| ho | code is here: https://github.com/andrewgaul/object-store-comparison | 00:00 |
|---|---|---|
| notmyname | here's runabove /info https://gist.github.com/notmyname/4c1ff71bf7129bf29c55 | 00:00 |
| notmyname | ho: I know him! | 00:00 |
| ho | notmyname: cool! | 00:01 |
| notmyname | actually clayg and I had coffee with him a moth or so ago | 00:01 |
| clayg | yeah those guys are cool - SF is such a weird place | 00:02 |
| clayg | saw this guys thing on HN this morning -> https://github.com/begriffs/postgrest#performance | 00:03 |
| clayg | like the joke about slow dynamic lanuages - i should totally look him up - but won't | 00:03 |
| clayg | who writes real thing in Haskell!? | 00:04 |
| ho | yeah, SF is best place for software engineer. i'm jealous. | 00:05 |
| *** bill_az has joined #openstack-swift | 00:06 | |
| notmyname | wow. I have a voucher code for OVH. it has a zero, and both capitol and lower letter "oh". isn't that exactly what you aren't supposed to do? | 00:06 |
| occupant | one time I heard two canadians on a bus talking about their decision to move to SF. One of 'em said "I mean, what would I have done back home? Work for Blackberry?" and then they just laughed for like 30 seconds | 00:10 |
| notmyname | lol | 00:10 |
| notmyname | the other day on the bus/train in SF I met a guy who work at ta company who's product is built on swift. pretty cool | 00:12 |
| notmyname | (https://www.bigcommerce.com) | 00:12 |
| occupant | where you work? you one of the actual rackspacers? | 00:13 |
| notmyname | "actual" rackspacers? | 00:15 |
| clayg | notmyname: you're not an *actual* racker - you're an ex-racker - big diference | 00:15 |
| clayg | dfg_: redbo: ahale: and mattoliverau are *actual* rackspacers | 00:16 |
| occupant | my apartment is like a block away from the office. you guys would covet hard. | 00:16 |
| notmyname | and glange and hurricanerix | 00:16 |
| clayg | and prometheanfire when he's around | 00:16 |
| clayg | notmyname: oh and scotticus ! | 00:16 |
| notmyname | occupant: ah, cool. clayg and me work at swiftstack. we used to be in south part | 00:16 |
| clayg | occupant: but none of them work at 2nd & Folsom | 00:16 |
| clayg | we used to have a... yeah right over there | 00:16 |
| notmyname | now we're next to the montgomery bart station | 00:16 |
| occupant | yup, I'm at 1st and folsom on the same street as that gigantic high rise they're building | 00:17 |
| occupant | I thought they were just going to build another five story condo like everything else on the block. I was really surprised when it kept going up and up. | 00:17 |
| clayg | we're all snobby in the FiDi now | 00:17 |
| notmyname | clayg: at half the rent of SoMA | 00:18 |
| notmyname | well, not *half*, but significantly less | 00:18 |
| clayg | you know how like twice of nothing is still nothing - rent in SF works the same way but on the other end - half of infinity... | 00:19 |
| occupant | my roomies and I are in a 3br rent control for $3300, which is pretty amazing for the location. | 00:23 |
| *** mahatic_ has joined #openstack-swift | 00:27 | |
| notmyname | time for me to head home | 00:27 |
| *** mahatic_ has quit IRC | 00:28 | |
| *** nadeem has quit IRC | 00:30 | |
| *** ndk has quit IRC | 00:32 | |
| *** gyee has quit IRC | 00:33 | |
| *** SkyRocknRoll has quit IRC | 00:36 | |
| *** ndk has joined #openstack-swift | 00:38 | |
| *** jlhinson has quit IRC | 00:42 | |
| *** jasondotstar has quit IRC | 00:47 | |
| *** SkyRocknRoll has joined #openstack-swift | 00:48 | |
| *** SkyRocknRoll has joined #openstack-swift | 00:48 | |
| *** jasondotstar has joined #openstack-swift | 00:49 | |
| *** DericHorn-HP has joined #openstack-swift | 00:49 | |
| *** chlong has quit IRC | 00:52 | |
| *** jasondotstar has quit IRC | 00:54 | |
| clayg | notmyname: idk, I don't think I'm happy with the async update patch yet - i'm doing something wrong :'( | 00:56 |
| clayg | notmyname: on my master test with some misbehaving standins for a couple of priamry containers I see put requests all over the place 0.5 - 3.5 - but on the async patch they're all like 1.7? | 00:57 |
| clayg | maybe i'm tired/distracted | 00:57 |
| clayg | i'm going to break for dinner and try again later | 00:58 |
| *** ho_ has joined #openstack-swift | 01:11 | |
| *** kota_ has quit IRC | 01:12 | |
| *** haomaiwang has quit IRC | 01:12 | |
| *** ho has quit IRC | 01:12 | |
| clayg | phew ok - there finally got it | 01:44 |
| *** wbhuber has joined #openstack-swift | 01:49 | |
| *** EmilienM|off is now known as EmilienM | 01:52 | |
| *** ho has joined #openstack-swift | 01:52 | |
| *** ho_ has quit IRC | 01:53 | |
| *** ho has quit IRC | 01:54 | |
| *** Nyny has quit IRC | 01:55 | |
| *** ho has joined #openstack-swift | 01:57 | |
| *** thurloat is now known as thurloat_isgone | 01:58 | |
| *** shangxdy has joined #openstack-swift | 02:00 | |
| *** wbhuber has quit IRC | 02:03 | |
| *** chlong has joined #openstack-swift | 02:11 | |
| *** kota_ has joined #openstack-swift | 02:11 | |
| *** ChanServ sets mode: +v kota_ | 02:11 | |
| *** DericHorn-HP has quit IRC | 02:17 | |
| *** janonymous has quit IRC | 02:19 | |
| *** annegentle has quit IRC | 02:28 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Cleanup nits from container_update_timeout https://review.openstack.org/204813 | 02:32 |
| clayg | nice work notmyname! | 02:33 |
| *** annegentle has joined #openstack-swift | 02:33 | |
| clayg | kota_: so why in patch 201055 can't we move the fix_response_headers to not try and "fix" the error responses? | 02:35 |
| patchbot | clayg: https://review.openstack.org/#/c/201055/ | 02:35 |
| *** DericHorn-HP has joined #openstack-swift | 02:38 | |
| openstackgerrit | Azhagu Selvan SP proposed openstack/swift: Respect 'Accept' header in error responses https://review.openstack.org/204196 | 02:55 |
| *** silor has joined #openstack-swift | 02:55 | |
| *** bill_az has quit IRC | 03:09 | |
| *** shangxdy has quit IRC | 03:16 | |
| *** annegentle has quit IRC | 03:17 | |
| *** sanchitmalhotra has joined #openstack-swift | 03:21 | |
| openstackgerrit | Merged openstack/swift: unit tests: Replace "self.assert_" by "self.assertTrue" https://review.openstack.org/204092 | 03:25 |
| openstackgerrit | Merged openstack/swift: Removing commented out code in test/unit/account/test_backend.py https://review.openstack.org/204575 | 03:26 |
| *** DericHorn-HP has quit IRC | 03:40 | |
| jrichli | clayg: all the gory details are there https://etherpad.openstack.org/p/swift_encryption_issues | 03:41 |
| *** jrichli has quit IRC | 03:43 | |
| *** ppai has joined #openstack-swift | 03:47 | |
| *** annegentle has joined #openstack-swift | 03:48 | |
| peluse | clayg, you don't recall and recent bugs and/or fixes for FI of None being passed in to validate_fragment_index() do you (would result in the old 'None: invalid literal for int() with base 10') | 03:52 |
| *** jrichli has joined #openstack-swift | 03:57 | |
| *** jrichli has quit IRC | 04:00 | |
| *** annegentle has quit IRC | 04:09 | |
| *** HenryG has quit IRC | 04:09 | |
| *** mordred has quit IRC | 04:11 | |
| *** HenryG has joined #openstack-swift | 04:12 | |
| *** mutoulbj has joined #openstack-swift | 04:18 | |
| *** mordred has joined #openstack-swift | 04:24 | |
| jith_ | Hi all, i have configured the swift (SAIO) for devstack and used the seperate swift(SAIO) for glance backend.. but in horizon->system information , all the services are getting listed but the ip address and Enabled status of swift is not there.. how to make enable in horizon? Also i am getting error while creating the instance "Failed to allocate the network(s), not rescheduling." | 04:35 |
| *** DericHorn-HP has joined #openstack-swift | 04:39 | |
| mattoliverau | jith_: I'll predicate again with I'm not a devstack expert, but horizon uses the keystone catelog, so you need to make sure you have added the swift endpoint to keystone. | 04:41 |
| *** Kennan has quit IRC | 04:41 | |
| *** Kennan has joined #openstack-swift | 04:42 | |
| jith_ | mattoliverau: That i have done.. thanks... Any idea about this error "Failed to allocate the network(s), not rescheduling." | 04:42 |
| mattoliverau | jith_: sounds like a neutron or nova-network problem (depending on what your using for networking) | 04:44 |
| jith_ | mattoliverau: Thanks... I am using neutron | 04:44 |
| *** DericHorn-HP has quit IRC | 04:53 | |
| clayg | peluse: nothing comes to mind - is it on the branch where the GET path now includes a frag-index? | 04:55 |
| *** zaitcev has quit IRC | 05:03 | |
| clayg | torgomatic: so If-None-Match: "xyzzy", "r2d2xxxx", "c3piozzzz" - is a thing, so I sorta think X-If-Not-Frag-Index will work with a plain old etag based If-Match header | 05:05 |
| clayg | And X-If-Not-Frag-Index just sends down the list of all the fi's we're already got connections for | 05:06 |
| *** bkopilov has quit IRC | 05:11 | |
| *** bkopilov has joined #openstack-swift | 05:13 | |
| *** DericHorn-HP has joined #openstack-swift | 05:25 | |
| openstackgerrit | Azhagu Selvan SP proposed openstack/swift: Respect 'Accept' header in error responses https://review.openstack.org/204196 | 05:32 |
| openstackgerrit | Azhagu Selvan SP proposed openstack/swift: Respect 'Accept' header in error responses https://review.openstack.org/204196 | 05:33 |
| *** DericHorn-HP has quit IRC | 05:41 | |
| *** hrou has quit IRC | 05:50 | |
| *** SkyRocknRoll has quit IRC | 05:51 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: Fix EC range GET/COPY handling https://review.openstack.org/204892 | 05:56 |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: Fix EC range GET/COPY handling https://review.openstack.org/204892 | 06:00 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Imported Translations from Transifex https://review.openstack.org/204914 | 06:11 |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: Fix EC range GET/COPY handling https://review.openstack.org/204892 | 06:11 |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: Fix the missing SLO state on fast-post https://review.openstack.org/182564 | 06:30 |
| *** ppai has quit IRC | 06:31 | |
| *** mahatic has joined #openstack-swift | 06:36 | |
| *** krypto has joined #openstack-swift | 06:36 | |
| krypto | can swift be installed on 2 servers with 2 replicas | 06:40 |
| krypto | will there be any impact on uploads if one of the server goes down | 06:41 |
| *** ppai has joined #openstack-swift | 06:41 | |
| kota_ | krypto: we can make the setting but I guess it is not recommended. | 06:42 |
| krypto | kota_ thanks so will 3 replicas with 3 server provide high data availability | 06:43 |
| kota_ | krypto: the upload impact when one of the servers down depends on # of disks in the server. | 06:44 |
| kota_ | krypto: yes, high durability and availability. | 06:44 |
| krypto | thanks +kota_ | 06:45 |
| kota_ | if each server has more than 1 disk, swift can store the 2 replica when the one of the servers down. | 06:45 |
| kota_ | into unique 2 disks. | 06:45 |
| krypto | i have a server with 2 disks in raid 1 | 06:46 |
| *** ppai has quit IRC | 06:46 | |
| krypto | total 3 servers like this | 06:46 |
| kota_ | well, assuming, if just 1 disk attached to Swift in each server and there are 2 servers and config as 2 replica | 06:47 |
| kota_ | when one of them is shutdown, anyone uploads objects anymore. | 06:48 |
| kota_ | if 3 servers and 2 replica we can upload objects until two of them halted. | 06:49 |
| *** chlong has quit IRC | 06:49 | |
| *** ppai has joined #openstack-swift | 06:50 | |
| mattoliverau | krypto: 2 replica on 2 servers (one drive each) wont work well, if one is down all puts will fail as a qourum for replication is (repl /2) +1.. so 2. | 06:50 |
| mattoliverau | So what kota_said :) | 06:50 |
| kota_ | mattoliverau: nice, additional info :D | 06:51 |
| mattoliverau | swift will handle your durability for you, so better to give it more disks (i.e. jbod). As kota_ said 3x servers if a drive each is better. but even better would be multiple drives per server :) | 06:52 |
| krypto | thanks +kota_ and +mattoliverau for helping me with this much information. I will be using 3 servers with 2 replicas and let me try if i can get more disks :) | 06:54 |
| mattoliverau | so instead of using raid 1 to save your files, you could split it, add them all to swift and up the replica count.. but that obviously leaves your system more vulnrable (if its living on the raid1) | 06:55 |
| mattoliverau | krypto: your welcome | 06:55 |
| *** jasondotstar has joined #openstack-swift | 06:55 | |
| kota_ | krypto: :-) | 06:56 |
| *** SkyRocknRoll has joined #openstack-swift | 07:05 | |
| *** rledisez has joined #openstack-swift | 07:09 | |
| *** proteusguy has quit IRC | 07:14 | |
| *** proteusguy has joined #openstack-swift | 07:27 | |
| *** vinsh_ has joined #openstack-swift | 07:29 | |
| *** vinsh has quit IRC | 07:31 | |
| *** zhill_desktop has quit IRC | 07:31 | |
| *** zhill_desktop has joined #openstack-swift | 07:33 | |
| *** vinsh has joined #openstack-swift | 07:36 | |
| *** vinsh_ has quit IRC | 07:36 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: Fix EC range GET/COPY handling https://review.openstack.org/204892 | 07:58 |
| *** vinsh_ has joined #openstack-swift | 07:58 | |
| *** vinsh has quit IRC | 07:59 | |
| *** vinsh has joined #openstack-swift | 07:59 | |
| *** jistr has joined #openstack-swift | 08:02 | |
| *** mahatic has quit IRC | 08:02 | |
| *** mahatic has joined #openstack-swift | 08:02 | |
| *** vinsh_ has quit IRC | 08:03 | |
| *** jordanP has joined #openstack-swift | 08:07 | |
| *** geaaru has joined #openstack-swift | 08:10 | |
| *** mahatic_ has joined #openstack-swift | 08:11 | |
| *** proteusguy_ has joined #openstack-swift | 08:12 | |
| *** proteusguy_ has quit IRC | 08:12 | |
| *** proteusguy has quit IRC | 08:13 | |
| *** SkyRocknRoll has quit IRC | 08:13 | |
| *** mahatic has quit IRC | 08:13 | |
| *** Kennan has quit IRC | 08:26 | |
| *** SkyRocknRoll has joined #openstack-swift | 08:26 | |
| *** Kennan has joined #openstack-swift | 08:26 | |
| openstackgerrit | Merged openstack/swift: Add two functional tests for delimiter. https://review.openstack.org/202303 | 08:28 |
| *** haomaiwang has joined #openstack-swift | 08:34 | |
| *** proteusguy has joined #openstack-swift | 08:48 | |
| *** sanchitmalhotra1 has joined #openstack-swift | 08:55 | |
| *** sanchitmalhotra has quit IRC | 08:57 | |
| *** aix has joined #openstack-swift | 09:02 | |
| *** kota_ has quit IRC | 09:07 | |
| *** sanchitmalhotra has joined #openstack-swift | 09:08 | |
| *** sanchitmalhotra1 has quit IRC | 09:10 | |
| *** peterbusque has quit IRC | 09:15 | |
| *** marzif_ has joined #openstack-swift | 09:22 | |
| *** marzif_ has quit IRC | 09:23 | |
| *** marzif_ has joined #openstack-swift | 09:24 | |
| *** marzif_ has quit IRC | 09:24 | |
| *** marzif_ has joined #openstack-swift | 09:25 | |
| *** jasondotstar has quit IRC | 09:25 | |
| *** mahatic__ has joined #openstack-swift | 09:26 | |
| *** marzif_ has quit IRC | 09:30 | |
| *** mahatic_ has quit IRC | 09:30 | |
| *** silor has quit IRC | 09:32 | |
| openstackgerrit | Ondrej Novy proposed openstack/swift: Time synchronization check in recon. https://review.openstack.org/192562 | 09:35 |
| *** jasondotstar has joined #openstack-swift | 09:38 | |
| *** sanchitmalhotra1 has joined #openstack-swift | 09:42 | |
| *** jasondotstar has quit IRC | 09:43 | |
| *** sanchitmalhotra has quit IRC | 09:43 | |
| *** sanchitmalhotra has joined #openstack-swift | 09:49 | |
| *** sanchitmalhotra1 has quit IRC | 09:51 | |
| *** sanchitmalhotra1 has joined #openstack-swift | 09:56 | |
| *** sanchitmalhotra has quit IRC | 09:58 | |
| mahatic__ | mattoliverau, hello! did you happen to set up encryption? | 10:00 |
| *** jasondotstar has joined #openstack-swift | 10:31 | |
| *** geaaru has quit IRC | 10:31 | |
| *** silor has joined #openstack-swift | 10:37 | |
| *** pksingh has joined #openstack-swift | 10:37 | |
| *** silor1 has joined #openstack-swift | 10:39 | |
| pksingh | Hi folks, clayg , cschwede , swifterdarrell , torgomatic , could you please review https://review.openstack.org/185061 and https://review.openstack.org/183899, these are there from very long time , thanks in advance :) | 10:40 |
| *** silor has quit IRC | 10:42 | |
| *** geaaru has joined #openstack-swift | 10:44 | |
| *** silor1 has quit IRC | 10:44 | |
| *** joeljwright has joined #openstack-swift | 10:45 | |
| *** ChanServ sets mode: +v joeljwright | 10:45 | |
| *** krypto has quit IRC | 10:56 | |
| *** chlong has joined #openstack-swift | 11:08 | |
| *** silor has joined #openstack-swift | 11:10 | |
| peluse | clayg, wrt the Non fi question - no its something our bencmarking guys ran into in their testing. Getting more info... | 11:11 |
| *** sanchitmalhotra has joined #openstack-swift | 11:17 | |
| *** thurloat_isgone is now known as thurloat | 11:19 | |
| *** sanchitmalhotra1 has quit IRC | 11:19 | |
| *** sanchitmalhotra1 has joined #openstack-swift | 11:22 | |
| *** marzif has joined #openstack-swift | 11:23 | |
| *** sanchitmalhotra has quit IRC | 11:24 | |
| *** aix has quit IRC | 11:27 | |
| *** sanchitmalhotra has joined #openstack-swift | 11:33 | |
| *** robefran has quit IRC | 11:35 | |
| *** robefran_ has joined #openstack-swift | 11:35 | |
| *** sanchitmalhotra1 has quit IRC | 11:35 | |
| *** kutija has quit IRC | 11:35 | |
| *** daemontool_ has joined #openstack-swift | 11:36 | |
| *** daemontool_ has quit IRC | 11:36 | |
| *** daemontool_ has joined #openstack-swift | 11:37 | |
| *** kutija has joined #openstack-swift | 11:38 | |
| *** ppai has quit IRC | 12:00 | |
| *** jasondotstar has quit IRC | 12:11 | |
| *** ppai has joined #openstack-swift | 12:12 | |
| *** robefran_ has quit IRC | 12:13 | |
| *** ho has quit IRC | 12:15 | |
| *** thurloat is now known as thurloat_isgone | 12:16 | |
| *** kei_yama has quit IRC | 12:18 | |
| *** tongli has joined #openstack-swift | 12:18 | |
| *** tongli has quit IRC | 12:22 | |
| *** km has quit IRC | 12:24 | |
| *** aix has joined #openstack-swift | 12:28 | |
| *** tongli has joined #openstack-swift | 12:28 | |
| *** ppai has quit IRC | 12:30 | |
| *** dmorita has quit IRC | 12:33 | |
| *** janonymous has joined #openstack-swift | 12:43 | |
| *** ppai has joined #openstack-swift | 12:43 | |
| janonymous | Hi All, Please find some links for reviews : https://review.openstack.org/#/q/owner:self+status:open+project:openstack/swift,n,z :) | 12:52 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 12:54 |
| *** robefran has joined #openstack-swift | 12:54 | |
| haypo | janonymous: i don't think that your link is right. "owner: self" | 12:59 |
| *** robefran has quit IRC | 13:03 | |
| *** mahatic__ has quit IRC | 13:05 | |
| *** mahatic__ has joined #openstack-swift | 13:06 | |
| janonymous | ohh..ryt.. | 13:13 |
| janonymous | wait | 13:13 |
| *** annegentle has joined #openstack-swift | 13:14 | |
| janonymous | Py3 Link : https://review.openstack.org/#/q/status:open+project:openstack/swift+branch:master+topic:py3,n,z | 13:14 |
| janonymous | My previous link has an additional bandit topic feature as well.. | 13:14 |
| janonymous | haypo: hey | 13:15 |
| janonymous | What's the progress about voting py3 gate ... | 13:16 |
| janonymous | Is there something i could do for that? | 13:16 |
| *** jasondotstar has joined #openstack-swift | 13:30 | |
| *** annegentle has quit IRC | 13:36 | |
| *** hrou has joined #openstack-swift | 13:36 | |
| *** annegentle has joined #openstack-swift | 13:36 | |
| *** SkyRocknRoll has quit IRC | 13:41 | |
| *** ppai has quit IRC | 13:43 | |
| *** robefran has joined #openstack-swift | 13:45 | |
| *** DericHorn-HP has joined #openstack-swift | 13:49 | |
| haypo | janonymous: i updated https://review.openstack.org/#/c/200416/ | 13:52 |
| haypo | janonymous: my python pull request was merged into pyeclib, i'm waiting for a a release | 13:52 |
| haypo | janonymous: we need these two stuff to port swift to python 3 | 13:53 |
| janonymous | Yes. | 13:54 |
| *** bill_az has joined #openstack-swift | 13:55 | |
| *** jrichli has joined #openstack-swift | 13:56 | |
| *** SkyRocknRoll has joined #openstack-swift | 13:57 | |
| *** dgrauet has joined #openstack-swift | 13:58 | |
| *** mahatic__ has quit IRC | 13:58 | |
| *** mahatic__ has joined #openstack-swift | 13:59 | |
| *** wbhuber has joined #openstack-swift | 14:03 | |
| mahatic__ | jrichli: good morning. got a min? | 14:04 |
| jrichli | mahatic__: hello. sure | 14:04 |
| mahatic__ | jrichli: http://paste.openstack.org/show/401160/ - you gave me that yesterday. When I dig into failures when I ran, on a test like this "test.functional.test_object.TestObject" most of them are failing with AssertionError | 14:06 |
| *** lpabon_ has joined #openstack-swift | 14:06 | |
| *** lpabon_ has quit IRC | 14:06 | |
| *** lpabon has quit IRC | 14:07 | |
| mahatic__ | jrichli: but from your paste, only few are failing from that test. I'm not quite sure. I did check that I have the code you pointed to me and the proxy-config right | 14:07 |
| mahatic__ | jrichli: any thoughts on that? | 14:07 |
| jrichli | mahatic__: I suggest picking the first one that fails on your list that is not on mine, and debugging to find out why the assertion fails | 14:08 |
| mahatic__ | sure | 14:08 |
| mahatic__ | jrichli: I wanted to check that I'm not missing any extra code or config options | 14:08 |
| jrichli | mahatic__: I have told you all that I know of. The proxy config that I sent to you is all you need beyond the code from the last patch. So I am afraid that debugging is the only way we can find out what is happening. | 14:10 |
| mahatic__ | alright | 14:10 |
| jrichli | did you start with a clean cluster? | 14:11 |
| jrichli | when is the last time you did a "resetswift"? | 14:11 |
| jrichli | mahatic__ ^^ | 14:12 |
| mahatic__ | jrichli: hey, sorry. yesterday :) | 14:13 |
| *** zul has quit IRC | 14:14 | |
| jrichli | mahatic__: great. So, did you run 'sudo python setup.py develop' and 'sudo python setup.py install'? | 14:15 |
| mahatic__ | jrichli: yup all that done | 14:17 |
| *** jlhinson has joined #openstack-swift | 14:17 | |
| mahatic__ | jrichli: i'm looking at the errors. Will get back if I find something weird | 14:18 |
| jrichli | mahatic__: ok. I am looking at some things too. | 14:20 |
| *** rbrooker has joined #openstack-swift | 14:20 | |
| jrichli | mahatic__: I am looking at the failures in the automated tests in the reviews. But I assume these are different than what you would get. | 14:21 |
| jrichli | mahatic__: for instance, there are problems there because "Unable to read test config /etc/swift/test.conf - file not found" | 14:22 |
| jrichli | mahatic__: be sure to look at the error logs | 14:22 |
| mahatic__ | jrichli: oh okay. Let me cross check | 14:24 |
| *** zul has joined #openstack-swift | 14:27 | |
| jrichli | mahatic__: I also assume you would have different errors than the automated tests because they aren't setup to use encryption middleware yet | 14:28 |
| mahatic__ | jrichli: yeah, I thought so | 14:29 |
| jrichli | mahatic__: Actually, that does bring up a good idea. Maybe you should change back to a normal pipeline first and then run the functests. they should be successful then. | 14:30 |
| jrichli | that should isolate the errors that are not due to encryption middleware | 14:30 |
| mahatic__ | jrichli: I did run the functests when I first set up SAIO and they were successful | 14:31 |
| jrichli | when was that? | 14:31 |
| mahatic__ | which is last week I believe | 14:31 |
| *** briancurtin has quit IRC | 14:31 | |
| *** briancurtin has joined #openstack-swift | 14:31 | |
| jrichli | but you have a new code set now | 14:31 |
| jrichli | so trying the normal pipeline is worth a spin | 14:31 |
| mahatic__ | jrichli: yeah, but what did you mean by normal pipeline? without the encryption stuff? | 14:32 |
| jrichli | yes. just comment out your pipeline, and add back the line without the new middleware. | 14:33 |
| mahatic__ | okay | 14:34 |
| *** lpabon has joined #openstack-swift | 14:36 | |
| *** marzif has quit IRC | 14:43 | |
| *** marzif has joined #openstack-swift | 14:44 | |
| *** annegentle has quit IRC | 14:56 | |
| *** nadeem has joined #openstack-swift | 14:58 | |
| *** nadeem has quit IRC | 15:00 | |
| *** Fin1te has joined #openstack-swift | 15:01 | |
| *** mahatic has joined #openstack-swift | 15:02 | |
| *** mahatic__ has quit IRC | 15:05 | |
| *** minwoob has joined #openstack-swift | 15:09 | |
| *** daemontool_ is now known as marzif_ | 15:10 | |
| *** jasondotstar has quit IRC | 15:12 | |
| *** david-lyle has joined #openstack-swift | 15:17 | |
| *** dgrauet has quit IRC | 15:17 | |
| *** nadeem has joined #openstack-swift | 15:18 | |
| *** annegentle has joined #openstack-swift | 15:18 | |
| *** annegentle has quit IRC | 15:18 | |
| *** annegentle has joined #openstack-swift | 15:19 | |
| *** marzif_ has quit IRC | 15:28 | |
| *** rbrooker has quit IRC | 15:30 | |
| *** mutoulbj has quit IRC | 15:33 | |
| *** mahatic_ has joined #openstack-swift | 15:37 | |
| *** mahatic has quit IRC | 15:37 | |
| *** rbrooker has joined #openstack-swift | 15:38 | |
| *** jasondotstar has joined #openstack-swift | 15:45 | |
| *** DericHorn-HP has quit IRC | 15:52 | |
| *** zhiyan has quit IRC | 15:58 | |
| *** zhiyan has joined #openstack-swift | 15:58 | |
| *** rledisez has quit IRC | 16:00 | |
| *** jistr has quit IRC | 16:01 | |
| *** marzif has quit IRC | 16:02 | |
| *** SkyRocknRoll has quit IRC | 16:04 | |
| peluse | http://newsroom.intel.com/community/intel_newsroom/blog/2015/07/23/intel-announces-cloud-for-all-initiative-to-deliver-benefits-of-the-cloud-to-more-businesses | 16:06 |
| *** jordanP has quit IRC | 16:08 | |
| *** rbrooker has quit IRC | 16:12 | |
| notmyname | interesting | 16:12 |
| *** serverascode has quit IRC | 16:14 | |
| *** marzif has joined #openstack-swift | 16:15 | |
| notmyname | scotticus: are your teams part of this? ^ | 16:16 |
| *** serverascode has joined #openstack-swift | 16:16 | |
| *** jordanP has joined #openstack-swift | 16:17 | |
| openstackgerrit | Carlos Cavanna proposed openstack/swift: Improving statistics sent to Graphite. https://review.openstack.org/202657 | 16:20 |
| *** breitz has quit IRC | 16:28 | |
| *** breitz has joined #openstack-swift | 16:28 | |
| *** nottrobin has quit IRC | 16:29 | |
| *** nottrobin has joined #openstack-swift | 16:29 | |
| redbo | mark | 16:31 |
| redbo | oops I meant /mark | 16:32 |
| openstackgerrit | Christian Schwede proposed openstack/python-swiftclient: Revert "Update mock requirements" https://review.openstack.org/205150 | 16:35 |
| *** aix has quit IRC | 16:37 | |
| *** silor has quit IRC | 16:38 | |
| *** blmartin has joined #openstack-swift | 16:42 | |
| cschwede | timburke: please have a look at that patch if you have a second or two^^ | 16:42 |
| *** Fin1te has quit IRC | 16:46 | |
| *** jordanP has quit IRC | 17:01 | |
| *** jasondotstar has quit IRC | 17:03 | |
| *** DericHorn-HP has joined #openstack-swift | 17:18 | |
| clayg | honestly that rax/intel thing sounds kinda cool - kudos! | 17:25 |
| *** mahatic has joined #openstack-swift | 17:31 | |
| *** mahatic_ has quit IRC | 17:35 | |
| *** mahatic has quit IRC | 17:36 | |
| *** CaioBrentano has joined #openstack-swift | 17:37 | |
| *** mahatic has joined #openstack-swift | 17:43 | |
| *** marzif has quit IRC | 17:45 | |
| minwoob | Just out of curiosity, it seems that most distributed storage system alternatives (such as Ceph, GlusterFS, Lustre, etc) are written in C/C++. | 17:46 |
| minwoob | Were there some reasons originally, for Swift's python origins? | 17:46 |
| MooingLemur | Pretty sure it was OpenStack in general | 17:47 |
| redbo | Well swift predates openstack by a year or so. Mostly we just all liked python. | 17:48 |
| minwoob | MooingLemur: Swift was designed/implemented before that, no? | 17:48 |
| MooingLemur | aha, wasn't sure about that | 17:48 |
| *** janonymous has quit IRC | 17:51 | |
| minwoob | redbo: Yeah, it makes me wonder in the other direction, actually -- as to why the others weren't initially implemented using Python like Swift. | 17:54 |
| redbo | I don't know. I'm not so hot on python these days. Its refusal to acknowledge the existence of multi-core processors makes for a lot of challenges. | 17:58 |
| minwoob | redbo: That's true. | 18:11 |
| *** eandersson has quit IRC | 18:11 | |
| *** Fin1te has joined #openstack-swift | 18:13 | |
| clayg | minwoob: I think some of it at least must have had something to do with swift picking *web* storage - ended up with api-systems/web-services guys that knew http and python? other storage systems started in kernal modules and research projects with contributions from people with old storage/file-system backgrounds? | 18:16 |
| clayg | I honestly can't understand why people write things in c++ on purpose. c sure - but c++ seems more like "I like c, but it takes a frustrating amount of time to do anything useful, and I'm not really going to explore other options - so I'll pretend like learning c++ is a thing that you do a lifetime!" | 18:17 |
| *** Fin1te has quit IRC | 18:20 | |
| *** geaaru has quit IRC | 18:28 | |
| minwoob | clayg: So you're suggesting, perhaps that their rationale for c/c++ wasn't for the performance reasons one would initially suspect, but rather because they had built up a proclivity towards it? | 18:28 |
| notmyname | I'd guess that most projects use a particular language because that's what the original devs were comfortable with | 18:29 |
| *** hrou has quit IRC | 18:30 | |
| *** hrou has joined #openstack-swift | 18:30 | |
| minwoob | notmyname: Right. | 18:31 |
| openstackgerrit | Michael Barton proposed openstack/swift: go: allow profiling https://review.openstack.org/205232 | 18:49 |
| openstackgerrit | Michael Barton proposed openstack/swift: go: goroutine leak https://review.openstack.org/205234 | 18:50 |
| *** sanchitmalhotra1 has joined #openstack-swift | 18:53 | |
| *** bitblt has joined #openstack-swift | 18:55 | |
| *** sanchitmalhotra has quit IRC | 18:55 | |
| *** bitblt has quit IRC | 18:58 | |
| *** DericHorn-HP has quit IRC | 19:13 | |
| *** petertr7_away is now known as petertr7 | 19:24 | |
| clayg | redbo: nadeem: ^ FASCINATING! | 19:26 |
| nadeem | Go Profiler is fun :) | 19:28 |
| clayg | oohhhhhh - topic for the hackathon?! | 19:28 |
| nadeem | I am still on waitlist for hackathon. | 19:29 |
| notmyname | shouldn't be | 19:29 |
| clayg | notmyname: ^ fix this | 19:29 |
| nadeem | however redbo will be there I guess | 19:29 |
| notmyname | "still". you signed up today | 19:29 |
| nadeem | yep sorry I was late :) | 19:30 |
| clayg | nadeem: redbo's not that great | 19:30 |
| *** mahatic has quit IRC | 19:30 | |
| *** annegentle has quit IRC | 19:37 | |
| *** jlhinson_ has joined #openstack-swift | 19:39 | |
| *** jlhinson has quit IRC | 19:39 | |
| *** DericHorn-HP has joined #openstack-swift | 19:40 | |
| *** Fin1te has joined #openstack-swift | 19:42 | |
| *** DericHorn-HP has quit IRC | 19:44 | |
| *** DericHorn-HP has joined #openstack-swift | 19:45 | |
| *** wer has quit IRC | 19:51 | |
| *** wer has joined #openstack-swift | 19:55 | |
| *** minwoob_ has joined #openstack-swift | 20:00 | |
| redbo | no argument | 20:02 |
| *** minwoob has quit IRC | 20:02 | |
| *** annegentle has joined #openstack-swift | 20:09 | |
| *** blmartin has quit IRC | 20:10 | |
| *** blmartin has joined #openstack-swift | 20:10 | |
| *** annegentle has quit IRC | 20:11 | |
| *** DericHorn-HP has quit IRC | 20:12 | |
| *** annegentle has joined #openstack-swift | 20:12 | |
| *** annegentle has quit IRC | 20:13 | |
| *** annegentle has joined #openstack-swift | 20:13 | |
| *** petertr7 is now known as petertr7_away | 20:17 | |
| *** petertr7_away is now known as petertr7 | 20:20 | |
| clayg | lol | 20:32 |
| *** zhill has joined #openstack-swift | 20:35 | |
| clayg | ok, what's worth reviewing today? | 20:38 |
| clayg | tdasilva: merge conflict!? | 20:38 |
| tdasilva | clayg: yeah...just noticed that...haven't tried a rebase just yet, but will in a bit...working on the matrix test atm | 20:40 |
| clayg | tdasilva: did we ever decide how we think manifests and segments in containers with versions turned on are *supposed* to work - is there any functional tests that prove it? | 20:40 |
| clayg | tdasilva: oh yeah :'( | 20:40 |
| tdasilva | clayg: check it out: https://docs.google.com/spreadsheets/d/1hyo3Q7SS70kG0ZCR3NvXJZ6ikeqqQ04RcRYjOLb44m8/edit?usp=sharing | 20:40 |
| clayg | tdasilva: yeah I think that's the ticket! | 20:41 |
| clayg | tdasilva: oic "container configuration" refers to the option in the container server *config* | 20:42 |
| tdasilva | clayg: yeah, let me try to make that clear | 20:42 |
| clayg | tdasilva: there's another vector that has to do with a container that has x-version-location stored in it directly (true/false) | 20:42 |
| tdasilva | clayg: "directly"??? | 20:43 |
| tdasilva | clayg: not sure what you mean by that | 20:43 |
| clayg | tdasilva: when you turn on the container server option and then post x-versions-location to a a container - the container db gets a metadata value for "x-versions-location" instead of "x-sysmeta-versions-location" | 20:44 |
| tdasilva | right, ok, that's what i thought...so yeah..that is sort of implicit on that matrix, I was trying to figure out how to call that out | 20:45 |
| clayg | tdasilva: like one of the weird cases was when you have a container with x-versions-location then you upgrade and turn on the middleware and change x-sysmeta-versions-location and then turn off th emiddleware :\ | 20:45 |
| clayg | i mean on one hand - screw 'em - dont' turn off the middleware after you turn it on (or turn of the container you jerk) - but OTOH i guess it sucks for you? | 20:46 |
| clayg | tdasilva: or does the thing that sets x-sysmeta-versions-location always set x-versions-location? | 20:47 |
| tdasilva | yeah, if you turn on middleware, then turn off, it's OFF, no? | 20:47 |
| clayg | ^ cause that acctually seems pretty sweet | 20:47 |
| clayg | no middleware off depends on the container server setting? | 20:48 |
| tdasilva | crap this thing is confusing :/ | 20:48 |
| tdasilva | looking... | 20:48 |
| clayg | I mean I think I like documenting that once you turn on the middleware you should turn off the container and you can't ever turn versions on the container back on | 20:48 |
| clayg | like you turn on or off versions - but you *have* to use the middleware once you turn it on - and you MUST turn off the container option | 20:49 |
| tdasilva | yeah, once you decide to move over to the middleware, stick with it | 20:49 |
| clayg | idk, maybe just something a doc would be good enough | 20:49 |
| clayg | tdasilva: yeah I think once you turn on the middleware - as long as you turn off the container - you're fine | 20:50 |
| clayg | the only time you should ahve container on if you have middleware off | 20:50 |
| clayg | ^ there that's it | 20:50 |
| clayg | ou could dfg_: ou | 20:50 |
| tdasilva | clayg: that certainly simplifies the code | 20:53 |
| clayg | how so? | 20:54 |
| tdasilva | clayg: like you said, then we don't have to handle the case of the user turning middleware on and off and still expecting things to work | 20:58 |
| clayg | tdasilva: ah yeah - so we should document that | 20:58 |
| tdasilva | clayg: the other thing i need to look into is mattoliverau question about where to place in the pipeline...which also deals with your question about dlo/slo | 21:01 |
| *** annegent_ has joined #openstack-swift | 21:02 | |
| clayg | ok, so maybe we have columns: "container-server-config-off & middleware-off", "container-server-config-on & middleare-off", "container-server-config-off & middleware-on" | 21:02 |
| clayg | tdasilva: that's why I asked | 21:02 |
| clayg | tdasilva: about slo | 21:02 |
| clayg | tdasilva: i *thought* in some swift meeting you were like "how is this supposed to work" and we all said "you tell us!" | 21:03 |
| clayg | maybe that was acoles_away? | 21:03 |
| clayg | tdasilva: anway given those columns you could enumerate a matrix for all the combinations of x-versions-location & x-container-sysmtea-versions-location | 21:04 |
| *** annegentle has quit IRC | 21:05 | |
| clayg | tdasilva: like another weird one was "turn on x-versions-locations, turn on middleware, change x-sysmeta-versions-location, remove x-sysmeta-versions-location" | 21:05 |
| *** robefran has quit IRC | 21:05 | |
| tdasilva | clayg: yeah, basically I was trying to keep the same behavior at is now and then if we think we need change the behavior we should do it in a separate patch | 21:05 |
| clayg | I suppose we could enumerate "container-server-config-on & middleware-on" as *not recommended* | 21:06 |
| clayg | tdasilva: +1000000 to don't change it in this patch! | 21:06 |
| clayg | tdasilva: was mattoliverau suggesting a change or enumerating a difference? that's why I asked about "what's expected and what's tested" | 21:06 |
| tdasilva | clayg: hehehe...don't want this thing going to 100 patchsets :D | 21:06 |
| clayg | tdasilva: god help us | 21:06 |
| tdasilva | clayg: what I can say is I don't think even current behavior is not very good | 21:07 |
| tdasilva | clayg: especially when it comes to versioning and slo and deleting segments and what not... | 21:08 |
| clayg | tdasilva: ok, current behavior sucks so bad it's not worth testing? | 21:08 |
| clayg | tdasilva: i just... if we change it we have to justify it - i mean... I don't want to make it worse? idk - i feel like the risk of breaking a client that is getting something out of the current behavior is enough of a reason to justify tests to enumerate what the current behavior is | 21:09 |
| clayg | if we don't have those tests then either mattoliverau or I could write them | 21:09 |
| tdasilva | clayg: no, what I mean is that there isn't even automated testing for it, I think it might be something that's been overlooked??? maybe??? | 21:09 |
| clayg | tdasilva: +1000 overlooked! | 21:09 |
| clayg | tdasilva: maybe the fact that it's lasted this long with no bug reports meaning no one is using it | 21:10 |
| tdasilva | clayg: yeah yeah, i agree with adding those tests and then we try to keep the behavior the same, I think that would be good. Then in a future work, we move to how we think things should work | 21:10 |
| clayg | tdasilva: if we've managed not to break clients through the slo extraction maybe we're fine - it seems unlikely we didn't slip a behavior change in there somewhere? | 21:11 |
| *** petertr7 is now known as petertr7_away | 21:11 | |
| clayg | tdasilva: hell we don't even *like* the current versions behavior -> https://review.openstack.org/#/c/155421/ | 21:11 |
| clayg | sigh | 21:12 |
| tdasilva | lol | 21:12 |
| clayg | we have got to get this crap out of the proxy | 21:12 |
| tdasilva | agreed | 21:12 |
| clayg | i'm just going to +A it and see what happens | 21:12 |
| clayg | :P | 21:12 |
| clayg | notmyname: ^ close your eyes real quick | 21:12 |
| tdasilva | rofl | 21:12 |
| * clayg forgot momentarily notmyname wrote most (all?) of the original versions code? | 21:12 | |
| notmyname | clayg: just means I'll be the first hater on the current state of things | 21:13 |
| tdasilva | clayg: so here is my plan: 1. test, figure out the current behavior when it comes to cross-functionality between slo and versioning. 2. Add automated tests against current master. 3. make sure those tests passes again middleware patch. 4. make necessary changes to patch | 21:15 |
| tdasilva | mattoliverau: ^^^ | 21:16 |
| clayg | mattoliverau: I just left on a review that says that plan should be ordered 3, 2, 5 - where 5 is "work on the undelete version of object versions middleware" | 21:20 |
| *** mgarza_ has joined #openstack-swift | 21:20 | |
| clayg | oh wait... no what number is merge patch 134347 - that should be number 1 | 21:20 |
| patchbot | clayg: https://review.openstack.org/#/c/134347/ | 21:20 |
| mattoliverau | That would be 6 :p | 21:22 |
| mattoliverau | And morning :) | 21:22 |
| clayg | yay matt! | 21:22 |
| clayg | mattoliverau: yeah so i'm advocating we just rip this crap out of the proxy ASAP and then figure out where we're at? | 21:22 |
| mattoliverau | If you moved the middle ware to the right of slo, slo deletes work with versioning.. Slo + versioning (slo being manifested and etagged) does seem edge casey tho | 21:23 |
| clayg | I don't even know what it means to version a *slo - am I versioning the whole thing? just the manifest? what if the segments are in a container that's different form the manifest? my head hurts | 21:23 |
| clayg | slo deletes don't "work with versioning" - what is the expected behavior? whose stored where - what's versioned? | 21:24 |
| clayg | how do you define "work with versioning"? | 21:25 |
| clayg | tdasilva: new plan - if we move the middleware and mattoliverau +2's I think we might be able to merge!? | 21:25 |
| mattoliverau | Yeah I know, but if you can now version manifests and use the same container you could version entire slos.. Kinda.. | 21:25 |
| tdasilva | yeah, the problem becomes if users uploaded new versions of a segment but not if others | 21:25 |
| tdasilva | i think that breaks slo deletion | 21:25 |
| tdasilva | s/not if others/not of others | 21:26 |
| clayg | mattoliverau: please don't feel threating by my nay-saying - if you think moving the middleware is better I'm all for it - anyone on the planet knows what versions + *lo's should do better than me | 21:27 |
| *** jrichli has quit IRC | 21:27 | |
| mattoliverau | Lol, what if you use segments for other things (not just slo), they are versioned. You create and slo, then she time delete it, those old segments might still be important, but I'm the current state they'll be missing from the container (but versions would still exist) | 21:28 |
| mattoliverau | S/she/some | 21:28 |
| *** zhill has quit IRC | 21:28 | |
| mattoliverau | S/I'm/in | 21:29 |
| mattoliverau | I'm failing at writing.. Need coffee and breakfast :p | 21:29 |
| tdasilva | lol | 21:29 |
| *** Fin1te has quit IRC | 21:30 | |
| tdasilva | just spent a couple of seconds deciphering that | 21:30 |
| mattoliverau | So I think its not that slos are versionible but so the version delete behavior works as expected | 21:31 |
| *** DericHorn-HP has joined #openstack-swift | 21:33 | |
| clayg | nadeem: notmyname says he squared it | 21:33 |
| nadeem | yep true. Thanks notmyname ! | 21:34 |
| tdasilva | clayg: just gave you permissions on the spreadsheet | 21:34 |
| clayg | tdasilva: no it's too late - i'm doing my *own* spreadsheet now :P | 21:35 |
| tdasilva | lol...is there an etherpad like tool for spreadsheets? | 21:35 |
| *** jasondotstar has joined #openstack-swift | 21:36 | |
| *** tongli has quit IRC | 21:36 | |
| clayg | tdasilva: isn't google docs the eatherpad like tool for google docs? Or do you just mean "sharable by url by default" | 21:38 |
| tdasilva | yeah, the last part | 21:38 |
| clayg | tdasilva: i can't copy and paste from my doc to yours :\ | 21:40 |
| tdasilva | clayg: so you can't write to my sheet at all or just copy and paste? | 21:41 |
| *** blmartin has quit IRC | 21:44 | |
| *** lpabon has quit IRC | 21:44 | |
| *** CaioBrentano has quit IRC | 21:44 | |
| clayg | tdasilva: no we're collaborting now! | 21:46 |
| peluse | you can always print screen and paste :) | 21:50 |
| *** robefran has joined #openstack-swift | 21:57 | |
| openstackgerrit | Merged openstack/swift: Imported Translations from Transifex https://review.openstack.org/204914 | 22:06 |
| notmyname | I just got the openstack summit session voting email | 22:08 |
| *** DericHorn-HP has quit IRC | 22:08 | |
| *** DericHorn-HP has joined #openstack-swift | 22:10 | |
| clayg | notmyname: goodness it's that time again *already* | 22:14 |
| notmyname | someone named Pete who works at Red Hat has "The Plot to Destroy OpenStack Swift Using C++: Enhancements of Swift API Compatibility in Ceph RADOS Gateway" | 22:15 |
| notmyname | https://www.openstack.org/summit/tokyo-2015/vote-for-speakers/presentation/5281 | 22:15 |
| *** mgarza_ has quit IRC | 22:15 | |
| clayg | what a great topic for the Openstack Summit! | 22:15 |
| notmyname | wait what?! dfg_ and redbo have a talk submitted? https://www.openstack.org/summit/tokyo-2015/vote-for-speakers/presentation/5366 | 22:16 |
| clayg | "The Plot to Destroy Openstack Swift Usgin go"? | 22:16 |
| notmyname | lol | 22:16 |
| notmyname | "The Plit to Destroy Openstack Swift Using SMR Drives" | 22:16 |
| hrou | Sorry notmyname, there's only room for one SMR talk ; - ) Though I'm open to renaming it. | 22:19 |
| *** nadeem has quit IRC | 22:20 | |
| notmyname | hrou: I think that's the subtitle of the existing one | 22:20 |
| notmyname | or maybe, "How to destroy SMR drives with Swift" would be more appropriate ;-) | 22:20 |
| hrou | We suspected that actually happened at one point, but alas, it was our inexperience and it was simply driving an internal defrag. But I agree, an appropriate title indeed : ) | 22:22 |
| tdasilva | "Swift the Social Butterfly" ??? | 22:22 |
| tdasilva | i wonder if there will be voting for best talk tittles | 22:23 |
| notmyname | yeah, that caught my eye too | 22:23 |
| clayg | redbo: can you define what you mean by "orphaned data" - I can think of few things I've seen that might be called that - but you hinted at monitoring it - so I'm... curious ;) | 22:23 |
| redbo | objects that dont exizt in containers | 22:24 |
| redbo | but are on object nodes | 22:25 |
| *** jome has joined #openstack-swift | 22:25 | |
| clayg | redbo: you can *track* that - like ... how much? | 22:26 |
| *** mahatic has joined #openstack-swift | 22:27 | |
| clayg | i mean don't get me wrong - i've got my scripts - but I'm way to scared to let them loose - depending on how much you see in otherwise stable environments - maybe we should go looking for it more aggresively? | 22:27 |
| redbo | not exactly but we pretty regularly go see how much we can find and try to match it back to events or periods of high loaf or whatever | 22:27 |
| redbo | load | 22:27 |
| clayg | swift always just be loafin' around | 22:28 |
| jome | Hi. I'm new to swift and would like to experiment with it for a project we are working on. There is so much documentation out there that its quite overwhelming. My use case is developing a photo sharing app based on django | 22:28 |
| notmyname | jome: that's great! | 22:28 |
| jome | I get the part where I can save images via the cli, but I'm quite stuck as to how to retrieve these via a web browser for example. | 22:29 |
| jome | So it anyone could point me in the right directions and hold my hand aling the way, that would be helpful | 22:30 |
| jome | So if anyone could point me in the right directions and hold my hand along the way, that would be helpful | 22:30 |
| notmyname | jome: do you want anyone to be able to access them (ie public) or only give out limited access to certain people? | 22:30 |
| jome | a mix of both actually... I would like things like static files to be accessible publicly (css, jpgs, js) and users can set permissions on their generated content. | 22:32 |
| clayg | tdasilva: yeah so I think https://docs.google.com/spreadsheets/d/1hyo3Q7SS70kG0ZCR3NvXJZ6ikeqqQ04RcRYjOLb44m8/edit#gid=0 is pretty good | 22:32 |
| clayg | i'm pretty confident of about half the behaviors - the others I have my expections... | 22:33 |
| clayg | mattoliverau: did you decide what we're going to do about the middleware placement versions + *.lo? | 22:33 |
| notmyname | jome: ok, public access is a little simpler | 22:33 |
| jome | Ok... any info how I would go about that? Like some example code | 22:36 |
| clayg | zackmdavis: ^ swift-django app! | 22:36 |
| notmyname | sorry. got distracted. seems that the Apple "swift" CLI has taken over on my mac | 22:37 |
| jome | ha ha ha | 22:37 |
| *** mahatic has quit IRC | 22:37 | |
| clayg | notmyname: rewrite python-swiftclient in swift? | 22:37 |
| notmyname | of course! | 22:38 |
| jome | @notmyname any ideas? | 22:39 |
| notmyname | jome: you've already got some objects in a container in swift? | 22:40 |
| jome | yep | 22:40 |
| *** jasondotstar has quit IRC | 22:40 | |
| notmyname | ok, then `swift post <container> -r ".r:*" will make the container public | 22:41 |
| notmyname | let me know when you have that done | 22:42 |
| jome | That would also need to include the auth details. right? | 22:43 |
| notmyname | yes, if you haven't set them up as environment variables | 22:43 |
| notmyname | if you have the latest version of the swift cli tool, you can do `swift auth -v` to get the bash commands to set the environment variables | 22:44 |
| jome | doing that now... | 22:46 |
| jome | Done. No response though. | 22:47 |
| clayg | ugh - so i was going to review 120709 and realized I'll have to bust out sqlite explain and shit and try to wrap my head around a bunch of sqlite index nonsense :\ | 22:47 |
| *** annegent_ has quit IRC | 22:47 | |
| notmyname | jome: no response for the container post or the auth command? | 22:47 |
| clayg | notmyname: the swift auth command merged! | 22:47 |
| * clayg hugs timburke | 22:48 | |
| notmyname | clayg: yeah, and it's in the latest release! thanks :-) | 22:48 |
| jome | the container post | 22:48 |
| notmyname | jome: ah, that's expected | 22:48 |
| jome | ok | 22:48 |
| * timburke hugs clayg back | 22:48 | |
| notmyname | jome: ok, now you can get the data via a browser. of course the first question is "what's the url?" | 22:48 |
| timburke | now if we could get https://review.openstack.org/#/c/189387/ in too, we'd have a decent auth story! | 22:49 |
| notmyname | jome: to get your storage URL, do `swift auth` and the first line will have a url in it | 22:49 |
| jome | ok... | 22:49 |
| notmyname | jome: you can also get the url with `swift stat -v` | 22:49 |
| jome | Ok. got the url | 22:50 |
| notmyname | great | 22:50 |
| notmyname | jome: ok, my url on my dev environment is http://saio:8080/v1/AUTH_test. since the container named I used is "c" and I upload an object "1MB", then I can access http://saio:8080/v1/AUTH_test/c/1MB in a broswer | 22:50 |
| notmyname | jome: replace with your appropriate values, and you should be good | 22:51 |
| jome | Yeah!!! | 22:51 |
| jome | Worked like a charm! | 22:51 |
| notmyname | yay! | 22:51 |
| *** chlong has quit IRC | 22:52 | |
| jome | So what this would mean is that for any container I need to be public, i run the command there? | 22:53 |
| *** robefran has quit IRC | 22:53 | |
| jome | All well and good there... | 22:53 |
| jome | Now the hard part... :) | 22:53 |
| notmyname | jome: correct. you'd need to set a container as public. you only need to do that once | 22:54 |
| notmyname | jome: the hard part? | 22:55 |
| jome | To access files that require authentication/authorization | 22:56 |
| notmyname | via a browser? | 22:57 |
| jome | Yep. I'd like to build a django app that would serve up such files | 22:57 |
| notmyname | jome: https://swiftstack.com/blog/2015/01/29/swift-feature-highlight-tempurls/ | 22:58 |
| notmyname | jome: use tempurls. that link is how/what/etc | 22:58 |
| jome | Ok... | 22:59 |
| *** haypo has left #openstack-swift | 22:59 | |
| notmyname | jome: basically, you get a time-limited URL that is good for only one kind of operation. you can make that server-side and then pass it to the browser | 22:59 |
| *** david-lyle has quit IRC | 23:02 | |
| jome | I see... So if I upload a photo, say via a form to save in swift, I save it with some accompanying metadata and then when a request is made for that photo, say by my friends, the app generally would create that tempurl for the photo to be served up the browser/client | 23:03 |
| notmyname | yes | 23:04 |
| *** km has joined #openstack-swift | 23:05 | |
| notmyname | jome: the temp-url keys don't have to be created for every object. you can create the tempurl key once (or use 2 keys and rotate them on a schedule if you want) | 23:05 |
| *** hrou has quit IRC | 23:05 | |
| jome | Thanks for this man... Certainly a good nudge in the right direction. | 23:09 |
| jome | I'll go through this and give you a shout later? | 23:10 |
| mattoliverau | clayg: i think the middleware should move to the otherside of slo, not just because it makes slos+versions work (if that even makes sense) but mainly so anything deleting a versioned object is handled correctly. Then I'm happy to +2 to get versions in. I.e. atm you might have been using the slo segments as individual versioned objects. You place a slo, later delete it.. Now the container holding the latest version of | 23:11 |
| mattoliverau | the segments is now missing them, however the versions container still hold older versions of the files.. Moving the middle ware fixes this. Edge case yes, but I think is expected behaviour (at least in my mind) | 23:11 |
| notmyname | jome: sure. and if I'm not around, there's normally someone who can answer. we've got pretty decent global coverage, even if it is quieter at times :-) | 23:12 |
| jome | Great stuff... Say I wanted someone who I needed to formally consult for support, would I find such here? | 23:13 |
| notmyname | jome: most of us work for companies that have some sort of business interest in swift | 23:14 |
| clayg | notmyname: what'd I do to make swiftclient say "error in setup command: 'tests_require' must be a string or list of strings containing valid project/version requirement specifiers | 23:14 |
| clayg | " | 23:14 |
| notmyname | jome: eg I work at swiftstack. we have a software product we sell. others work for service providers (rackspace, hp, ibm, etc). others, like red hat, have different kinds of support options | 23:15 |
| jome | Aah... | 23:15 |
| notmyname | jome: so yes, here's better than anywhere else. but we aren't sales people ;-) | 23:15 |
| notmyname | clayg: I don't know. and it took a lot of "FINE. I'll update ALL THE THINGS" and lots of grumpiness to make it work again | 23:16 |
| jome | I'll take helpful people over sales anyday! :) | 23:17 |
| *** gyee has joined #openstack-swift | 23:20 | |
| *** wbhuber has quit IRC | 23:22 | |
| *** kei_yama has joined #openstack-swift | 23:23 | |
| *** DericHorn-HP has quit IRC | 23:24 | |
| jome | Is there no other way to achieve the same thing other than tempurls? Unless I am missing something the python client does not seem to provide such a function. | 23:33 |
| notmyname | the trick is that you can't send custom headers in a browser. ie you can't give someone a url that can only be accessed with an auth token in a header and expect them to be able to load it in a browser (without additional js work). so the auth info has to be in the url itself. that's where you get the signed urls | 23:35 |
| *** gyee has quit IRC | 23:36 | |
| notmyname | so you've got (1) public, no-token access (2) signed url access, and (3) private, with a token access. that's your choice | 23:36 |
| *** david-lyle has joined #openstack-swift | 23:36 | |
| jome | Hmm... ok. | 23:36 |
| notmyname | you could proxy the data from swift through your django app, this using 3 on the backend but allowing non-tken access behind a login on the front end | 23:37 |
| notmyname | but then you've added a lot of complexity and you'll be limited by the ability of the django app to scale | 23:37 |
| *** gyee has joined #openstack-swift | 23:38 | |
| notmyname | it's normally better to have the end-users talk directly to swift | 23:38 |
| jome | So, you'd advise 1 or 2 from a scalability perspective? | 23:39 |
| notmyname | and a trust perspective. you can hand public and temporary signed urls to untrusted users. you can't hand the creds or a token to untrusted users | 23:39 |
| notmyname | but yes, from a scale perspective too :-) | 23:39 |
| jome | great stuff. Thanks... I'll dabble a bit more | 23:40 |
| *** darrenc is now known as darrenc_afk | 23:44 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Fix string formatting in dispersion cli command https://review.openstack.org/205326 | 23:48 |
| *** darrenc_afk is now known as darrenc | 23:54 | |
| *** jasondotstar has joined #openstack-swift | 23:56 | |
| *** ho has joined #openstack-swift | 23:57 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!