok - so abstractly - coolio - yeah fucking swift api sucks! | 01:34 |
| timburke | it's not actually that hard to differentiate in middleware -- peak at the env cache, see whether the container exists | 01:34 |
| clayg | timburke: good thinking! less sure how robust that will be to future maintance as oppsed to a header - but WFM if it works for you! | 01:35 |
| clayg | also obviously no one is trying to keep the information from the user - I think an X-Container-Does-Not-Exist header or somethign would be fine | 01:35 |
| clayg | i remember the first time I tried to PUT and got a 404 I was confused - not sure if a header would have helped | 01:36 |
| timburke | object-server.conf config value? dont_404_deletes_dumb_dumb = True, only it defaults to False? | 01:36 |
| clayg | timburke: no | 01:36 |
| timburke | :P | 01:36 |
| clayg | timburke: we should not encourage deployments to have non-conforming api's | 01:36 |
| kota_ | clayg: k, let me think, still thinking better idea | 01:38 |
| timburke | clayg: so... golang proxy-server! middleware be damned! | 01:41 |
| clayg | timburke: i... think... you saying something current in-tree official middleware that is "optional" being... hypocritical to my earlier statement? | 01:43 |
| clayg | timburke: i feel like we've got off track from how swift should handle DELETE of tombstone :D | 01:43 |
| notmyname | clayg: timburke: are we talking about DELETEs and 404 vs 204? | 01:45 |
| timburke | notmyname: i've gotten us a bit... sidetracked :P | 01:45 |
| clayg | notmyname: yes current thinking is to rewrite the proxy in golang and not allow any custom middleware | 01:45 |
| notmyname | lol | 01:45 |
| timburke | totally solves the DELETE problem | 01:46 |
| notmyname | I kinda like the always 204 option. if it writes a .ts, make it return 204 (but being smarter with the returned timestamps is always an option) | 01:46 |
| notmyname | timburke: so would disabline DELETE altogether | 01:46 |
| notmyname | disabline? | 01:46 |
| clayg | it's like the southern bell version of disabling | 01:47 |
| clayg | dis-ab-line | 01:47 |
| notmyname | yeah, that's how in went in my head | 01:47 |
| notmyname | anyway, "return 204 if .ts is written" seems reasonable. and not a breaking change to the api | 01:48 |
| clayg | my heaves i don't know why would eAvear want to go dis-ab-line DELETE altogether | 01:48 |
| notmyname | family just got home. I'm afk for a bit | 01:50 |
| timburke | notmyname: no, that'd totally break the api -- DELETE to an object that doesn't exist used to 404, then it would 204 | 01:54 |
| *** Guest36 has quit IRC | 01:54 | |
| kota_ | oh... wait? | 02:07 |
| kota_ | why object_update returns HTTP_INTERNAL_SEVER_ERROR for the first item in the return tupple in failure case? | 02:07 |
| kota_ | that should be true/false | 02:08 |
| kota_ | :/ | 02:08 |
| clayg | kota_: you can't fix *all* the bugs in one patch! ;) | 02:10 |
| kota_ | clayg: sure | 02:11 |
| clayg | kota_: i'm worried you're thinking too hard on this issue! save your braincells - we have many issues unfortuantely :'( | 02:16 |
| *** klrmn has quit IRC | 02:17 | |
| clayg | kota_: e.g. is there anything I can do to help with isa-l-rs-cauchy? | 02:17 |
| kota_ | clayg: yeah, i can add +2 for that soon but I'm in pazzling we could may remove the debug line, anyway? | 02:17 |
| kota_ | clayg: could you wait a bit? | 02:17 |
| clayg | kota_: of course! | 02:18 |
| *** isotope has joined #openstack-swift | 02:24 | |
| *** isotope has quit IRC | 02:29 | |
| kota_ | clayg: not yet tested just dump my thought into code, https://gist.github.com/bloodeagle40234/2d747f44909d3d3f153a27a70afb7a39 | 02:35 |
| kota_ | ok, i got an error, TypeError: 'set' object does not support indexing | 02:37 |
| kota_ | need to fix a bit | 02:37 |
| kota_ | fixed and url link is same, https://gist.github.com/bloodeagle40234/2d747f44909d3d3f153a27a70afb7a39 | 02:39 |
| kota_ | er, maybe we can keep debug log line in the failure case | 02:41 |
| kota_ | but, I need to think more for the case the udater process in the second time (with attempts to less than whole replicas) | 02:42 |
| *** winggundamth has quit IRC | 02:48 | |
| *** winggundamth has joined #openstack-swift | 02:49 | |
| *** bkopilov has quit IRC | 03:13 | |
| *** dmorita has quit IRC | 03:16 | |
| *** briancline has joined #openstack-swift | 03:29 | |
| *** klrmn has joined #openstack-swift | 03:31 | |
| openstackgerrit | Tim Burke proposed openstack/pyeclib: Fix checksum memory leak https://review.openstack.org/407302 | 03:48 |
| *** links has joined #openstack-swift | 03:52 | |
| *** arch-nemesis has quit IRC | 03:53 | |
| notmyname | timburke: no, currently you DELETE and get a 404 sometimes (depending on things not in your control)--but yes for unknown objects you get a 404--yet we always actually do something for you. I'm proposing that telling the client that we wrote the tombstone isn't a horrible idea | 03:56 |
| notmyname | no, I'm not arguing that the result on a delete to a 404 isn't different. just that returning a 204 is more consistent and more accurate. and that since currently we require that clients handle both 204 and 404 for reasons beyond their understanding--which we've seen both with customers and the community--simply returning a 204 is reasonable and very low risk | 04:02 |
| *** tqtran has quit IRC | 04:03 | |
| *** psachin has joined #openstack-swift | 04:03 | |
| *** m_kazuhiro_ has joined #openstack-swift | 04:07 | |
| *** m_kazuhiro has quit IRC | 04:10 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 04:37 |
| *** bkopilov has joined #openstack-swift | 04:41 | |
| *** mmuffinman has joined #openstack-swift | 04:50 | |
| *** klrmn has quit IRC | 04:55 | |
| *** dmorita has joined #openstack-swift | 05:16 | |
| *** dmorita has quit IRC | 05:21 | |
| *** mingyu has joined #openstack-swift | 05:21 | |
| *** mingyu has quit IRC | 05:26 | |
| *** sanchitmalhotra has joined #openstack-swift | 05:31 | |
| *** sanchitmalhotra1 has joined #openstack-swift | 05:43 | |
| *** sanchitmalhotra has quit IRC | 05:44 | |
| *** sanchitmalhotra1 is now known as sanchitmalhotra | 05:44 | |
| *** ppai has joined #openstack-swift | 05:51 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/liberasurecode: ISA-L Cauchy support https://review.openstack.org/393263 | 05:59 |
| *** tqtran has joined #openstack-swift | 06:03 | |
| *** tqtran has quit IRC | 06:07 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/pyeclib: DON'T MERGE: just checking if the test failed https://review.openstack.org/407333 | 06:09 |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/pyeclib: Fix checksum memory leak https://review.openstack.org/407302 | 06:12 |
| *** tovin07 has joined #openstack-swift | 06:25 | |
| *** ppai has quit IRC | 06:31 | |
| *** sams-gleb has joined #openstack-swift | 06:33 | |
| *** sams-gleb has quit IRC | 06:33 | |
| *** sams-gleb has joined #openstack-swift | 06:34 | |
| *** rcernin has joined #openstack-swift | 06:34 | |
| *** sams-gleb has quit IRC | 06:38 | |
| *** sanchitmalhotra has quit IRC | 06:39 | |
| *** ppai has joined #openstack-swift | 06:48 | |
| *** SkyRocknRoll has joined #openstack-swift | 06:53 | |
| *** SkyRocknRoll has quit IRC | 06:58 | |
| *** ChubYann has quit IRC | 07:06 | |
| *** m_kazuhiro_ has quit IRC | 07:09 | |
| *** SkyRocknRoll has joined #openstack-swift | 07:10 | |
| *** ppai has quit IRC | 07:11 | |
| *** rcernin has quit IRC | 07:12 | |
| *** ppai has joined #openstack-swift | 07:24 | |
| *** bikmak has quit IRC | 07:25 | |
| *** Guest66666 has quit IRC | 07:27 | |
| *** Guest66666 has joined #openstack-swift | 07:29 | |
| *** rcernin has joined #openstack-swift | 07:34 | |
| *** Jeffrey4l has quit IRC | 07:36 | |
| *** ppai has quit IRC | 07:39 | |
| *** hseipp has joined #openstack-swift | 07:41 | |
| *** pcaruana has joined #openstack-swift | 07:42 | |
| *** Jeffrey4l has joined #openstack-swift | 07:52 | |
| *** d0ugal has joined #openstack-swift | 07:57 | |
| *** d0ugal has quit IRC | 07:57 | |
| *** d0ugal has joined #openstack-swift | 07:57 | |
| *** mvk has quit IRC | 08:01 | |
| *** sanchitmalhotra has joined #openstack-swift | 08:16 | |
| *** silor has joined #openstack-swift | 08:26 | |
| *** silor1 has joined #openstack-swift | 08:31 | |
| *** silor has quit IRC | 08:32 | |
| *** silor1 is now known as silor | 08:32 | |
| *** mvk has joined #openstack-swift | 08:33 | |
| *** geaaru has joined #openstack-swift | 08:48 | |
| *** Guest36 has joined #openstack-swift | 08:50 | |
| *** si1ver has quit IRC | 08:57 | |
| *** si1ver has joined #openstack-swift | 08:58 | |
| *** Guest36 has quit IRC | 08:59 | |
| *** Guest36 has joined #openstack-swift | 09:01 | |
| *** dmorita has joined #openstack-swift | 09:01 | |
| *** oshritf has joined #openstack-swift | 09:01 | |
| *** psachin has quit IRC | 09:04 | |
| *** dmorita has quit IRC | 09:06 | |
| *** asettle has joined #openstack-swift | 09:09 | |
| *** psachin has joined #openstack-swift | 09:17 | |
| *** hseipp has quit IRC | 09:21 | |
| *** hseipp has joined #openstack-swift | 09:28 | |
| *** daemontool has joined #openstack-swift | 09:35 | |
| *** tqtran has joined #openstack-swift | 10:06 | |
| *** ppai has joined #openstack-swift | 10:07 | |
| *** abalfour has quit IRC | 10:08 | |
| *** abalfour has joined #openstack-swift | 10:08 | |
| *** Trpger has joined #openstack-swift | 10:09 | |
| *** tovin07 has quit IRC | 10:09 | |
| *** tqtran has quit IRC | 10:10 | |
| *** Trpger has left #openstack-swift | 10:12 | |
| *** daemontool has quit IRC | 10:23 | |
| *** SkyRocknRoll has quit IRC | 10:48 | |
| *** Guest36 has quit IRC | 10:51 | |
| *** dfflanders has quit IRC | 10:53 | |
| openstackgerrit | Sachin Patil proposed openstack/swift: get_part_nodes should raise error on invalid part https://review.openstack.org/402522 | 11:04 |
| *** daemontool has joined #openstack-swift | 11:11 | |
| openstackgerrit | Gábor Antal proposed openstack/swift: Use more specific asserts in test/unit/obj tests https://review.openstack.org/342830 | 11:32 |
| *** vint_bra has joined #openstack-swift | 11:41 | |
| *** jamielennox is now known as jamielennox|away | 11:44 | |
| *** cdelatte has joined #openstack-swift | 11:46 | |
| *** ppai has quit IRC | 12:00 | |
| *** ppai has joined #openstack-swift | 12:09 | |
| *** tingo has joined #openstack-swift | 12:13 | |
| *** tingo has quit IRC | 12:17 | |
| *** acoles_ is now known as acoles | 12:18 | |
| *** kei_yama has quit IRC | 12:19 | |
| *** bkopilov has quit IRC | 12:39 | |
| *** cdelatte has quit IRC | 12:51 | |
| *** cdelatte has joined #openstack-swift | 12:58 | |
| *** xionchen_ has joined #openstack-swift | 13:19 | |
| *** peterlisak has quit IRC | 13:42 | |
| *** onovy has quit IRC | 13:42 | |
| *** _JZ_ has joined #openstack-swift | 13:43 | |
| *** catintheroof has joined #openstack-swift | 13:43 | |
| *** ppai has quit IRC | 13:52 | |
| *** daemontool has quit IRC | 14:04 | |
| *** joeljwright has joined #openstack-swift | 14:11 | |
| *** ChanServ sets mode: +v joeljwright | 14:11 | |
| *** StraubTW has joined #openstack-swift | 14:18 | |
| *** StraubTW has quit IRC | 14:21 | |
| *** StraubTW has joined #openstack-swift | 14:22 | |
| *** links has quit IRC | 14:40 | |
| openstackgerrit | Merged openstack/swift-specs: Show team and repo badges on README https://review.openstack.org/402859 | 14:50 |
| *** peterlisak has joined #openstack-swift | 14:52 | |
| *** onovy has joined #openstack-swift | 14:52 | |
| *** StraubTW has quit IRC | 14:54 | |
| *** psachin has quit IRC | 15:00 | |
| *** cdelatte has quit IRC | 15:00 | |
| *** siva_krish has joined #openstack-swift | 15:00 | |
| *** sams-gleb has joined #openstack-swift | 15:02 | |
| *** cdelatte has joined #openstack-swift | 15:03 | |
| *** StraubTW has joined #openstack-swift | 15:07 | |
| *** tqtran has joined #openstack-swift | 15:10 | |
| *** tqtran has quit IRC | 15:15 | |
| *** tuan_luong has joined #openstack-swift | 15:19 | |
| *** StraubTW has quit IRC | 15:20 | |
| *** xionchen_ has quit IRC | 15:24 | |
| *** StraubTW has joined #openstack-swift | 15:36 | |
| *** isotope has joined #openstack-swift | 15:41 | |
| *** StraubTW has quit IRC | 15:46 | |
| *** jmccarthy has joined #openstack-swift | 15:46 | |
| jmccarthy | Hiya, quick I hope question about storage network and swift, the docs seem to suggest bind_ip = MANAGEMENT_INTERFACE_IP_ADDRESS that I've found so far, how do I enable/configure portions to function in the Storage Network ? | 15:49 |
| jmccarthy | Oh - guess I should have kept reading :) | 15:51 |
| jmccarthy | Set the bind_ip configuration option to STORAGE_LOCAL_NET_IP. | 15:51 |
| jmccarthy | Edit the account-server.conf, container-server.conf and object-server.conf files in the /etc/swift directory. In each file, update the [DEFAULT] section as follows: | 15:51 |
| jmccarthy | [DEFAULT] | 15:51 |
| jmccarthy | bind_ip = STORAGE_LOCAL_NET_IP | 15:51 |
| *** isotope has quit IRC | 15:53 | |
| notmyname | good morning | 16:02 |
| jmccarthy | Morning ! | 16:05 |
| jmccarthy | I was just asking this a minute ago :) Is it really only a case of the bind_ip setting for account-server.conf, container-server.conf and object-server.conf to have swift use storage network 'in the back' ? | 16:06 |
| notmyname | jmccarthy: I'm not 100% clear on your question, but I think the answer is yes | 16:07 |
| jmccarthy | I'm sure I'm not asking it clearly :) Hehe I mean for things that don't *have to be on the public network (like calls to api/mgmt interface) maybe stuff like replicator auditor and internal things to swift, so they can do that stuff on storage network (not public) | 16:09 |
| notmyname | yep | 16:10 |
| jmccarthy | Ok cool I'm going to try that out :) Thanks ! | 16:10 |
| notmyname | the bind_ip in the proxy conf is what you put in your load balancer and your clients talk to. the bind_ip in the account, container, object (ACO) conf should match what's in the respective ring files | 16:11 |
| jmccarthy | Oh whoa ok great point re:ring files - didn't think of that ! | 16:11 |
| notmyname | the proxy uses the ring to find the right IP/port and then talks to that server | 16:11 |
| notmyname | but here's the extra complexity... | 16:11 |
| notmyname | you can add a separate replication ip/port in the ring too | 16:11 |
| notmyname | to start with, you might not need that | 16:12 |
| notmyname | it's there so you can have a separate network for the internal background processes that won't interfere with client-originated requests | 16:12 |
| notmyname | it's normally used by having a separate object server config that's listening on the replication ip/port | 16:13 |
| notmyname | ie you'd have 2 object servers running on a box: one for client traffic and one for replication/background/internal traffic | 16:13 |
| notmyname | make sense? | 16:13 |
| *** catinthe_ has joined #openstack-swift | 16:14 | |
| jmccarthy | Defo that's the sort of thing I'm thinking about - ok so if I make the bind_ip config changes, and update rings to point to ips on that storage network, I'm up to that point :) With only those changes, where would replication happen ? | 16:14 |
| jmccarthy | In the storage network ? | 16:15 |
| *** bkopilov has joined #openstack-swift | 16:16 | |
| notmyname | yes. if you don't designate a separate replication network, both client-originated traffic (ie via the proxy server) and internal-only traffic (eg from replication) will be on the same interface | 16:16 |
| *** daemontool has joined #openstack-swift | 16:16 | |
| *** catintheroof has quit IRC | 16:16 | |
| jmccarthy | Hmm ok I lost it :) | 16:16 |
| notmyname | FWIW, https://www.swiftstack.com/docs/admin/hardware.html#example-configuration has some words and pictures that may help. note that those are docs related to the swiftstack product, not just the open-source swift storage engine (and my employer) | 16:17 |
| jmccarthy | I have proxy server on mgmt net, and with reconfigure, no other bind_ip on that net lets say - the others are on storage net now | 16:17 |
| *** rcernin has quit IRC | 16:17 | |
| jmccarthy | The client traffic is only on mgmt net, and proxy server talks to rest on storage right ? | 16:17 |
| *** catinthe_ has quit IRC | 16:18 | |
| notmyname | yeah, so the proxy server has 2 interfaces, right? one for the external-facing and one for the cluster-facing | 16:19 |
| *** pcaruana has quit IRC | 16:19 | |
| jmccarthy | proxy has 2 yep, so I mean in this case, no api traffic happens on storage net if I'm thinking right, then where does replication happen ? (i.e. I have not setup a third network for this) | 16:20 |
| notmyname | you've got a public and an internet network? then the replication traffic would be on the internal network | 16:20 |
| jmccarthy | I was thinking it would happen on the storage net | 16:21 |
| notmyname | ah! http://learn.swiftstack.com/rs/034-CBF-009/images/Cisco-and-SwiftStack-Reference-Design-Document.pdf has a better picture on page 12 | 16:21 |
| jmccarthy | Ok yes the internal one I'm calling the storage net :) | 16:21 |
| notmyname | right | 16:21 |
| notmyname | :-) | 16:21 |
| *** zul has quit IRC | 16:22 | |
| notmyname | jmccarthy: I need to step away for a few minutes | 16:22 |
| jmccarthy | :) It's cool - that helped a lot | 16:23 |
| jmccarthy | Thanks ! | 16:23 |
| *** oshritf has quit IRC | 16:23 | |
| *** zul has joined #openstack-swift | 16:25 | |
| *** mvk has quit IRC | 16:26 | |
| *** tuan_luong has quit IRC | 16:34 | |
| *** chsc has joined #openstack-swift | 16:36 | |
| *** chsc has quit IRC | 16:36 | |
| *** chsc has joined #openstack-swift | 16:36 | |
| *** chsc has quit IRC | 16:50 | |
| *** vinsh has joined #openstack-swift | 16:54 | |
| * notmyname back | 16:58 | |
| * notmyname back | 17:00 | |
| *** catintheroof has joined #openstack-swift | 17:03 | |
| *** catintheroof has quit IRC | 17:03 | |
| *** catintheroof has joined #openstack-swift | 17:04 | |
| *** tqtran has joined #openstack-swift | 17:12 | |
| *** geaaru has quit IRC | 17:18 | |
| *** hseipp has quit IRC | 17:22 | |
| *** rcernin has joined #openstack-swift | 17:26 | |
| *** diogogmt has joined #openstack-swift | 17:32 | |
| *** chsc has joined #openstack-swift | 17:34 | |
| *** chsc has joined #openstack-swift | 17:34 | |
| *** asettle has quit IRC | 17:42 | |
| *** dmorita has joined #openstack-swift | 17:43 | |
| *** bkopilov has quit IRC | 17:51 | |
| *** chsc has quit IRC | 17:53 | |
| *** cbartz has left #openstack-swift | 17:56 | |
| *** bkopilov has joined #openstack-swift | 17:58 | |
| *** mvk has joined #openstack-swift | 18:00 | |
| *** isotope has joined #openstack-swift | 18:00 | |
| *** arch-nemesis has joined #openstack-swift | 18:00 | |
| openstackgerrit | Thiago da Silva proposed openstack/swift: add object_post_as_copy to saio. https://review.openstack.org/407627 | 18:02 |
| *** siva_krish has quit IRC | 18:04 | |
| *** siva_krish has joined #openstack-swift | 18:06 | |
| *** joeljwright has quit IRC | 18:08 | |
| tdasilva | jrichli: hello, are you around? | 18:13 |
| *** acoles is now known as acoles_ | 18:16 | |
| *** dmorita has quit IRC | 18:30 | |
| *** dmorita has joined #openstack-swift | 18:31 | |
| *** asettle has joined #openstack-swift | 18:35 | |
| jmccarthy | Oh is this current ? | 18:37 |
| jmccarthy | Note | 18:37 |
| jmccarthy | Syntax of adding device has been changed: R<ip_replication>:<port_replication> was added between z<zone>-<ip>:<port> and /<device_name>_<meta> <weight>. Added devices will use <ip_replication> and <port_replication> for replication activities. | 18:37 |
| jmccarthy | Like: | 18:37 |
| jmccarthy | -ring-builder object.builder add z1-127.0.0.1:6010R127.0.0.1:6050/sdb1 1 | 18:37 |
| jmccarthy | (from http://docs.openstack.org/developer/swift/replication_network.html) | 18:37 |
| jrichli | tdasilva: hello | 18:49 |
| jrichli | tdasilva: I am consuming mass quantities. We are having something we call our "progressive lunch" day at work. There is food at each floor of the building. | 18:51 |
| tdasilva | jrichli: hehe, no worries, after pinging you I imagined you would be at lunch | 18:51 |
| tdasilva | jrichli: was just pinging about symlinks, left a note for you at the patch | 18:51 |
| jrichli | great, thanks | 18:51 |
| jrichli | tdasilva: are we wanting to allow a symlink's target header to be updated? i guess not, right? | 18:53 |
| tdasilva | correct, at some point (i think in san antonio) we decide to allow users to only create symlinks with a PUT request | 18:54 |
| jrichli | ah, ok | 18:54 |
| *** asettle has quit IRC | 18:54 | |
| tdasilva | jrichli: that's at least how I remember, but I couldn't find a place where we wrote that down :( | 18:54 |
| *** asettle has joined #openstack-swift | 18:55 | |
| jrichli | tdasilva: the suggestion you mentioned sounds good. I am not thinking really deeply about symlinks at the moment, though, so I will mull it over later tonight :-) | 18:55 |
| tdasilva | jrichli: cool, no worries, thanks! :) | 18:56 |
| *** asettle has quit IRC | 18:59 | |
| *** dmorita has quit IRC | 19:00 | |
| *** klrmn has joined #openstack-swift | 19:01 | |
| *** dmorita has joined #openstack-swift | 19:02 | |
| *** arch-nemesis has quit IRC | 19:07 | |
| *** jamielennox|away is now known as jamielennox | 19:07 | |
| *** silor has quit IRC | 19:13 | |
| *** catinthe_ has joined #openstack-swift | 19:14 | |
| *** bapalm_ has quit IRC | 19:14 | |
| *** cnf has quit IRC | 19:15 | |
| *** catintheroof has quit IRC | 19:16 | |
| *** bapalm has joined #openstack-swift | 19:18 | |
| *** cnf has joined #openstack-swift | 19:27 | |
| *** cnf has quit IRC | 19:29 | |
| *** cnfer has joined #openstack-swift | 19:29 | |
| *** jmccarthy has quit IRC | 19:33 | |
| *** mmotiani has quit IRC | 19:33 | |
| *** ChubYann has joined #openstack-swift | 19:33 | |
| *** jmccarthy has joined #openstack-swift | 19:34 | |
| *** cnfer has quit IRC | 19:34 | |
| *** a1|away has quit IRC | 19:35 | |
| *** a1|away has joined #openstack-swift | 19:35 | |
| *** cnf has joined #openstack-swift | 19:35 | |
| *** mmotiani has joined #openstack-swift | 19:36 | |
| *** jmccarthy has left #openstack-swift | 19:38 | |
| *** Jeffrey4l has quit IRC | 19:42 | |
| *** Jeffrey4l has joined #openstack-swift | 19:42 | |
| *** cnf has quit IRC | 19:43 | |
| *** siva_krish has quit IRC | 19:48 | |
| *** cnf has joined #openstack-swift | 19:49 | |
| *** dmorita has quit IRC | 19:58 | |
| *** dmorita has joined #openstack-swift | 19:59 | |
| *** catintheroof has joined #openstack-swift | 20:02 | |
| *** catinthe_ has quit IRC | 20:04 | |
| *** siva_krish has joined #openstack-swift | 20:06 | |
| *** chsc has joined #openstack-swift | 20:07 | |
| *** chsc has joined #openstack-swift | 20:07 | |
| *** isotope has quit IRC | 20:11 | |
| *** isotope has joined #openstack-swift | 20:11 | |
| *** asettle has joined #openstack-swift | 20:11 | |
| clayg | ntata: do changes like patch 407627 get picked up by your https://github.com/ntata/swift-setup-scripts project? | 20:13 |
| patchbot | https://review.openstack.org/#/c/407627/ - swift - add object_post_as_copy to saio. | 20:13 |
| *** asettle has quit IRC | 20:14 | |
| clayg | kota_: I agree with everything you said about patch #405450 | 20:35 |
| patchbot | https://review.openstack.org/#/c/405450/ - swift - Fixups for ghost listing fix | 20:35 |
| clayg | thanks! | 20:35 |
| *** nikivi has joined #openstack-swift | 20:36 | |
| *** isotope has quit IRC | 20:38 | |
| clayg | kota_: timburke: tdasilva: what's the status on patch #393263? That'd be pretty good to be able to do isa-l with parity > 4 again right!? Reviews of pyeclib & liberasurecode are *easy and fun* using vagrant-swift-all-in-one's new "reec" sciprt!? | 20:39 |
| patchbot | https://review.openstack.org/#/c/393263/ - liberasurecode - ISA-L Cauchy support | 20:39 |
| *** nikivi has quit IRC | 20:40 | |
| *** nikivi has joined #openstack-swift | 20:40 | |
| *** asettle has joined #openstack-swift | 20:41 | |
| clayg | oh... acctually you *still* have to get isa-l installed some home - I cloned git@github.com:01org/isa-l.git and followed the readme to get it `make install`ed | 20:43 |
| *** daemontool_ has joined #openstack-swift | 20:44 | |
| *** isotope has joined #openstack-swift | 20:45 | |
| *** daemontool has quit IRC | 20:47 | |
| *** diogogmt has quit IRC | 20:48 | |
| *** siva_krish has quit IRC | 20:51 | |
| *** david-lyle has quit IRC | 20:55 | |
| *** karenc_ is now known as karenc | 21:02 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Split up tests for Concurrent Gets https://review.openstack.org/286589 | 21:04 |
| *** catintheroof has quit IRC | 21:06 | |
| *** cdelatte has quit IRC | 21:18 | |
| clayg | easy review of the day that could help a new contrib get some code landed - patch #406012 | 21:29 |
| patchbot | https://review.openstack.org/#/c/406012/ - swift - Fix swift-get-nodes arg parsing for missing ring | 21:29 |
| clayg | ... assuming you agree with me that swift-get-nodes on master sucks compared to this change | 21:29 |
| *** Jeffrey4l has quit IRC | 21:32 | |
| *** Jeffrey4l has joined #openstack-swift | 21:34 | |
| *** nikivi has quit IRC | 21:39 | |
| *** nikivi has joined #openstack-swift | 21:40 | |
| mattoliverau | morning | 21:40 |
| *** arch-nemesis has joined #openstack-swift | 21:42 | |
| *** chlong has joined #openstack-swift | 21:43 | |
| openstackgerrit | Thiago da Silva proposed openstack/swift: Symlink implementation. https://review.openstack.org/232162 | 21:45 |
| *** nikivi has quit IRC | 21:45 | |
| *** daemontool_ has quit IRC | 21:55 | |
| *** daemontool_ has joined #openstack-swift | 21:55 | |
| *** sams-gleb has quit IRC | 22:04 | |
| *** sams-gleb has joined #openstack-swift | 22:04 | |
| *** chlong has quit IRC | 22:06 | |
| *** chlong has joined #openstack-swift | 22:07 | |
| *** sams-gleb has quit IRC | 22:09 | |
| ntata | clayg, yes it does! My saio makes use of the master's copy of doc/saio/swift/proxy-server.conf :) | 22:10 |
| jrichli | ntata: great! clayg: good thought. I will try to keep the scripts in mind going forward | 22:11 |
| *** chlong has quit IRC | 22:12 | |
| tdasilva | ntata: so is that patch bad for your script? sorry, i think i'm missing something... | 22:15 |
| jrichli | tdasilva: I was saying that I want to be sure to think of those scripts. but in this case, I think we are fine since she copies the file you are changing, so the changes will be integrated. | 22:17 |
| jrichli | of course, ntata is the one who knows! I just wanted to be clear on what my statement meant | 22:18 |
| tdasilva | jrichli: cool, hopefully it is not breaking anything | 22:18 |
| *** vint_bra has quit IRC | 22:22 | |
| ntata | jrichli, thank you for helping me out :). tdasilva, you're not disturbing my scripts at all. Even if it does, I'll make sure to tweak them to work around your solution | 22:22 |
| *** david-lyle has joined #openstack-swift | 22:24 | |
| *** daemontool_ has quit IRC | 22:33 | |
| *** daemontool_ has joined #openstack-swift | 22:34 | |
| tdasilva | ntata: :) | 22:35 |
| *** asettle has quit IRC | 22:40 | |
| *** david-lyle has quit IRC | 22:41 | |
| *** isotope has quit IRC | 22:49 | |
| *** asettle has joined #openstack-swift | 23:00 | |
| *** asettle has quit IRC | 23:01 | |
| *** rcernin has quit IRC | 23:02 | |
| *** jamielennox is now known as jamielennox|away | 23:05 | |
| *** jamielennox|away is now known as jamielennox | 23:06 | |
| *** isotope has joined #openstack-swift | 23:07 | |
| *** isotope has quit IRC | 23:16 | |
| openstackgerrit | Merged openstack/swift: add object_post_as_copy to saio. https://review.openstack.org/407627 | 23:22 |
| notmyname | hello, world | 23:25 |
| *** chsc has quit IRC | 23:27 | |
| mattoliverau | notmyname: o/ | 23:28 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 23:35 |
| *** kei_yama has joined #openstack-swift | 23:36 | |
| *** vinsh has quit IRC | 23:46 | |
| *** catintheroof has joined #openstack-swift | 23:58 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!