| *** theanalyst has joined #openstack-swift | 00:00 | |
| openstackgerrit | paul luse proposed openstack/swift: Erasure Code Reconstructor https://review.openstack.org/131872 | 00:24 |
|---|---|---|
| *** dmorita has joined #openstack-swift | 00:25 | |
| *** ho has joined #openstack-swift | 00:28 | |
| ho | morning! | 00:29 |
| *** rdaly2 has joined #openstack-swift | 00:30 | |
| *** rdaly2 has quit IRC | 00:35 | |
| *** jrichli has joined #openstack-swift | 00:44 | |
| *** gyee has quit IRC | 00:44 | |
| openstackgerrit | Merged openstack/swift: Fix param description on _make_request https://review.openstack.org/162592 | 00:50 |
| *** wshao has quit IRC | 00:53 | |
| mattoliverau | ho: morning | 01:00 |
| *** thumpba has quit IRC | 01:04 | |
| *** bkopilov has joined #openstack-swift | 01:24 | |
| *** zhill_ has joined #openstack-swift | 01:33 | |
| *** zhill_ has quit IRC | 01:34 | |
| *** tellesnobrega_ has joined #openstack-swift | 02:14 | |
| *** tellesnobrega_ has quit IRC | 02:14 | |
| *** Anticimex has quit IRC | 02:43 | |
| *** wshao has joined #openstack-swift | 02:54 | |
| *** ho has quit IRC | 02:55 | |
| *** ho has joined #openstack-swift | 02:55 | |
| *** wshao has quit IRC | 02:58 | |
| *** haomaiwang has joined #openstack-swift | 03:19 | |
| *** bkopilov has quit IRC | 03:30 | |
| *** theanalyst has quit IRC | 03:48 | |
| *** jrichli has quit IRC | 03:55 | |
| *** ppai has joined #openstack-swift | 04:05 | |
| *** theanalyst has joined #openstack-swift | 04:20 | |
| *** panbalag has quit IRC | 04:26 | |
| *** wshao has joined #openstack-swift | 04:48 | |
| *** SkyRocknRoll has joined #openstack-swift | 04:53 | |
| *** wshao_ has joined #openstack-swift | 04:53 | |
| *** wshao has quit IRC | 04:57 | |
| *** daniel___ has joined #openstack-swift | 04:59 | |
| *** zaitcev has quit IRC | 05:04 | |
| *** daniel___ has quit IRC | 05:07 | |
| *** bkopilov has joined #openstack-swift | 05:16 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Prefer policy to policy_index https://review.openstack.org/162775 | 05:40 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Return 400 from object server if policy does not exist https://review.openstack.org/162696 | 05:54 |
| *** ho__ has joined #openstack-swift | 05:56 | |
| *** ho has quit IRC | 05:58 | |
| *** km_ has joined #openstack-swift | 05:58 | |
| *** km has quit IRC | 05:59 | |
| *** ppai has quit IRC | 06:06 | |
| clayg | tdasilva: I didn't even try to go find the object-versioning patch today - hope that's still going well | 06:09 |
| *** wshao_ has quit IRC | 06:10 | |
| *** ppai has joined #openstack-swift | 06:19 | |
| *** zhill_ has joined #openstack-swift | 06:27 | |
| *** david-lyle has quit IRC | 06:30 | |
| *** rdaly2 has joined #openstack-swift | 06:35 | |
| *** david-lyle has joined #openstack-swift | 06:39 | |
| *** rdaly2 has quit IRC | 06:39 | |
| *** zhill_ has quit IRC | 06:48 | |
| *** wshao has joined #openstack-swift | 06:50 | |
| *** wshao has quit IRC | 06:57 | |
| *** silor has joined #openstack-swift | 07:03 | |
| *** bkopilov has quit IRC | 07:07 | |
| openstackgerrit | Yuan Zhou proposed openstack/swift: Fix copy from different type policy https://review.openstack.org/162890 | 07:10 |
| *** david-lyle is now known as david-lyle_afk | 07:11 | |
| *** chlong has quit IRC | 07:21 | |
| *** bkopilov has joined #openstack-swift | 07:22 | |
| *** nshaikh has joined #openstack-swift | 07:26 | |
| *** jistr has joined #openstack-swift | 07:31 | |
| *** openstackgerrit has quit IRC | 07:35 | |
| *** openstackgerrit has joined #openstack-swift | 07:36 | |
| *** ppai has quit IRC | 07:39 | |
| *** km_ has quit IRC | 07:44 | |
| *** mmcardle has joined #openstack-swift | 07:52 | |
| *** ppai has joined #openstack-swift | 07:54 | |
| *** rledisez has joined #openstack-swift | 08:09 | |
| openstackgerrit | Yuan Zhou proposed openstack/swift: Fix Etag check on EC put for if_none_match https://review.openstack.org/162905 | 08:13 |
| *** silor has quit IRC | 08:14 | |
| *** Anticimex has joined #openstack-swift | 08:20 | |
| *** geaaru has joined #openstack-swift | 08:23 | |
| *** ho__ has quit IRC | 08:27 | |
| *** haomaiwang has quit IRC | 08:32 | |
| *** rdaly2 has joined #openstack-swift | 08:36 | |
| *** haomaiwang has joined #openstack-swift | 08:37 | |
| *** rdaly2 has quit IRC | 08:41 | |
| openstackgerrit | Yuan Zhou proposed openstack/swift: Fix EC PUT on HTTP_CONFLICT or HTTP_PRECONDITION_FAILED https://review.openstack.org/162047 | 08:58 |
| *** acoles_away is now known as acoles | 08:58 | |
| *** silor has joined #openstack-swift | 08:59 | |
| *** ho has joined #openstack-swift | 09:01 | |
| openstackgerrit | Daisuke Morita proposed openstack/swift: object-updater runs for all async_pending directories https://review.openstack.org/141252 | 09:05 |
| *** nellysmitt has joined #openstack-swift | 09:06 | |
| *** aix has joined #openstack-swift | 09:19 | |
| *** nshaikh has quit IRC | 09:20 | |
| *** david-lyle_afk has quit IRC | 09:30 | |
| *** david-lyle_afk has joined #openstack-swift | 09:30 | |
| *** jistr has quit IRC | 09:34 | |
| *** jistr has joined #openstack-swift | 09:46 | |
| *** bkopilov has quit IRC | 10:15 | |
| *** chlong has joined #openstack-swift | 10:17 | |
| *** me has joined #openstack-swift | 10:30 | |
| *** me is now known as Guest2514 | 10:30 | |
| donagh | acoles: I just re-read the crypto spec...there was a paragraph I could not parse; otherwise looking good. | 10:31 |
| Guest2514 | good morning everybody! :) | 10:31 |
| *** aix has quit IRC | 10:32 | |
| acoles | donagh: thanks! | 10:32 |
| Guest2514 | i'm deploying swift on a ubuntu14.04 VM using this doc http://docs.openstack.org/developer/swift/development_saio.html#common-post-device-setup | 10:32 |
| Guest2514 | i get an error when i try to build swift | 10:33 |
| Guest2514 | cd $HOME/swift; sudo python setup.py develop; cd - | 10:34 |
| Guest2514 | running develop | 10:34 |
| Guest2514 | running egg_info | 10:34 |
| Guest2514 | writing pbr to swift.egg-info/pbr.json | 10:34 |
| Guest2514 | writing requirements to swift.egg-info/requires.txt | 10:34 |
| Guest2514 | writing swift.egg-info/PKG-INFO | 10:34 |
| Guest2514 | writing top-level names to swift.egg-info/top_level.txt | 10:34 |
| Guest2514 | writing dependency_links to swift.egg-info/dependency_links.txt | 10:34 |
| Guest2514 | writing entry points to swift.egg-info/entry_points.txt | 10:34 |
| Guest2514 | [pbr] Reusing existing SOURCES.txt | 10:34 |
| Guest2514 | running build_ext | 10:34 |
| Guest2514 | Creating /usr/local/lib/python2.7/dist-packages/swift.egg-link (link to .) | 10:34 |
| Guest2514 | swift 2.2.2.post113 is already the active version in easy-install.pth | 10:34 |
| Guest2514 | Installing swift-account-audit script to /usr/local/bin | 10:34 |
| Guest2514 | Installing swift-account-auditor script to /usr/local/bin | 10:34 |
| Guest2514 | Installing swift-account-info script to /usr/local/bin | 10:34 |
| Guest2514 | Installing swift-account-reaper script to /usr/local/bin | 10:34 |
| Guest2514 | Installing swift-account-replicator script to /usr/local/bin | 10:34 |
| Guest2514 | Installing swift-account-server script to /usr/local/bin | 10:34 |
| Guest2514 | Installing swift-config script to /usr/local/bin | 10:34 |
| Guest2514 | Installing swift-container-auditor script to /usr/local/bin | 10:34 |
| Guest2514 | Installing swift-container-info script to /usr/local/bin | 10:34 |
| Guest2514 | Installing swift-container-replicator script to /usr/local/bin | 10:34 |
| Guest2514 | Installing swift-container-server script to /usr/local/bin | 10:34 |
| Guest2514 | Installing swift-container-sync script to /usr/local/bin | 10:34 |
| Guest2514 | Installing swift-container-updater script to /usr/local/bin | 10:34 |
| Guest2514 | Installing swift-container-reconciler script to /usr/local/bin | 10:34 |
| ekarlso | Guest2514: ... please use pastebin :p | 10:34 |
| Guest2514 | Installing swift-reconciler-enqueue script to /usr/local/bin | 10:34 |
| Guest2514 | Installing swift-dispersion-populate script to /usr/local/bin | 10:35 |
| ekarlso | paste.openstack.org | 10:35 |
| Guest2514 | Installing swift-dispersion-report script to /usr/local/bin | 10:35 |
| Guest2514 | Installing swift-drive-audit script to /usr/local/bin | 10:35 |
| Guest2514 | Installing swift-form-signature script to /usr/local/bin | 10:35 |
| Guest2514 | Installing swift-get-nodes script to /usr/local/bin | 10:35 |
| Guest2514 | Installing swift-init script to /usr/local/bin | 10:35 |
| Guest2514 | Installing swift-object-auditor script to /usr/local/bin | 10:35 |
| Guest2514 | Installing swift-object-expirer script to /usr/local/bin | 10:35 |
| Guest2514 | Installing swift-object-info script to /usr/local/bin | 10:35 |
| Guest2514 | Installing swift-object-replicator script to /usr/local/bin | 10:35 |
| Guest2514 | Installing swift-object-server scr | 10:35 |
| Guest2514 | Installing swift-object-updater script to /usr/local/bin | 10:35 |
| Guest2514 | Installing swift-oldies script to /usr/local/bin | 10:35 |
| Guest2514 | Installing swift-orphans script to /usr/local/bin | 10:35 |
| Guest2514 | Installing swift-proxy-server script to /usr/local/bin | 10:35 |
| Guest2514 | Installing swift-recon script to /usr/local/bin | 10:35 |
| Guest2514 | Installing swift-recon-cron script to /usr/local/bin | 10:35 |
| Guest2514 | Installing swift-ring-builder script to /usr/local/bin | 10:35 |
| Guest2514 | Installing swift-temp-url script to /usr/local/bin | 10:35 |
| Guest2514 | Installed /home/med/swift | 10:35 |
| Guest2514 | Processing dependencies for swift==2.2.2.post113 | 10:35 |
| Guest2514 | Searching for xattr>=0.4 | 10:35 |
| Guest2514 | Reading https://pypi.python.org/simple/xattr/ | 10:35 |
| Guest2514 | Best match: xattr 0.7.5 | 10:35 |
| Guest2514 | Downloading https://pypi.python.org/packages/source/x/xattr/xattr-0.7.5.tar.gz#md5=42e727fef24c8eda20ea00d89da2c9e1 | 10:35 |
| Guest2514 | Processing xattr-0.7.5.tar.gz | 10:35 |
| Guest2514 | Writing /tmp/easy_install-RHVL6g/xattr-0.7.5/setup.cfg | 10:35 |
| Guest2514 | Running xattr-0.7.5/setup.py -q bdist_egg --dist-dir /tmp/easy_install-RHVL6g/xattr-0.7.5/egg-dist-tmp-Yu1xM9 | 10:35 |
| Guest2514 | Searching for cffi>=0.4 | 10:36 |
| Guest2514 | Reading https://pypi.python.org/simple/cffi/ | 10:36 |
| Guest2514 | Best match: cffi 0.9.1 | 10:36 |
| Guest2514 | Downloading https://pypi.python.org/packages/source/c/cffi/cffi-0.9.1.tar.gz#md5=8dbdf23c600845b75654024e434601ce | 10:36 |
| Guest2514 | Processing cffi-0.9.1.tar.gz | 10:36 |
| Guest2514 | Writing /tmp/easy_install-RHVL6g/xattr-0.7.5/temp/easy_install-M1u9rq/cffi-0.9.1/setup.cfg | 10:36 |
| Guest2514 | Running cffi-0.9.1/setup.py -q bdist_egg --dist-dir /tmp/easy_install-RHVL6g/xattr-0.7.5/temp/easy_install-M1u9rq/cffi-0.9.1/egg-dist-tmp-WnXLZu | 10:36 |
| Guest2514 | Package libffi was not found in the pkg-config search path. | 10:36 |
| Guest2514 | Perhaps you should add the directory containing `libffi.pc' | 10:36 |
| Guest2514 | to the PKG_CONFIG_PATH environment variable | 10:36 |
| Guest2514 | No package 'libffi' found | 10:36 |
| Guest2514 | Package libffi was not found in the pkg-config search path. | 10:36 |
| Guest2514 | Perhaps you should add the directory containing `libffi.pc' | 10:36 |
| Guest2514 | to the PKG_CONFIG_PATH environment variable | 10:36 |
| Guest2514 | No package 'libffi' found | 10:36 |
| Guest2514 | Package libffi was not found in the pkg-config search path. | 10:36 |
| Guest2514 | Perhaps you should add the directory containing `libffi.pc' | 10:36 |
| Guest2514 | to the PKG_CONFIG_PATH environment variable | 10:36 |
| Guest2514 | No package 'libffi' found | 10:36 |
| Guest2514 | Package libffi was not found in the pkg-config search path. | 10:36 |
| Guest2514 | Perhaps you should add the directory containing | 10:36 |
| Guest2514 | to the PKG_CONFIG_PATH environment variable | 10:36 |
| Guest2514 | No package 'libffi' found | 10:36 |
| Guest2514 | compiling '_configtest.c': | 10:36 |
| Guest2514 | __thread int some_threadlocal_variable_42; | 10:36 |
| Guest2514 | x86_64-linux-gnu-gcc -pthread -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -fPIC -c _configtest.c -o _configtest.o | 10:36 |
| Guest2514 | success! | 10:37 |
| Guest2514 | removing: _configtest.c _configtest.o | 10:37 |
| Guest2514 | c/_cffi_backend.c:13:17: fatal error: ffi.h: No such file or directory | 10:37 |
| Guest2514 | #include <ffi.h> | 10:37 |
| Guest2514 | ^ | 10:37 |
| Guest2514 | compilation terminated. | 10:37 |
| Guest2514 | error: Setup script exited with error: command 'x86_64-linux-gnu-gcc' failed with exit status 1 | 10:37 |
| Guest2514 | can anybody help me on that ? | 10:37 |
| *** rdaly2 has joined #openstack-swift | 10:38 | |
| *** nellysmitt has quit IRC | 10:41 | |
| *** rdaly2 has quit IRC | 10:42 | |
| *** chlong has quit IRC | 10:51 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: Allow swift-object-info to inspect .meta and .ts files https://review.openstack.org/162306 | 10:51 |
| *** ho has quit IRC | 10:52 | |
| *** haomaiwang has quit IRC | 10:53 | |
| *** aix has joined #openstack-swift | 10:57 | |
| ahale | apt-get install python-cffi ? should probably find a pastebin too or something | 10:58 |
| Guest2514 | @ahale: thank you it worked !! xD | 11:02 |
| Guest2514 | that package have to be mentioned in the 'dependencies section' of the doc | 11:03 |
| ahale | yep I'm surprised it wasn't pulled in by something else, should probs get added indeed | 11:08 |
| acoles | Guest2514: which doc are you following? did you install the packages listed here? | 11:09 |
| ahale | fwiw, I build a SAIO by copy pasting the docs at the weekend with no problems | 11:09 |
| *** nellysmitt has joined #openstack-swift | 11:09 | |
| *** Krast has quit IRC | 11:09 | |
| acoles | Guest2514: sorry, forgot link http://docs.openstack.org/developer/swift/development_saio.html#installing-dependencies | 11:10 |
| Guest2514 | i following the official doc http://docs.openstack.org/developer/swift/development_saio.html | 11:10 |
| Guest2514 | i dont know,may be "python-cffi" is installed by defaut in some distros. I had to install it manually in my 14.04 VM | 11:14 |
| ahale | hm, from your paste you have xattr getting installed by pip pulling in cffi, while my 14.04 vm got it in the second apt-get install | 11:20 |
| openstackgerrit | Alistair Coles proposed openstack/swift-specs: Updates to encryption spec https://review.openstack.org/154318 | 11:23 |
| *** bkopilov has joined #openstack-swift | 11:25 | |
| *** SkyRocknRoll has quit IRC | 11:28 | |
| rsFF | Hi there, | 11:29 |
| rsFF | could someone give me some information on this, https://bugs.launchpad.net/swift/+bug/1428866 | 11:29 |
| openstack | Launchpad bug 1428866 in OpenStack Object Storage (swift) "swift-object-info display for sysmeta" [Wishlist,In progress] - Assigned to Ricardo Ferreira (rsff) | 11:29 |
| *** Guest2514 is now known as notme | 11:30 | |
| *** notme is now known as Guest11309 | 11:30 | |
| *** Guest11309 is now known as m_h | 11:31 | |
| *** m_h is now known as m_han | 11:31 | |
| acoles | rsFF: i can try to help, do you have a specific question? | 11:44 |
| *** panbalag has joined #openstack-swift | 11:47 | |
| rsFF | yes acoles, the output is somewhat like this:http://pastebin.com/938SSZHK | 11:49 |
| rsFF | Im not understanding what should be in a line | 11:49 |
| rsFF | cos I can grep each field of that output... | 11:50 |
| acoles | rsFF: if i understand the bug report from clayg the goal is to separate 'user metadata' from 'sysmeta' in the output | 11:52 |
| acoles | rsFF: see http://docs.openstack.org/developer/swift/development_middleware.html#swift-metadata for info on sysmeta | 11:52 |
| rsFF | hummmm ok, separate what is sysmeta vs meta... | 11:53 |
| acoles | rsFF: so print_obj_metadata would need to iterate over the metadata dict and parse keys to match either x-object-meta-* or x-object-sysmeta-* | 11:54 |
| acoles | rsFF: and group them separately in the output | 11:54 |
| rsFF | roger | 11:54 |
| rsFF | ty | 11:55 |
| acoles | rsFF: and, secondly, looks like clayg would like them listed one per line | 11:55 |
| acoles | rsFF: check out https://github.com/openstack/swift/blob/master/swift/common/request_helpers.py#L136-136 for useful helper methods | 11:56 |
| openstackgerrit | Denis Cavalcante proposed openstack/swift: Add storage policy support for sorting method https://review.openstack.org/160877 | 11:56 |
| acoles | rsFF: and note that a client request cannot set sysmeta (it is reserved for internal use and stripped from client requests by proxy server) so actually getting some sysmeta into an object .data file to test will require a request direct to the object server - ask if you need help | 11:59 |
| rsFF | yeah i would like that, also if you have any more tips on setting up and env for testing, i currently use vim and print for the debug process, is it possible to integrate some IDE with remote debugging? | 12:02 |
| *** SkyRocknRoll has joined #openstack-swift | 12:03 | |
| *** SkyRocknRoll has quit IRC | 12:03 | |
| *** SkyRocknRoll has joined #openstack-swift | 12:03 | |
| *** dmorita has quit IRC | 12:24 | |
| *** rdaly2 has joined #openstack-swift | 12:40 | |
| *** rdaly2 has quit IRC | 12:44 | |
| *** jistr has quit IRC | 12:47 | |
| *** jistr has joined #openstack-swift | 12:48 | |
| *** nellysmitt has quit IRC | 12:49 | |
| *** annegentle has joined #openstack-swift | 12:51 | |
| openstackgerrit | Yuan Zhou proposed openstack/swift: Fix EC PUT on HTTP_CONFLICT or HTTP_PRECONDITION_FAILED https://review.openstack.org/162047 | 12:53 |
| *** cdelatte has joined #openstack-swift | 13:01 | |
| *** bkopilov has quit IRC | 13:04 | |
| *** zul has quit IRC | 13:04 | |
| *** jistr has quit IRC | 13:04 | |
| *** jistr has joined #openstack-swift | 13:05 | |
| *** zul has joined #openstack-swift | 13:09 | |
| *** fifieldt has joined #openstack-swift | 13:14 | |
| *** bkopilov has joined #openstack-swift | 13:18 | |
| acoles | rsFF: this may help with creating some sysmeta in a .meta file to test against https://gist.github.com/alistairncoles/b7d67db06eda62d4bed2 | 13:19 |
| acoles | rsFF: or maybe swiftly https://github.com/gholt/swiftly | 13:20 |
| *** mahatic has joined #openstack-swift | 13:21 | |
| *** ppai has quit IRC | 13:27 | |
| openstackgerrit | Stuart McLaren proposed openstack/python-swiftclient: Allow reading from object body on download https://review.openstack.org/155291 | 13:33 |
| openstackgerrit | Elvis Teixeira proposed openstack/swift: Use assertTrue/False instead of assertEqual(*, True/False) https://review.openstack.org/163000 | 13:33 |
| *** mmcardle has quit IRC | 13:34 | |
| *** rdaly2 has joined #openstack-swift | 13:44 | |
| *** mmcardle has joined #openstack-swift | 13:48 | |
| *** annegentle has quit IRC | 13:51 | |
| *** lcurtis has joined #openstack-swift | 13:57 | |
| *** mahatic has quit IRC | 14:01 | |
| *** nshaikh has joined #openstack-swift | 14:03 | |
| *** rdaly2 has quit IRC | 14:03 | |
| notmyname | good morning | 14:07 |
| *** annegentle has joined #openstack-swift | 14:08 | |
| *** rdaly2 has joined #openstack-swift | 14:09 | |
| *** annegentle has quit IRC | 14:11 | |
| rsFF | acoles - thanks! | 14:11 |
| acoles | notmyname: morning! | 14:14 |
| *** mahatic has joined #openstack-swift | 14:15 | |
| *** SkyRocknRoll has quit IRC | 14:22 | |
| openstackgerrit | Elvis Teixeira proposed openstack/swift: Use assertTrue/False instead of assertEqual(*, True/False) https://review.openstack.org/163000 | 14:23 |
| openstackgerrit | Donagh McCabe proposed openstack/swift: Support HTTP_X_SERVICE_IDENTITY_STATUS in keystoneauth https://review.openstack.org/156634 | 14:25 |
| *** Guest75125 has joined #openstack-swift | 14:30 | |
| cschwede | Hello everyone! | 14:36 |
| acoles | cschwede: hi | 14:39 |
| cschwede | acoles: hello Alistair! | 14:41 |
| notmyname | guten tag | 14:43 |
| *** Guest75125 is now known as notsogentle | 14:43 | |
| *** notsogentle is now known as agentle_ | 14:43 | |
| *** nshaikh has quit IRC | 14:44 | |
| *** agentle_ has quit IRC | 15:05 | |
| *** david-lyle_afk is now known as david-lyle | 15:23 | |
| *** jrichli has joined #openstack-swift | 15:28 | |
| notmyname | reminder that there is no meeting tomorrow | 15:31 |
| *** thurloat has quit IRC | 15:35 | |
| *** swat30 has quit IRC | 15:35 | |
| *** delattec has joined #openstack-swift | 15:37 | |
| *** cdelatte has quit IRC | 15:39 | |
| *** swat30 has joined #openstack-swift | 15:39 | |
| *** thurloat has joined #openstack-swift | 15:40 | |
| *** gyee has joined #openstack-swift | 15:46 | |
| *** fifieldt has quit IRC | 15:57 | |
| *** bill_az has joined #openstack-swift | 16:01 | |
| *** bkopilov has quit IRC | 16:02 | |
| *** mahatic has quit IRC | 16:09 | |
| *** zhill_ has joined #openstack-swift | 16:09 | |
| *** agentle_ has joined #openstack-swift | 16:10 | |
| *** delatte has joined #openstack-swift | 16:10 | |
| *** delattec has quit IRC | 16:13 | |
| straycat | Hi | 16:13 |
| straycat | I'm running into trouble with swift, I've configured it to use v2 apis as per http://docs.openstack.org/juno/install-guide/install/apt/content/swift-install-controller-node.html but the logs tell me it's using v3, I haven't been able to find out why yet | 16:14 |
| *** homegrown_ has quit IRC | 16:18 | |
| acoles | straycat: what did you do to specifically configure v2 api? looks like that doc configures using keystone, but not a specific api version | 16:19 |
| straycat | acoles, I thought the api version was specified by the auth_uri, which in this case is 'auth_uri = http://controller:5000/v2.0' | 16:20 |
| straycat | If that's not the case how is the api version used determined? | 16:24 |
| acoles | straycat: its identity_uri that specifies the keystone service endpoint | 16:24 |
| acoles | straycat: auth_uri is a value that gets returned in www-authenticate header if request is denied | 16:25 |
| straycat | Ahh okay | 16:25 |
| acoles | straycat: the identity_uri api version is auto-discovered and these days defaults to v3 | 16:25 |
| straycat | So can/should I specify the api version in the identity_uri? | 16:26 |
| acoles | straycat: so auth_uri should be publically visible endpoint your client can go to to re-authorize, whereas identity_uri may only be visible to the swift service | 16:27 |
| * straycat nods | 16:27 | |
| acoles | straycat: err, i'm not sure about specifying api in identity_uri :/ if you did, you'd add either /v2.0 or /v3 | 16:28 |
| straycat | Right | 16:28 |
| acoles | straycat: but as i say afaik it now 'discovers' the latest available api and defaults to that (preferring v3 over v2.0) | 16:29 |
| straycat | Okay that might explain the problem I'm having, I'll check my keystone config out more carefully | 16:30 |
| acoles | straycat: i'll just try adding v2.0 to identity_uri and get back to you... | 16:31 |
| clayg | morning | 16:32 |
| straycat | acoles, wow thanks, i appreciate the help | 16:32 |
| clayg | acoles: evening | 16:32 |
| acoles | clayg: hi | 16:33 |
| clayg | yuanz: yay! 162890 copy fix! | 16:35 |
| * clayg runs off to run functests | 16:35 | |
| *** delattec has joined #openstack-swift | 16:35 | |
| *** mahatic has joined #openstack-swift | 16:36 | |
| *** delatte has quit IRC | 16:38 | |
| acoles | straycat: adding /v2.0 to identity_uril wasn't good. adding auth_version = v2.0 inf [filter:authtoken] seems to work and judging by keystone logs is making calls to v2.0 api. ymmv of course :) | 16:38 |
| *** rdaly2 has quit IRC | 16:39 | |
| acoles | straycat: for ref, config opts are here https://github.com/openstack/keystonemiddleware/blob/master/keystonemiddleware/auth_token/__init__.py#L240-240 . sorry to point you at code but in my experience is that unfortunately the authtoken middleware docs may not be up to date | 16:40 |
| straycat | acoles, no problem thanks for all your help :) | 16:40 |
| acoles | straycat: ok good luck | 16:43 |
| *** rdaly2 has joined #openstack-swift | 16:46 | |
| *** delatte has joined #openstack-swift | 16:51 | |
| *** delattec has quit IRC | 16:51 | |
| *** agentle_ has quit IRC | 16:58 | |
| acoles | clayg: been on other stuff so far today, not looked at the policy/policy-index patch(es) - but thanks for the comments | 16:59 |
| *** gyee has quit IRC | 17:00 | |
| *** jistr has quit IRC | 17:01 | |
| acoles | clayg: i kinda get you about int(policy) - big hint about the type | 17:01 |
| *** gyee has joined #openstack-swift | 17:02 | |
| *** thumpba has joined #openstack-swift | 17:03 | |
| *** thumpba_ has joined #openstack-swift | 17:07 | |
| *** gyee has quit IRC | 17:07 | |
| *** thumpba has quit IRC | 17:10 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Prefer policy to policy_index https://review.openstack.org/162775 | 17:20 |
| clayg | acoles: yeah no worries | 17:20 |
| acoles | clayg: i just started looking over then again | 17:21 |
| clayg | acoles: ok, well I'd like to chat soon without you about the per-policy-diskfile idea - obviously tons of functional stuff we still need to work out but i really appreciate your help and would love to also be helpful to you | 17:21 |
| clayg | s/without/with :P | 17:21 |
| torgomatic | holy scrollback, Batman | 17:21 |
| acoles | torgomatic: lol you hit the setup.py stacktrace i presume | 17:22 |
| torgomatic | acoles: yup | 17:22 |
| acoles | clay: ack. | 17:22 |
| acoles | clay lol - chatting *without* me could be the better way to go | 17:23 |
| clayg | no no no | 17:23 |
| *** zhill_ has quit IRC | 17:26 | |
| *** rledisez has quit IRC | 17:29 | |
| *** silor has quit IRC | 17:31 | |
| acoles | clayg: peluse: torgomatic: notmyname: about the .durable files...presumably we should be fsync'ing them? because without them we have no object. | 17:40 |
| torgomatic | makes sense to me | 17:40 |
| notmyname | +1 | 17:40 |
| acoles | in which case we would't want to go fsyncing the dir until both the data and durable are written, like if we were to fscync the dir ever | 17:41 |
| acoles | one dir fsync is bad enough, two would be profligate | 17:43 |
| *** Nadeem has joined #openstack-swift | 17:44 | |
| clayg | acoles: in an ideal world - sure | 17:45 |
| *** geaaru has quit IRC | 17:49 | |
| *** fbo has quit IRC | 18:01 | |
| *** tab_ has joined #openstack-swift | 18:01 | |
| *** rdaly2 has quit IRC | 18:03 | |
| *** tsg_ has quit IRC | 18:04 | |
| *** bkopilov has joined #openstack-swift | 18:05 | |
| *** rdaly2 has joined #openstack-swift | 18:10 | |
| *** dencaval has joined #openstack-swift | 18:15 | |
| *** gyee has joined #openstack-swift | 18:17 | |
| *** fbo has joined #openstack-swift | 18:25 | |
| acoles | clayg: ok i had a quick look but a bit pressed or time this evening - no problem at all with you taking/squashing 162696 into 162775 or whatever - if i don't have to go through the multi-FI patch checking for policy != None all over the place then I'm happy :) | 18:28 |
| acoles | clayg: and yeah lets chat on per-policy-diskfile soon (tomorrow?) but i gotta go now, sorry | 18:29 |
| straycat | It's odd that swift detects a v3 api, there's no v3 endpoint. | 18:33 |
| *** thumpba_ has quit IRC | 18:38 | |
| *** thumpba has joined #openstack-swift | 18:40 | |
| *** thumpba_ has joined #openstack-swift | 18:41 | |
| torgomatic | I'm a little confused about the direction some of the EC work is going; what's up with all the storage-policy knowledge getting into the object servers? | 18:42 |
| torgomatic | it started out being do-what-the-proxy-tells-you and now it's getting more smarts-in-the-object-server | 18:43 |
| *** thumpba has quit IRC | 18:45 | |
| *** panbalag has quit IRC | 18:45 | |
| *** panbalag has joined #openstack-swift | 18:45 | |
| clayg | acoles: tommorrow wfm | 18:46 |
| acoles | clayg: ok | 18:46 |
| *** agentle_ has joined #openstack-swift | 18:50 | |
| *** agentle_ has quit IRC | 18:55 | |
| *** agentle_ has joined #openstack-swift | 18:55 | |
| *** wshao has joined #openstack-swift | 19:00 | |
| *** wshao has left #openstack-swift | 19:02 | |
| *** wshao has joined #openstack-swift | 19:03 | |
| *** acoles is now known as acoles_away | 19:05 | |
| peluse | torgomatic, tell me about it! | 19:15 |
| *** erlon has joined #openstack-swift | 19:22 | |
| *** bill_az has quit IRC | 19:24 | |
| *** wshao has quit IRC | 19:35 | |
| *** wshao has joined #openstack-swift | 19:37 | |
| *** vishy has joined #openstack-swift | 19:38 | |
| *** wshao has quit IRC | 19:41 | |
| clayg | tdasilva: hey! you know what I want - anything I can do to help? | 19:42 |
| tdasilva | clayg: Hey! sorry, i've just been really busy with some other work | 19:42 |
| tdasilva | but getting back on obj. versioning today | 19:42 |
| clayg | tdasilva: it's no problem, i'm really not trying to rush - I'm really just looking for anything I can do to help - but if you don't have anything that I could be useful doing - I'm happy to wait - just let me know | 19:45 |
| tdasilva | clayg: I did have a question for you tho. Last friday we talked about that issue I raised about the Delete with ACLs and we decided to fix it so that users don't need any acls on the versioned container | 19:46 |
| tdasilva | but I was thinking that the current behavior could actually be used as a feature :-) | 19:46 |
| tdasilva | similar to what cschwede has proposed before | 19:47 |
| tdasilva | where users would be able to write to a container and objects get versioned and maybe never be deleted | 19:47 |
| tdasilva | without admin access | 19:47 |
| clayg | tdasilva: ok, yeah let me think about that | 19:48 |
| tdasilva | clayg: ok...feel free to tell it's a really dumb idea :P | 19:48 |
| clayg | because - yes - i totally agree there's an overlap with the undelete and versions - and I like the idea that an operator or account admin could give access to a container that is "safe" w/o having to give access to the backups/shadow-copies/versioned objects that are protecting it | 19:49 |
| tdasilva | right! | 19:49 |
| clayg | i just don't know if that requires the existing version middleware to not work in this state... | 19:50 |
| clayg | brb | 19:50 |
| tdasilva | ok | 19:50 |
| *** NM has joined #openstack-swift | 19:53 | |
| NM | Hello guys. | 19:55 |
| NM | Do you use any kind of script to test a storage node after an upgrade? I was thinking about writing a code do upload a file and then using swift-ring-builder check if everything is fine. | 19:55 |
| *** Tahmina has joined #openstack-swift | 19:55 | |
| *** aix has quit IRC | 19:57 | |
| *** mahatic has quit IRC | 19:59 | |
| *** dencaval has quit IRC | 20:12 | |
| clayg | NM: swift-dispersion-report sends out object writes to some % of the partition space (maybe up to 100%) - and it talks to nodes directly on the backend - be a good script for some other reasons and may cover some of the goals you're describing | 20:13 |
| clayg | tdasilva: I'm pretty much convinced myself that allowing zero-byte PUT but making DELETE return some sort of 40X is only the illusion of protection | 20:14 |
| tdasilva | well..not if the object gets versioned, no? | 20:14 |
| tdasilva | i'd agree with you if there was no versioning | 20:14 |
| clayg | tdasilva: if anything the lack of the x-version-location container listing calls into question the usefulness of the existing version implementation - there's no good way to do a restore | 20:15 |
| NM | clayg: That seens nice. I'll read about it. | 20:15 |
| clayg | tdasilva: well i just mean it doesn't prevent the source location for having data removed - but i guess i see what you mean about making it so they can't delete *any* versions of the object - maybe from like an audit perspective | 20:16 |
| clayg | tdasilva: I think it's a pretty janky use-case - but if it's existing behavior I suppose there's some change some client is expecting it | 20:16 |
| clayg | tdasilva: more likely if someone has a read/write acl'd container with a version location they're storing objects in it but don't realize they can't delete - i'd guess | 20:17 |
| tdasilva | clayg: Yeah, the thought just came to mind that with the current behavior we can actually meet different use cases | 20:19 |
| *** delatte has quit IRC | 20:19 | |
| tdasilva | 1. regular object versioning: jsut give read and write to both containers | 20:19 |
| tdasilva | 2. auditing: just give write source container | 20:19 |
| tdasilva | but I also did think of the fact that we might already be trying to provide a similar feature with undelete | 20:20 |
| tdasilva | so no need to fuss with it in object versioning | 20:20 |
| clayg | tdasilva: i dont know - you might have a case | 20:22 |
| clayg | tdasilva: if it didn't *already* work this way I think i'd be pretty skeptical - but justifying any change in behavior needs to be pretty solid - http://xkcd.com/1172/ | 20:24 |
| tdasilva | rofl..do you keep a log of these links somewhere? you always seem to have the perfect analogy for the discussed topic | 20:25 |
| *** cdelatte has joined #openstack-swift | 20:41 | |
| clayg | google just finds them for me | 20:46 |
| mattoliverau | Morning | 21:13 |
| *** Tahmina has quit IRC | 21:17 | |
| *** rdaly2 has quit IRC | 21:26 | |
| tdasilva | mattoliverau: hi! how's the long weekend? | 21:30 |
| *** rdaly2 has joined #openstack-swift | 21:31 | |
| mattoliverau | tdasilva: nice, my parents were visiting so was great catching up :) nice work on all the EC stuff you've been working on, I for one have been impressed :) | 21:32 |
| mattoliverau | And an extra day of is always nice ;) | 21:33 |
| tdasilva | mattoliverau: yes, definitely | 21:33 |
| tdasilva | mattoliverau: looks like there's a a thon of activity with EC in the past few days so I need to catch up with reviews | 21:34 |
| mattoliverau | You and me both :) | 21:36 |
| *** tsg has joined #openstack-swift | 21:39 | |
| *** chlong has joined #openstack-swift | 21:40 | |
| *** NM has quit IRC | 21:48 | |
| clayg | tdasilva: it would be a huge contribution to the EC effort to get get versioned objects extracted to middleware - boarderline superhuman if you manage to eek out COPY as well | 21:57 |
| tdasilva | clayg: I think I found a couple of issues with that ACL scheme that makes it not work as a good auditing tool. 1. If user writes just one version of the object, then he will be able to delete it, because we only version on the 2nd PUT. 2. If only read access is given to x-versions-location, then on a DELETE system will copy previous version to source container, but not DELETE the object in x-versions-location | 21:58 |
| clayg | torgomatic: did you see yuanz's fix for COPY - this controller routing stuff has teeth :\ (https://review.openstack.org/#/c/162890/) | 21:58 |
| tdasilva | clayg: have you seen this: https://review.openstack.org/#/c/156923/ ? | 21:59 |
| clayg | tdasilva: idk, I thought we do the container listing before the delete - won't it blow up early enough that DELETE's just wont work at all w/o read access to the target container | 21:59 |
| clayg | tdasilva: i haven't looked at it much - it seemed not quite finished | 21:59 |
| clayg | tdasilva: have you tried it? you think it'll work? | 22:00 |
| tdasilva | no, it is not...was just pointing out that the work has started | 22:00 |
| tdasilva | planning to help ppai with it | 22:00 |
| clayg | versions first - I'd be happy to help as well - if it seems manageable it's in front of the ECObjectController PUT refactor | 22:00 |
| tdasilva | yep | 22:01 |
| clayg | notmyname: I took a stab at making loadapp work with a default config string | 22:01 |
| clayg | notmyname: it would... "work" - but... well I don't see any better options :'( | 22:01 |
| clayg | notmyname: I mean upstream patch obviously - we should get config_dir upstream too because it would make an olso migration much more palitable if paste supported loading from config dir's naitively | 22:02 |
| tdasilva | clayg: so back to versions: yes, the container listing is done before the DELETE, but the COPY as well...so if an admin gave the user read (but not write) then they would be able to do listing and even the copy, but not the final delete, which i think leaves it in a weird state where you overwrite the object in the source container, but did not delete it from the x-versions-location | 22:03 |
| *** tsg has quit IRC | 22:03 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Trick paste.deploy's loadapp into reading a config from a string https://review.openstack.org/163209 | 22:05 |
| clayg | ^ defintely the worst hack evar - anyone that needs a laugh should go check it out | 22:06 |
| clayg | tdasilva: oic, so DELETE, DELETE, DELETE just keeps copying the same object over - but you get an error back right? | 22:07 |
| *** chlong has quit IRC | 22:07 | |
| tdasilva | yes | 22:07 |
| *** agentle_ has quit IRC | 22:07 | |
| tdasilva | and I think point 1 above is really bad too | 22:07 |
| tdasilva | at least to claim this as a feature for auditing purposes | 22:08 |
| clayg | tdasilva: anyway I think that behavior sucks - and at lest with read access to the container you can do some sort of sane restore workflow | 22:08 |
| clayg | tdasilva: oh, if they have read access to the version location the listing will pass, turn up empty and remove the object without ever having a copy in the version location where it's safe - got it | 22:09 |
| clayg | yeah I think the new rev of versions that works more like shadow copy will cover this usecase better | 22:10 |
| clayg | it'll be great to have in middleware - easier to experiment with different workflows and use-cases | 22:10 |
| clayg | so yeah point 1 kills the audit workflow - it's amazing it almost worked as well as it did since that was obviously not the intention | 22:11 |
| clayg | tdasilva: but if you don't give read or write access to the version location that does efectively implement no data is ever deleted - only moved when you PUT | 22:11 |
| clayg | but it's so wonky I don't think it's worth trying to maintain | 22:12 |
| *** rdaly2 has quit IRC | 22:12 | |
| tdasilva | true and true | 22:12 |
| clayg | let's assume anyone with a version location that doesn't have read/write is broken, so we're back to either fixing PUT or fixing DELETE - and I think we honestly could surprise someone with 40X if we make PUT not work until the ACL the version location | 22:13 |
| clayg | i'm as skeptical as ever that someone would be surprised if DELETE started "working" - but I guess it's possible - fuuuuuu | 22:14 |
| tdasilva | yeah..i was thinking of just fixing DELETE to use preauth, then I think we just need to document it really well | 22:15 |
| clayg | of course at least if we fix PUT they have a way to fix it - assuming they weren't relying somehow on the fact that a user can write to the source but not the version location | 22:15 |
| clayg | it's such a non feature to begin with - if you don't have read access to the version location you can't do a restore - if we allow DELETE to pop entries out of the version location you basically have write - so it's sorta crazy to maintain the preauthed behavior | 22:17 |
| clayg | i feel like fixing PUT might somehow be more defensible - i'm pretty sure we were all surprised it worked at all | 22:18 |
| tdasilva | do you mean "fix PUT" to be something like return 40X if user doesn't have read/write to versions-location? | 22:19 |
| clayg | tdasilva: anyway, fix it whichever way it's easier to code - but leaving the current behavior I think is wrong | 22:19 |
| clayg | tdasilva: yeah you can't PUT to a versioned container unless you have write access to the version location, you can't DELETE from a versioned container unless you have read access to the version location - something like that would make perfect sense I think | 22:20 |
| clayg | it's not a security audit feature - I'm not really sure how the write path managed to sneak past an authorization check on the version location container really | 22:20 |
| *** agentle_ has joined #openstack-swift | 22:25 | |
| tdasilva | clayg: sorry, but just thought of something else: how big of a concern or how much do you think ops rely on having a single versions-location container for multiple containers. The reason I ask is because we have code specifically to handle that scenario and my concern is that if ops like that feature, then they might not be happy with giving read access to x-versions-location. If this is a true concern, then we might | 22:36 |
| tdasilva | need to fix DELETE instead | 22:36 |
| *** jrichli has quit IRC | 22:41 | |
| *** agentle_ has quit IRC | 22:47 | |
| clayg | tdasilva: well version location isn't | 22:49 |
| clayg | cross account | 22:49 |
| clayg | so it'd be multiple containers in the account - so id' think an admin/swift_owner still has read access | 22:50 |
| clayg | because of DELETE and restore being broken I think *maybe* someone doesn't have *write* access to version location - but they probably have *read* - maybe? | 22:50 |
| *** lcurtis has quit IRC | 22:52 | |
| tdasilva | clayg: I was concerned for the case of one account and multiple "regular" users, with each just having access to their own container | 22:52 |
| tdasilva | clayg: is that a scenario you have seen before? | 22:53 |
| clayg | multiple non-swift-owner users with read/write to subset of containers sure - but I'm not sure if any of those containers would have an x-version-location to a container that they don't have read access too | 22:54 |
| clayg | tdasilva: i'm spinning up master now to play around with current behavior | 22:54 |
| tdasilva | ok...i'm sending up a new patchset for you to take a look and we can adjust accordingly | 22:54 |
| tdasilva | sorry if it seems like we are just spinning our wheels but new use cases just keep coming up | 22:55 |
| clayg | tdasilva: isn't having multiple containers pointing to the same x-versions-location broken because of the duplicate object name issue? like the little 3 char hash maybe only uses the source object name - not the source container/object name? | 22:58 |
| tdasilva | yes, but I think that was done in the first place to handle that exact scenario, it seems like it was just done with a bug :) | 22:59 |
| clayg | oh weird - it's just a length of the object name - i have no idea why that's there - notmyname? | 23:02 |
| *** erlon has quit IRC | 23:11 | |
| openstackgerrit | Thiago da Silva proposed openstack/swift: versioned writes middleware https://review.openstack.org/134347 | 23:13 |
| clayg | man it's really hard to reason about which object is where because there's no x-timestamp mangling - it's just the time when the copy happened nothing referencing the original objects life cycle | 23:24 |
| clayg | with read access to the x-versions-location DELETE is quite confusing indeed | 23:25 |
| clayg | write access to the x-version-location doesn't seem to have any effect on the behavior | 23:26 |
| clayg | tdasilva: i'm seeing no read access to x-versions-location disables all deletes (even the first one if there's no old versions) | 23:30 |
| clayg | tdasilva: so no read access to x-versions-location is a possible audit in that you can upload objects forever and never delete them (but you also can't restore old versions :\) | 23:31 |
| tdasilva | clayg: i see, because that would return an error right away | 23:32 |
| tdasilva | as the listing fails | 23:33 |
| *** david-lyle is now known as david-lyle_afk | 23:33 | |
| clayg | tdasilva: yeah | 23:40 |
| clayg | tdasilva: you know I'm not sure the ability to write to the x-versions-location isn't a regression... I don't *really* see the code that is allowing it - when PUT constructs the COPY I think it's expecting the PUT to the x-version-location to call authorize... | 23:41 |
| *** tab_ has quit IRC | 23:44 | |
| tdasilva | but that COPY request does not container "swift.authorize", right? | 23:45 |
| tdasilva | clayg: ^https://github.com/openstack/swift/blob/master/swift/proxy/controllers/obj.py#L541 | 23:45 |
| clayg | tdasilva: yeah i just confirmed that it does not | 23:46 |
| *** chlong has joined #openstack-swift | 23:54 | |
| clayg | tdasilva: so if i was going to try and fix master it'd be much easier to use make_subrequest when building the copy_req I think than it would be to add preauth requests to DELETE's listing and COPY requests | 23:55 |
| *** gyee has quit IRC | 23:55 | |
| tdasilva | you mean use make_subrequest on the PUT, right? | 23:56 |
| clayg | what'd I say? | 23:56 |
| *** dmorita has joined #openstack-swift | 23:56 | |
| clayg | yeah I think the line you linked predates common.wsgi.make_subrequest - if that code was being written today i'd be more obvious to use that and let the ACL's work as expected | 23:57 |
| tdasilva | yes, got it! | 23:58 |
| clayg | I mean the comment that retuns client_error even says "or bad permissions" | 23:58 |
| tdasilva | true true | 23:58 |
| clayg | i'm not saying that means we have a more obvious path forward | 23:58 |
| tdasilva | well...i think we have determined that either way it sounds like we need to change some behavior | 23:59 |
| clayg | unless by some mircale the ability to write to the x-version-location is a regression - then we can totally just say we're fixing a regression in the name of security! | 23:59 |
| clayg | tdasilva: yeah current behavior is the worst | 23:59 |
| *** zul has quit IRC | 23:59 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!