| notmyname | akhivin: what's the error you got? | 00:01 |
|---|---|---|
| akhivin | notmyname, AssertionError: 400 != 204 | 00:01 |
| *** abhirc has quit IRC | 00:02 | |
| akhivin | self.assertEquals(resp.status_int, 204) | 00:02 |
| zaitcev | You don't happen a Swift installed on that box? | 00:04 |
| *** joeljwright has quit IRC | 00:05 | |
| zaitcev | If you do, and follow tutorial, it's going to test the installed Swift with in-tree tests, and may end with funny results. | 00:05 |
| zaitcev | My solution is to run "python setup.py build", then use PYTHONPATH=$(pwd) ./.unittests | 00:05 |
| akhivin | zaitcev, it is a clean virtual machine installed for this tutorial | 00:06 |
| notmyname | akhivin: tests just passed for me on my saio | 00:10 |
| *** jrichli_ has quit IRC | 00:11 | |
| zaitcev | okay. There's no class TestAuthMultiplePrefixes in test/unit/common/middleware/test_tempauth.py. | 00:12 |
| zaitcev | akhivin: Where did you get that "Swift"? | 00:12 |
| zaitcev | akhivin: or even better, paste the exact clone command, I'm curious. | 00:12 |
| *** gyee has joined #openstack-swift | 00:13 | |
| zaitcev | could be some kind of Mirantis or other hacked-up thing with broken tests | 00:13 |
| *** cpallares has quit IRC | 00:13 | |
| *** david-lyle is now known as david-lyle_afk | 00:14 | |
| notmyname | zaitcev: there is :-) | 00:14 |
| akhivin | zaitcev, no. no Mirantis hacks. just git clone https://github.com/openstack/swift.git | 00:15 |
| notmyname | zaitcev: just recently added by acoles_away donagh | 00:15 |
| zaitcev | okay, go go gadget git pull | 00:15 |
| zaitcev | Hmm. Now this works (cd test/unit && PYTHONPATH=/q/zaitcev/hail/swift-tip nosetests common/middleware/test_tempauth.py:TestAuthMultiplePrefixes.test_acls_propagate_to_sysmeta) | 00:16 |
| zaitcev | akhivin: sorry, seems to work fine here. | 00:16 |
| zaitcev | I wonder if we accidentially created a test that blows if $HOME does not support xattr. | 00:16 |
| *** Manshoon has joined #openstack-swift | 00:17 | |
| *** Manshoon has quit IRC | 00:18 | |
| *** dmorita has joined #openstack-swift | 00:25 | |
| akhivin | zaitcev, notmyname thanks for your help. Still fails. Will try to understand. May be I doing something wrong | 00:28 |
| zaitcev | well, what does getfattr -d setup.py say? | 00:29 |
| zaitcev | actually, n/m | 00:30 |
| zaitcev | it seems working on everything nowadays | 00:30 |
| notmyname | akhivin: sorry, been a little distracted. now I can focus on this | 00:30 |
| notmyname | akhivin: can you run the test by itself and pastebin the output? | 00:31 |
| notmyname | akhivin: eg https://gist.github.com/notmyname/53a6fd6b17e76d4f16e3 | 00:32 |
| akhivin | notmyname, yes. just a moment | 00:32 |
| akhivin | http://paste.openstack.org/show/182185/ | 00:35 |
| akhivin | notmyname, http://paste.openstack.org/show/182186/ | 00:38 |
| notmyname | akhivin: we're looking | 00:43 |
| akhivin | http://paste.openstack.org/show/182192/ is four failed | 00:43 |
| akhivin | this four | 00:43 |
| notmyname | what version of python are you running? | 00:43 |
| akhivin | Python 2.7.6 (default, Mar 22 2014, 22:59:56) | 00:45 |
| akhivin | [GCC 4.8.2] on linux2 | 00:45 |
| notmyname | akhivin: I honestly have no idea | 00:45 |
| zaitcev | akhivin: sorry if I step of notmyname's toes, but could you add "print resp.status" in front of that self.assertEquals(resp.status_int, 204) and run nosetests -s | 00:45 |
| openstackgerrit | Samuel Merritt proposed openstack/swift: EC: Multi-phase commit conversation for erasure-coded PUTs https://review.openstack.org/153843 | 00:45 |
| openstackgerrit | Samuel Merritt proposed openstack/swift: Bump eventlet version to 0.16.1 https://review.openstack.org/145403 | 00:45 |
| * notmyname tags zaitcev | 00:45 | |
| zaitcev | but | 00:45 |
| zaitcev | I need to step aside for a moment | 00:46 |
| torgomatic | tsg: okay, I think I got the multistep PUT thing all pieced together with the GET | 00:46 |
| torgomatic | mostly it was a matter of eliminating self._quorum_size as a method on the various controllers and just plumbing the right thing through at the right times | 00:47 |
| notmyname | torgomatic: did you see that eventlet 0.17[.1] was just released? check the changelog http://eventlet.net/doc/changelog.html | 00:47 |
| torgomatic | there is no one-size-fits-all quorum size for EC; it really depends on what you're doing | 00:47 |
| torgomatic | notmyname: yes; I'm looking forward to working on py3 compatibility *after* EC | 00:47 |
| notmyname | :-) | 00:47 |
| notmyname | also the ipv6 support is nice | 00:48 |
| torgomatic | my brain is not big enough to hold both; heck, it's barely big enough to hold EC in it | 00:48 |
| *** acoles_away has quit IRC | 00:51 | |
| notmyname | akhivin: what time zone are you in? | 00:51 |
| *** kota_ has joined #openstack-swift | 00:52 | |
| akhivin | notmyname, utc +3. 4am at this moment | 00:52 |
| notmyname | yikes :-) | 00:52 |
| tsg | torgomatic: I was getting ready to post an update to the multiphase patch - this is last night's version https://github.com/tsg-/swift-ec/commits/multiphase_ec_put, and I have some obj server tests added | 00:53 |
| torgomatic | tsg: nice; if I've goofed anything up in my rebase of your patch, please throw it away | 00:54 |
| * torgomatic is not always sure what's going on | 00:54 | |
| notmyname | akhivin: actually, that's good for what I'm about to say (assuming you don't sleep all during daylight hours). that's a very recent set of tests that you're seeing fail. which is in part why we're looking at it and not seeing an obvious problem. however, the author is in GMT+0 and may be able to help more | 00:54 |
| notmyname | akhivin: look for acoles | 00:54 |
| tsg | torgomatic: I am looking :) didn't see the update earlier - was trying to sort out an openstack-infra thing (getting liberasurecode in there) | 00:55 |
| torgomatic | tsg: it wasn't there earlier; I just pushed it 10 minutes ago :) | 00:55 |
| notmyname | ok, I'm out | 00:55 |
| notmyname | might be on later tonight. if not, then tomorrow | 00:56 |
| tsg | torgomatic: thank you - I will take a closer look at the patch .. a question for you on Putter in the meantime | 00:56 |
| torgomatic | sure | 00:56 |
| mattoliverau | Yay so eventlet finally has ipv6! | 00:56 |
| akhivin | zaitcev, http://paste.openstack.org/show/182194/ | 00:56 |
| tsg | in Putter.await_response(), we are checking "if not conn.resp" - I was wondering about the logic there | 00:56 |
| akhivin | zaitcev, printed 400 Bad Request | 00:57 |
| tsg | torgomatic: in a multiphase PUT, I ended up having to set the resp to None specifically to hit that condition (to read the 2nd phase response) | 00:57 |
| torgomatic | tsg: I think the idea was that conn.resp would be set only when the "final" response code (>= 200) came down the wire | 00:58 |
| torgomatic | I think if you call conn.getresponse(), it might set conn.resp as a side effect... maybe? | 00:58 |
| torgomatic | it's sort of oddly written right now | 00:59 |
| tsg | torgomatic: Gotcha. Do you think it is reasonable to set it to None specifically in get_put_responses(), in the 2-phase case? | 00:59 |
| tsg | torgomatic: ack :) I have been wondering how to best do it | 01:00 |
| torgomatic | but the idea was to call await_response() and that would read the next one off the wire, so you'd have current_status = 100, then await_response, then current_status = 100 (again), then await_response, then current_status = 200 | 01:00 |
| torgomatic | or 201 or whatever the heck it is | 01:00 |
| zaitcev | akhivin: what about resp.body, then | 01:00 |
| torgomatic | so I'd take await_response and s/conn.resp/self.resp/g in it | 01:01 |
| tsg | torgomatic: that makes sense - I noticed that we were keeping the self.resp.status around but I never saw it being used :) | 01:03 |
| torgomatic | tsg: I don't know if that'll actually work or not, but that's sort of the flow I imagined | 01:03 |
| tsg | (ie current_status() being called) | 01:03 |
| torgomatic | yeah, the battle plan did not survive contact with the enemy | 01:03 |
| *** comay has joined #openstack-swift | 01:03 | |
| akhivin | zaitcev, http://paste.openstack.org/show/182197/ | 01:03 |
| tsg | torgomatic: let me see if I can put self.resp to use | 01:04 |
| tsg | should be quick to try | 01:04 |
| torgomatic | zaitcev: simplejson strikes again | 01:05 |
| zaitcev | torgomatic: ouch | 01:05 |
| comay | i was curious about swift 2.2.1 or 2.2.2 - as part of the kilo development, are they only tested with the latter or is it expected they can/should be used on prior releases such as juno? | 01:05 |
| torgomatic | zaitcev: tempauth.py line 458 is what's bad... it's asserting that the values in that list are strings | 01:06 |
| torgomatic | which makes sense; you don't want them to be lists or hashes or anything | 01:06 |
| torgomatic | except that these are unicodes, which happens when either (a) you do not have simplejson installed, or (b) you do have simplejson installed but without the C bindings | 01:07 |
| zaitcev | oh. I have python-simplejson-3.5.3-2.fc21.x86_64 | 01:07 |
| torgomatic | so basically tempauth's acl validator doesn't work unless you are using simplejson *and* the C bindings | 01:07 |
| * torgomatic really really does not like simplejson | 01:08 | |
| zaitcev | so... how are we supposed to fix it? There's no way for akhivin to guess stuff like that. It should be in requirements.txt... Lessee | 01:09 |
| zaitcev | simplejson>=2.0.9 | 01:10 |
| torgomatic | zaitcev: remove simplejson from Swift; then decoded JSON strings are Python unicodes everywhere and always, so we don't have bugs like this | 01:10 |
| zaitcev | or... Migrate to py3 :-) | 01:11 |
| torgomatic | or that :) | 01:11 |
| torgomatic | tsg: I did this http://paste.openstack.org/show/PTpF6kfaIbjqqjRDxiqs/ and it didn't seem to break anything | 01:12 |
| torgomatic | of course I didn't add any callers of current_status, but I did get rid of that nasty thing you saw | 01:13 |
| akhivin | zaitcev, it works now | 01:13 |
| akhivin | i removed simplejson 3.6.5 and installed 3.5.3 | 01:13 |
| *** Anticimex has quit IRC | 01:14 | |
| *** Anticimex has joined #openstack-swift | 01:14 | |
| zaitcev | I presume you're not on Fedora, then, because we only have 3.5.3. | 01:17 |
| *** cpallares has joined #openstack-swift | 01:17 | |
| akhivin | so should I report a bug on launchpad? | 01:17 |
| zaitcev | akhivin: thanks for bringing this to my attention... | 01:17 |
| zaitcev | akhivin: well, you probably should. Just state the facts and then ask if they forgot C bindings. | 01:17 |
| *** gyee has quit IRC | 01:18 | |
| zaitcev | and/or file a bug against swift to migrate away from simplejson... I'm just not sure that's going to be resolved quickly. | 01:18 |
| akhivin | zaitcev, i really not understood about c bindings but requirements contans simplejson>=2.0.9, but may be <3.6 should be added? | 01:21 |
| zaitcev | I suspect we'll add a few .encode() eventually. Maybe. | 01:24 |
| torgomatic | I think the only reason we've still got simplejson is because it's significantly faster on py2.6 | 01:27 |
| tsg | torgomatic: yep, it does seem to work if we store resp in just one place :) thx | 01:28 |
| torgomatic | tsg: np :) | 01:29 |
| * torgomatic is out for the day | 01:29 | |
| *** zhill has quit IRC | 01:36 | |
| akhivin | zaitcev, notmyname https://bugs.launchpad.net/swift/+bug/1425776 | 01:42 |
| openstack | Launchpad bug 1425776 in OpenStack Object Storage (swift) "ACL tests fails with simplejson 3.6.5 " [Undecided,New] | 01:42 |
| notmyname | akhivin: thanks | 01:42 |
| tsg | torgomatic: still around? | 01:42 |
| akhivin | zaitcev, notmyname thanks for your help | 01:43 |
| akhivin | away | 01:43 |
| *** yuan has joined #openstack-swift | 01:59 | |
| *** jrichli_ has joined #openstack-swift | 02:03 | |
| *** kota_ has quit IRC | 02:07 | |
| *** bill_az has quit IRC | 02:14 | |
| *** cpallares has quit IRC | 02:16 | |
| *** SkyRocknRoll has joined #openstack-swift | 02:27 | |
| *** panbalag has quit IRC | 02:29 | |
| *** tsg has quit IRC | 02:33 | |
| *** chlong_ has joined #openstack-swift | 02:43 | |
| *** zhill has joined #openstack-swift | 02:47 | |
| *** zhill has quit IRC | 02:51 | |
| *** zhill has joined #openstack-swift | 03:03 | |
| *** zhill has quit IRC | 03:05 | |
| *** doxavore has joined #openstack-swift | 03:13 | |
| doxavore | is there a recommended way to get timing/performance information re what the proxy node is spending its time on? object storage nodes are reporting 1ms mean times for GET requests and the proxies are ~60ms (according to the statsd reporting) | 03:15 |
| doxavore | i already run account/container DBs on SSDs, and the 3 proxies all run memcached (in a cluster), but if there is something I can be doing to see if there's something in particular it's spending the bulk of that time on... | 03:17 |
| *** Anticimex has quit IRC | 03:24 | |
| *** Anticimex has joined #openstack-swift | 03:25 | |
| *** SkyRocknRoll has quit IRC | 03:26 | |
| *** bkopilov has quit IRC | 03:36 | |
| *** SkyRocknRoll has joined #openstack-swift | 03:39 | |
| *** cpallares has joined #openstack-swift | 03:50 | |
| *** tsg has joined #openstack-swift | 03:57 | |
| *** silor has joined #openstack-swift | 04:06 | |
| *** ho has joined #openstack-swift | 04:48 | |
| *** bkopilov has joined #openstack-swift | 04:56 | |
| *** zaitcev has quit IRC | 05:12 | |
| *** cpallares has quit IRC | 05:24 | |
| *** SkyRocknRoll has quit IRC | 05:26 | |
| *** doxavore has quit IRC | 05:27 | |
| *** SkyRocknRoll has joined #openstack-swift | 05:31 | |
| *** echevemaster has quit IRC | 05:38 | |
| *** jrichli_ has quit IRC | 05:40 | |
| *** comay has quit IRC | 05:47 | |
| *** wasmum has quit IRC | 05:49 | |
| *** ppai has joined #openstack-swift | 06:01 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Imported Translations from Transifex https://review.openstack.org/159012 | 06:10 |
| *** fifieldt has joined #openstack-swift | 06:26 | |
| *** mmcardle has quit IRC | 06:36 | |
| openstackgerrit | Prashanth Pai proposed openstack/swift: fsync() on directories https://review.openstack.org/126923 | 06:37 |
| *** pck has quit IRC | 06:37 | |
| openstackgerrit | Merged openstack/swift: Enable override of container update headers https://review.openstack.org/158821 | 07:09 |
| *** chlong_ has quit IRC | 07:42 | |
| *** chlong has quit IRC | 07:43 | |
| *** nshaikh has joined #openstack-swift | 08:23 | |
| *** rledisez has joined #openstack-swift | 08:24 | |
| *** geaaru has joined #openstack-swift | 08:34 | |
| *** jistr has joined #openstack-swift | 08:53 | |
| *** jordanP has joined #openstack-swift | 08:54 | |
| *** MasterPiece has quit IRC | 09:10 | |
| *** acoles_away has joined #openstack-swift | 09:13 | |
| *** acoles_away is now known as acoles | 09:14 | |
| *** ChanServ sets mode: +v acoles | 09:14 | |
| *** silor has quit IRC | 09:49 | |
| *** mmcardle has joined #openstack-swift | 09:57 | |
| *** arnaud_o has quit IRC | 10:03 | |
| *** arnaud_o has joined #openstack-swift | 10:04 | |
| *** sluo_laptop has quit IRC | 10:05 | |
| *** EmilienM|afk is now known as EmilienM | 10:11 | |
| *** mmcardle has quit IRC | 10:13 | |
| akhivin | binding | 10:22 |
| *** mmcardle has joined #openstack-swift | 10:35 | |
| *** nshaikh has quit IRC | 10:38 | |
| akhivin | acoles, hello! | 10:40 |
| akhivin | acoles, it seems i understand why tests fail | 10:40 |
| akhivin | acoles, https://bugs.launchpad.net/swift/+bug/1425776/ | 10:41 |
| openstack | Launchpad bug 1425776 in OpenStack Object Storage (swift) "ACL tests fails with simplejson 3.6.5 " [Undecided,New] | 10:41 |
| akhivin | acoles, is it ok? | 10:41 |
| acoles | akhivin: thanks for updating the bug report, glad you found a solution | 10:44 |
| acoles | akhivin: so did you need to take any special steps to have the c bindings installed on 14.04? - I don't remember doing anything special on my 14.04 machine | 10:45 |
| akhivin | acoles, just apt-get install | 10:46 |
| akhivin | acoles, but default package has no bindings compiled | 10:46 |
| akhivin | acoles, so It should previously be removed | 10:47 |
| acoles | akhivin: i see. pip install simplejson also builds the speedups extension | 10:50 |
| acoles | curious that the tox venv did not have it - did you run the tests with tox or directly with nosetests? | 10:50 |
| akhivin | acoles, yes, if there is no previously installed simplejson | 10:51 |
| akhivin | acoles, but it seems in Trusty simplejson installed is by default without the speedup extension | 10:53 |
| akhivin | acoles, so pip install just said "already satisfied" | 10:56 |
| acoles | akhivin: ok. understood. so we may need to add something to docs | 10:58 |
| akhivin | acoles, so it should be "pip install --upgrade" or something like that | 10:58 |
| acoles | akhivin: i'll build a fresh trusty to try to reproduce. | 10:59 |
| acoles | akhivin: have to go to a meeting now. thanks for the info. | 10:59 |
| *** nellysmitt has joined #openstack-swift | 11:10 | |
| *** pcaruana has joined #openstack-swift | 11:19 | |
| *** SkyRocknRoll has quit IRC | 11:22 | |
| *** SkyRocknRoll has joined #openstack-swift | 11:34 | |
| *** chlong has joined #openstack-swift | 11:50 | |
| *** chlong_ has joined #openstack-swift | 11:50 | |
| *** bill_az has joined #openstack-swift | 11:54 | |
| openstackgerrit | Donagh McCabe proposed openstack/swift-specs: Composite Tokens and Service Accounts are done https://review.openstack.org/159425 | 11:55 |
| openstackgerrit | Alistair Coles proposed openstack/swift-specs: Composite Tokens and Service Accounts are done https://review.openstack.org/159425 | 12:00 |
| acoles | donagh: ^^ i just added a link to the commit message | 12:01 |
| *** dmorita has quit IRC | 12:02 | |
| donagh | acoles: Good idea. | 12:02 |
| *** erlon has joined #openstack-swift | 12:10 | |
| *** SkyRocknRoll has quit IRC | 12:11 | |
| *** ppai has quit IRC | 12:14 | |
| *** panbalag has joined #openstack-swift | 12:17 | |
| *** panbalag has quit IRC | 12:17 | |
| *** panbalag has joined #openstack-swift | 12:22 | |
| *** SkyRocknRoll has joined #openstack-swift | 12:24 | |
| *** ppai has joined #openstack-swift | 12:28 | |
| *** annegentle has quit IRC | 12:33 | |
| *** ho has quit IRC | 12:48 | |
| *** km has quit IRC | 12:49 | |
| *** mmcardle has quit IRC | 12:58 | |
| *** aix has joined #openstack-swift | 13:13 | |
| *** bkopilov has quit IRC | 13:17 | |
| *** tsg has quit IRC | 13:19 | |
| *** chlong has quit IRC | 13:20 | |
| *** chlong_ has quit IRC | 13:20 | |
| *** ppai has quit IRC | 13:26 | |
| cschwede | akhivin: i was able to reproduce your bug https://bugs.launchpad.net/swift/+bug/1425776 and wrote a quick fix for this, do you mind if i submit it for review? | 13:43 |
| openstack | Launchpad bug 1425776 in OpenStack Object Storage (swift) "ACL tests fails on fresh Ubuntu 14.04" [Undecided,Confirmed] - Assigned to Christian Schwede (cschwede) | 13:43 |
| openstackgerrit | Alistair Coles proposed openstack/swift: Enable in-process functional test policy to be configured https://review.openstack.org/159205 | 13:45 |
| acoles | peluse: clayg: i improved this ^^ to allow any policy in a swift.conf to be specified for running in-process tests | 13:46 |
| *** mahatic has joined #openstack-swift | 13:50 | |
| akhivin | cschwede, I would happy if this thing will be fixed. As I understand, tests affected by this issue are written by acoles. So may be it would be better to talk with him how to fix this. Because I still dont understand why tests failed without _speedup extension | 13:58 |
| *** tsg has joined #openstack-swift | 14:00 | |
| *** annegentle has joined #openstack-swift | 14:02 | |
| cschwede | acoles: are you working on a bug-fix for the failing tests if the simplejson speedups extension is missing? if not i push a quick fix using skiptest (which works for me) | 14:05 |
| acoles | cschwede: no i've not had time yet. thanks. | 14:06 |
| cschwede | ok | 14:06 |
| akhivin | cschwede, do you think that affected tests should be skipped if extension is absent? | 14:09 |
| cschwede | akhivin: yes, we already do this in some other tests, for example https://github.com/openstack/swift/blob/master/test/unit/common/middleware/test_cname_lookup.py#L21 | 14:13 |
| cschwede | akhivin: i add a skiptest with an explanation why this is skipped, thus a user can upgrade the simplejson pkg | 14:14 |
| openstackgerrit | Christian Schwede proposed openstack/swift: Skip failing tests if simplejson speedups extension is missing https://review.openstack.org/159478 | 14:14 |
| cschwede | akhivin: but at the same time i try to find the underlying reason for this, and see if i can fix that one too | 14:15 |
| *** mmcardle has joined #openstack-swift | 14:18 | |
| *** doxavore has joined #openstack-swift | 14:21 | |
| openstackgerrit | Alistair Coles proposed openstack/python-swiftclient: Add improvements to MD5 validation. https://review.openstack.org/156677 | 14:27 |
| tdasilva | peluse: hello, just noticed that acoles patch got merged, once that makes its way to feature/ec branch I'll submit the patch for etag/size container update | 14:27 |
| *** lcurtis has joined #openstack-swift | 14:31 | |
| acoles | cschwede: i'm just looking at the tempauth tests | 14:34 |
| cschwede | acoles: me too, trying to dig into simplejson and find the speedup calls | 14:40 |
| *** j_king_ is now known as j_king | 14:43 | |
| lcurtis | hello all...im trying to get around the containers object limitation by creating thousands of containers. However, I will then have a need to search for objects, which for me will eventually scale into 10s of billions. Any such capability built into swift or a good open source utility others are using for such a purpose? | 14:45 |
| acoles | cschwede: akhivin : it appears that tempauth is checking that a value is a str, but without the speedups simplejson appears to deserialize the value as unicode, so tempauth rejects the request, hence 400 response to test | 14:47 |
| acoles | i'll try to fix tempauth | 14:47 |
| acoles | test is good though ;) | 14:47 |
| cschwede | acoles: yes, i think the same - failing str vs. unicode comparison | 14:49 |
| acoles | cschwede: i think i can fix | 14:49 |
| cschwede | acoles: upstream in simplejson? ;) | 14:50 |
| acoles | cschwede: no, make the tempauth check type is basestring | 14:50 |
| *** SkyRocknRoll_ has joined #openstack-swift | 14:52 | |
| *** SkyRocknRoll has quit IRC | 14:53 | |
| tdasilva | cschwede: just a question about your simplejson patch. I removed the file as you mentioned, but when running with tox I noticed that tests did not skip, forgot that tox downsloads it's own version of simplejson | 15:04 |
| cschwede | tdasilva: yes, you’re right - that’s the good thing with tox, it installs a clean environment | 15:04 |
| tdasilva | i guess my question is, for those environments that typically don't have the support for _speedups.so, would their tox environment also not have _speedups.so? | 15:05 |
| cschwede | at the end, i’m no longer convinced on my patch; i would prefer a fix in tempauth itself | 15:05 |
| cschwede | tdasilva: i think tox creates a new virtualenv, without relying on the hosts env? | 15:05 |
| tdasilva | right | 15:06 |
| *** joeljwright has joined #openstack-swift | 15:07 | |
| cschwede | tdasilva: virtualenv requires an extra parameter to use the system-wide packages: --system-site-packages | 15:08 |
| cschwede | and i couldn’t find a hint that this is used by tox | 15:08 |
| cschwede | tdasilva: found it: http://tox.readthedocs.org/en/latest/config.html#confval-sitepackages=True|False | 15:09 |
| *** dmsimard_away is now known as dmsimard | 15:10 | |
| *** bkopilov has joined #openstack-swift | 15:12 | |
| *** annegentle has quit IRC | 15:16 | |
| *** nellysmitt has quit IRC | 15:19 | |
| tdasilva | cschwede: I was wondering more if people that ran into this problem in the first place, would still run into it if they are running tox instead of .functests | 15:19 |
| cschwede | tdasilva: no, i don’t think so | 15:19 |
| tdasilva | cschwede: ok | 15:19 |
| *** rdaly2 has joined #openstack-swift | 15:21 | |
| acoles | cschwede: just running tests locally on a tempauth patch | 15:31 |
| acoles | cschwede: did you get func tests to pass with _speedups.so removed? | 15:42 |
| *** jrichli_ has joined #openstack-swift | 15:42 | |
| acoles | cschwede: i am seeing e.g. test_unicode_metadata (test.functional.test_account.TestAccount) ... ERROR | 15:43 |
| *** reed has joined #openstack-swift | 15:48 | |
| *** silor has joined #openstack-swift | 15:56 | |
| *** thgomes has joined #openstack-swift | 16:11 | |
| *** zaitcev has joined #openstack-swift | 16:12 | |
| *** ChanServ sets mode: +v zaitcev | 16:12 | |
| *** annegentle has joined #openstack-swift | 16:16 | |
| cschwede | acoles: i see two more account tests failing: test_swift_prohibits_garbage_account_acls and test_invalid_acls | 16:17 |
| cschwede | but with FAIL, not ERROR | 16:17 |
| acoles | yes those too | 16:18 |
| acoles | which suggests to me that there are code paths that rely on the speedup extension, not just tempauth | 16:19 |
| clayg | mattoliverau: acoles: cschwede: I don't see why the code can't just work w/o simplejson speedups - i would have swore there was some other code somewhere we realized didn't work with stock json so we just fixed it | 16:19 |
| cschwede | acoles: but it’s again in the acl part, and that uses simplejson, thus i think you’re right with the failing str-unicode comparison | 16:20 |
| acoles | cschwede: i didn't see a link to acl in this test TestAccount.test_unicode_metadata | 16:21 |
| *** annegentle has quit IRC | 16:22 | |
| *** silor has quit IRC | 16:23 | |
| *** gyee has joined #openstack-swift | 16:24 | |
| *** silor has joined #openstack-swift | 16:24 | |
| acoles | cschwede: and that test ^^ ERRORS even with my fix to tempauth acl validation | 16:24 |
| openstackgerrit | Alistair Coles proposed openstack/swift: Fix tempauth acl checks when simplejson has no speedups https://review.openstack.org/159530 | 16:25 |
| cschwede | acoles: hmm, i’m wondering if that is related to the simplejson stuff at all? or it is, but more during setup (missing ACL or similar)? but i didn’t looked into that test | 16:25 |
| acoles | cschwede: ok, i have to finish early today so I just pushed my fix ^^ but I am not convinced - either we track down the other causes of func test failures, or require speedups?? have we verified it is not installed by default on trusty? | 16:26 |
| *** bill_az has quit IRC | 16:28 | |
| cschwede | acoles: well, i see the speedups extension in trusty - at least on amd64: http://packages.ubuntu.com/trusty/amd64/python-simplejson/filelist | 16:29 |
| cschwede | same for i386 | 16:30 |
| cschwede | http://packages.ubuntu.com/trusty/i386/python-simplejson/filelist | 16:30 |
| clayg | did anyone get any more info on the probetest failure from yesterday - i still can't get it to fail on master when running rsync | 16:34 |
| acoles | cschwede: found this exception in proxy log when running that func test on master: http://paste.openstack.org/show/182554/ | 16:35 |
| acoles | clayg: i didn't see the probe tests fail with rsync, just ssync, fwiw | 16:35 |
| acoles | cschwede: ok, i have to leave to take son to guitar lesson, sorry! | 16:36 |
| *** david-lyle_afk is now known as david-lyle | 16:36 | |
| cschwede | acoles: thx and have a nice evening! | 16:36 |
| acoles | cschwede: and you! | 16:36 |
| peluse | clayg, I totally got the failure before your patch and totoally don't afterwards. haven't figured out why but didn't put that much time into it (why it fixed) but instead trying to focus on the goodness of the patch itself | 16:36 |
| peluse | clayg, and will be back looking at it here shortly.... | 16:37 |
| *** thgomes has quit IRC | 16:37 | |
| akhivin | cschwede, I am sorry but how to run test to see skip messages? I ran .unittest --verbosity=3, but don't see messages why tests skipped | 16:38 |
| *** acoles is now known as acoles_away | 16:39 | |
| *** thgomes has joined #openstack-swift | 16:40 | |
| *** NM has joined #openstack-swift | 16:44 | |
| *** tsg has quit IRC | 16:44 | |
| cschwede | akhivin: the cmd is correct, you should see a lot of output, and somewhere in the middle are these lines: | 16:45 |
| cschwede | test_acl_syntax_verification (test.unit.common.middleware.test_tempauth.TestAuthMultiplePrefixes) ... SKIP: C speedups extension in Python module simplejson missing | 16:45 |
| cschwede | test_acls_propagate_to_sysmeta (test.unit.common.middleware.test_tempauth.TestAuthMultiplePrefixes) ... SKIP: C speedups extension in Python module simplejson missing | 16:45 |
| *** pcaruana has quit IRC | 16:50 | |
| *** tsg_ has joined #openstack-swift | 16:51 | |
| akhivin | cschwede, found, thanks | 16:57 |
| *** jistr has quit IRC | 17:02 | |
| peluse | clayg, you there? | 17:07 |
| peluse | I'll be off and on for a few hrs, few clarrifying questions on your udpate to efficient replciation. ping me when around | 17:08 |
| *** tsg_ has quit IRC | 17:08 | |
| *** annegentle has joined #openstack-swift | 17:18 | |
| *** nellysmitt has joined #openstack-swift | 17:20 | |
| notmyname | good morning world | 17:22 |
| notmyname | acoles_away: cschwede: I saw that there was some conversation about the simplejson stuff | 17:23 |
| notmyname | cschwede: thanks for working on that | 17:23 |
| notmyname | cschwede: I'm looking now | 17:23 |
| jrichli_ | good morning | 17:23 |
| *** annegentle has quit IRC | 17:23 | |
| notmyname | jrichli_: hello | 17:23 |
| *** nellysmitt has quit IRC | 17:25 | |
| *** SkyRocknRoll_ has quit IRC | 17:25 | |
| *** rdaly2 has quit IRC | 17:28 | |
| notmyname | cschwede: so this patch to skip tests https://review.openstack.org/#/c/159478/ kinda scares me | 17:30 |
| notmyname | cschwede: I mean, I see how it "fixes" the problem of "not all unit tests pass in condition XYZ", but that seems like the wrong way, on the surface, to fix it | 17:30 |
| *** tsg_ has joined #openstack-swift | 17:32 | |
| *** rledisez has quit IRC | 17:40 | |
| *** silor has quit IRC | 17:40 | |
| *** nellysmitt has joined #openstack-swift | 17:41 | |
| *** zul has quit IRC | 17:43 | |
| openstackgerrit | Stuart McLaren proposed openstack/python-swiftclient: Retry download after streaming has started https://review.openstack.org/159208 | 17:45 |
| *** doxavore has quit IRC | 17:52 | |
| *** mahatic has quit IRC | 17:54 | |
| clayg | peluse: sorry i was offline for a few, i was about to get on the train | 17:56 |
| *** cpallares has joined #openstack-swift | 17:57 | |
| clayg | peluse: ah... yeah that patch is crazy - kota's a mad man, zaitcev and me totally fell for it | 17:57 |
| clayg | i think the clean up helps a bit - the probetests really are showing us a bug - I should write a unittest to demonstrate the issue | 17:58 |
| clayg | peluse: i'll ping you later - heading in | 18:01 |
| *** jordanP has quit IRC | 18:01 | |
| *** zul has joined #openstack-swift | 18:09 | |
| *** doxavore has joined #openstack-swift | 18:10 | |
| *** shri has joined #openstack-swift | 18:19 | |
| *** annegentle has joined #openstack-swift | 18:19 | |
| *** thgomes has left #openstack-swift | 18:21 | |
| *** annegentle has quit IRC | 18:24 | |
| *** dencaval has joined #openstack-swift | 18:26 | |
| *** mmcardle has quit IRC | 18:35 | |
| *** mmcardle has joined #openstack-swift | 18:40 | |
| *** cpallares has quit IRC | 18:42 | |
| lcurtis | All...would I gain a performance benefit by eventually creating millions of containers and lessening the amount of objects in them? or fewer containers with more obects in them | 18:45 |
| *** gyee has quit IRC | 18:48 | |
| notmyname | lcurtis: depends on (1) the overall capacity of your cluster and amount of hardware serving it and (2) the baseline of what "more" is (eg is it on the order of 100000 or 1000000000) | 18:51 |
| * torgomatic is around | 18:55 | |
| notmyname | torgomatic: sneaky sneaky | 18:56 |
| *** joeljwright has quit IRC | 18:58 | |
| *** mmcardle has quit IRC | 19:05 | |
| clayg | peluse: here | 19:07 |
| *** geaaru has quit IRC | 19:15 | |
| notmyname | so I saw the scrollback in the api-wg meeting today. they weren't very kind to swift | 19:22 |
| notmyname | eg "Well it may not be that we need to accommodate Swift, any more than Swift needs to realize why their current design is ill-considered and why they're currently breaking RFCs and should strive to be incompliance with them" | 19:22 |
| notmyname | from http://eavesdrop.openstack.org/meetings/api_wg/2015/api_wg.2015-02-26-16.00.log.html | 19:23 |
| clayg | well... we *strive* | 19:23 |
| notmyname | so, looks like there will be a mailing list discussion about the metadata guidelines. we should pay attention to that when it lands | 19:23 |
| notmyname | when I see it, I'll link it in here | 19:23 |
| notmyname | it's about https://review.openstack.org/#/c/141229/ | 19:24 |
| clayg | yeah that's short sighted - I'd be curious which RFC they think they're following? | 19:26 |
| notmyname | so most of the api-wg stuff is innocuous. some of it is even positive. and I totally support their goal. | 19:27 |
| clayg | and I don't just mean "short sighted" as unimplemntable in swift - I mean they're on the road to having the same problem we do as soon as they give the url's to the users as resource names | 19:27 |
| torgomatic | we do have the odd spot here or there where we don't follow RFC 2616 (or whatever it is now), but it's typically because we goofed a long time ago and have to maintain compatibility | 19:27 |
| notmyname | but I'm scared of some of it turning into "guidelines that are now requirements" and totally inappropriate for any data-api, much less swift | 19:28 |
| notmyname | so that's why I'm keeping my eye on it and bringing it up | 19:28 |
| clayg | torgomatic: oh sure, i man that's fine - but i thought they were specifically calling out the metadata proposal as if they were doing something standard there - that proposal is mostly not a good idea | 19:28 |
| torgomatic | it pretty much rules out using path parts as positional params; it restricts you to a style like /objects/$objectid/subobjects/$subobjectid where your params are only in even-numbered places | 19:28 |
| torgomatic | and of course forbids slashes in thingy names | 19:29 |
| clayg | notmyname: THANKS! yeah they throw around MUST like they got a lot of input from different projects API's but when someone says "hey we can't implement that" they act like the unapproved doc review is an RFC? | 19:29 |
| clayg | notmyname: I guess you try to just help them see wy they'll be unhappy with their own proposal followed to it's logical conclusion and in the end they might thing - hrmm... i wonder what the longest lived stable api in openstack is doing? | 19:30 |
| notmyname | ya, and any help from anyone else in the comments (gerrit or the upcoming mailing list) would be appreciated too. often you can explain things in a way I can't | 19:34 |
| notmyname | also, doesn't sticking your foot in that kind of mess just sound like so much fun?! | 19:35 |
| clayg | i'll wait for the ML thread | 19:39 |
| clayg | did we loose something in /info that said tempauth supported account acl's? I feel like I'm getting skip tests I didn't expect... | 19:39 |
| *** doxavore has quit IRC | 19:40 | |
| clayg | yeah we totally did :'( | 19:41 |
| *** dencaval has quit IRC | 19:41 | |
| clayg | composite tokens - i missed that, ffffff | 19:41 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Add tempauth back to /info https://review.openstack.org/159593 | 19:48 |
| clayg | ^ should be better | 19:49 |
| *** doxavore has joined #openstack-swift | 19:51 | |
| *** rdaly2 has joined #openstack-swift | 19:52 | |
| notmyname | thanks clayg | 19:53 |
| clayg | torgomatic: isn't passing in utf-8 encoded bytestrings through json sort of a no-no? I thought it had some sort of literal unicode syntax? | 20:05 |
| torgomatic | clayg: oh yeah, that's probably true | 20:05 |
| * torgomatic will fix it | 20:07 | |
| *** nellysmitt has quit IRC | 20:09 | |
| openstackgerrit | Thiago da Silva proposed openstack/swift: WIP: refactor the PUT method into separate classes https://review.openstack.org/156825 | 20:10 |
| clayg | separate classes! tdasilva what are you doing? | 20:10 |
| tdasilva | clayg: ha, was able to remove all the data transfer code to its own class, it really simplified all the single process code | 20:12 |
| tdasilva | that update is coming next :) | 20:12 |
| tdasilva | clayg: but please check it out, would like to hear/read your thoughts on it | 20:13 |
| clayg | tdasilva: it's pretty simple - i liked patch set 2 a lot better | 20:16 |
| tdasilva | clayg: well patchset 1 was just a copy of your patch :P | 20:16 |
| tdasilva | clayg: one of the benefits is that I think it clearly separates the public API handling from the data transfer to object servers | 20:17 |
| clayg | maybe | 20:19 |
| clayg | that change does two things at once | 20:19 |
| clayg | 1) extract methods from PUT to make it easier to see the abstraction lines 2) try to flesh out the abstraction | 20:20 |
| clayg | the first part should be stright forward code maintance the second part is going to be hard, and will require thinking about alternative implementations | 20:20 |
| *** rdaly2_ has joined #openstack-swift | 20:21 | |
| *** rdaly2 has quit IRC | 20:22 | |
| *** aix has quit IRC | 20:22 | |
| tdasilva | clayg: do you think I'd have a better chance at separating the two? | 20:23 |
| tdasilva | clayg: I liked the approach that EC was taking with multiple controller classes and wanted to build on that | 20:23 |
| clayg | yeah abosluately! | 20:23 |
| clayg | yeah I think EC will only help - you probably don't even have to drive it - but your input will be invaluable | 20:23 |
| clayg | I'd focus on just getting the PUT method abstraction change where you're happy with it - subclass ObjectController directly and go looking at GET | 20:24 |
| clayg | ^ might be informed from torgomatic's work on EC since we already have the GET path extracted in the ECObjectController | 20:24 |
| clayg | *personally* | 20:25 |
| clayg | i don't have everything you're looking at in your head - but my initial reaction is pretty down on patch set 3 - sorry | 20:26 |
| clayg | When the ReplicatedObjectController get's pulled out it will either be because (it's an obvious clean pull that removes *all* data-path operations/connections from the BaseObjectController for all verbs - OR - because we have two things in tree and they want to do different things (e.g. ECObjectController and ReplicatedObjectController) | 20:27 |
| clayg | then it matters less how much little common data-plane path gets left behind in BaseObjectController because it's shared | 20:27 |
| clayg | IMHO | 20:28 |
| tdasilva | clayg: I understand you, let me just share a little bit more of what I was thinking... | 20:29 |
| tdasilva | 1. I did plan on working on the other verbs too, that would be necessary for the single process work, IMO. | 20:30 |
| tdasilva | 2. I was going to introduce a new class (e.g., SingleReplicatedObjectController) that implements its own version of _store_object method | 20:31 |
| tdasilva | in the case of the PUT verb | 20:31 |
| tdasilva | haven't looked at the other verbs yet, but would dome something similar | 20:31 |
| tdasilva | 3. EC then could also have it's own class (which they are already doing) | 20:32 |
| tdasilva | that's it :) | 20:32 |
| clayg | ok, well none of that argues that patch set 3 isn't doing two things - in fact I think you're confirming that - so generally speaking it'd be better to have two patch sets doing two things than one | 20:33 |
| clayg | I think patch set 2 would probably be mergeable before evaluating other verbs, and I'm less convinced that would be true with a BaseObjectController that still has so much data-path opinionation built in | 20:34 |
| tdasilva | yes, I agree with you! especially that like you said: the second step might require more thinking about alternative implementation | 20:34 |
| clayg | esp. given that ECObjectController has started on the other end - seems like there's are more oppertunity to coordinate if we either a) wait for EC to come to master b) work on refactoring on the feature-ec branch to help that effort match our end goal better | 20:35 |
| tdasilva | the first step is very straight forward | 20:35 |
| clayg | if you go with a i think it's an excellent oppertunity to leave PUT refactoring as purely a method extraction w/o attempting the abstraction work - and experiement with single process GET and see how it will relate to ECObjectController GET | 20:35 |
| tdasilva | sounds good to me | 20:36 |
| tdasilva | do you think this PUT refactoring could benefit EC work? | 20:38 |
| clayg | yeah me too - I think patch 3 is the future! but I don't have my flux capaciter handy so i'm left wanting for patch set 2 | 20:38 |
| clayg | yes definately | 20:38 |
| *** bill_az has joined #openstack-swift | 20:38 | |
| clayg | torgomatic has it on his list to refactor the curmdged EC PUT path out over to the ECObjectController and if we can your (patch set 2) change reviewed and landed on master before he starts that work he'll probably be less unhappy in the end | 20:39 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Make functests retry auth immediately https://review.openstack.org/159600 | 20:39 |
| tdasilva | ok, so what's the process? Can I abandon just a patch set? to revert to previous one? | 20:40 |
| clayg | tdasilva: if you take the change diff from patchset 2 -> 3 and put it up as a dependent change you can continue to iterate on it wrt the other HTTP verbs as extract ReplicatedObjectController | 20:40 |
| clayg | tdasilva: oh - yeah you can checkout the sha from patch set 2 and git review it back up as the same change | 20:41 |
| clayg | git you branch your working tree first you can rebase it on patch set 2 as a seperate commit and git review it as a dependent change | 20:41 |
| clayg | tdasilva: let me know if you have some struggle with that I think I can pull down everything needed to git that into place | 20:42 |
| tdasilva | clayg: mm...little lost, haven't done this before, but will try here | 20:43 |
| clayg | torgomatic: what the f is u"á" | 20:43 |
| torgomatic | clayg: single-codepoint unicode string | 20:43 |
| clayg | WHY PYTHON WHY -> u'\u00e1' == u'\xe1' == u"á" | 20:47 |
| clayg | There should be one-- and preferably only one --obvious way to do it. (or three if you're a unicode string) | 20:47 |
| openstackgerrit | Thiago da Silva proposed openstack/swift: WIP: initial pass at refactoring the PUT method https://review.openstack.org/156825 | 20:53 |
| *** morganfainberg is now known as needscoffee | 20:57 | |
| *** lcurtis has quit IRC | 21:05 | |
| torgomatic | tsg_: were you going to push a new PUT-path patch today? (Not pestering, just curious) | 21:07 |
| tsg_ | torgomatic: yes, the github version I showed you yesterday .. cleaning up a obj server test - hard to get the multiphase stuff into the test cases :) | 21:08 |
| openstackgerrit | Thiago da Silva proposed openstack/swift: WIP: refactor the PUT method into separate classes https://review.openstack.org/159610 | 21:09 |
| tsg_ | torgomatic: I can post what I have for review right away | 21:09 |
| torgomatic | tsg_: if you like | 21:09 |
| tsg_ | torgomatic: the test can come last | 21:09 |
| torgomatic | that way I can look sooner :) | 21:09 |
| peluse | +2 | 21:09 |
| tdasilva | clayg: I think it worked, thanks! | 21:10 |
| tsg_ | torgomatic: getting rid of "prints" and pushing in a few :) | 21:10 |
| torgomatic | tsg_: thanks! | 21:10 |
| peluse | yes, gracias! | 21:10 |
| *** panbalag has quit IRC | 21:19 | |
| *** panbalag has joined #openstack-swift | 21:20 | |
| *** lcurtis has joined #openstack-swift | 21:21 | |
| openstackgerrit | Thiago da Silva proposed openstack/swift: WIP: refactor the PUT method into separate classes https://review.openstack.org/159610 | 21:24 |
| openstackgerrit | Tushar Gohad proposed openstack/swift: EC: Multi-phase commit conversation for erasure-coded PUTs https://review.openstack.org/153843 | 21:30 |
| openstackgerrit | Tushar Gohad proposed openstack/swift: Bump eventlet version to 0.16.1 https://review.openstack.org/145403 | 21:30 |
| mattoliverau | Morning | 21:35 |
| peluse | whooop whooop | 21:35 |
| *** abhirc has joined #openstack-swift | 21:37 | |
| clayg | glange: redbo: dfg: trying to free up some inodes, some disagreement on the best path forward -> https://review.openstack.org/#/c/134855/ any input? | 21:39 |
| redbo | I haven't looked at the code, but I'm fine with the description. | 21:41 |
| redbo | you know, with xfs bulkstat you could probably find all the expired tombstones on a disk in like 20 minutes | 21:42 |
| clayg | bulkstat includes the names? | 21:43 |
| redbo | no, but you can get the name from the metadata | 21:43 |
| clayg | oic | 21:43 |
| *** lcurtis has quit IRC | 21:44 | |
| *** abhirc has quit IRC | 21:45 | |
| peluse | clayg, so I think I'm good with your update to efficient replication patch - wish I had jumped in on the earlier discussions on the original patch. To amke sure I'm not missing the forrest for the trees, let me summarizze and you can correct me | 21:46 |
| peluse | for udpates: only hit one primary in a remote region and let that primary propgoate to others in the morete | 21:47 |
| peluse | for udpate_delete: first node that we talk to is regular sync who returns in-sync objs which is passed to | 21:47 |
| peluse | subsequent nodes and an overal intersection is used as the list of things that are assured to be in-sync | 21:47 |
| peluse | enough to delete from the local... | 21:47 |
| peluse | and anything on the remote side that wasn't outside of the intersectino won't be deleted and would | 21:47 |
| peluse | eventually become a candidate for deletion on subsequent repl passes | 21:47 |
| peluse | right? | 21:47 |
| peluse | "wasn't outside" was supposed to read "was outside" | 21:47 |
| clayg | peluse: that's how it's supposed to work - still don't see how it helps the probetests for rsync replication/handoff reversion - and that change wasn't ment to target specifically the issue with ssync and single region, so it ends up doing that plus the bug | 21:48 |
| peluse | yeah, usually I wouldn't be OK with a fix that works for an unknown reason but in this the fix is a side effect so hard to block it because of that :) | 21:49 |
| clayg | peluse: ok, i appreciate your thoughts - although in retrospect I think it could be argued I'm missing some unittests - either added to that change or as a followup | 21:50 |
| *** shri1 has joined #openstack-swift | 21:51 | |
| *** shri has quit IRC | 21:53 | |
| clayg | ha! | 21:56 |
| clayg | it's the *same* dangling elif | 21:56 |
| clayg | the second half is self.conf.get == 'rsync' - but if you have it comment out, no default, then None != rsync | 21:57 |
| clayg | there's an attribute that would have worked too | 21:57 |
| clayg | notmyname: THANKS! | 21:57 |
| *** joeljwright has joined #openstack-swift | 22:00 | |
| *** jrichli_ has quit IRC | 22:09 | |
| peluse | clayg, I guess I tried backing out a different one - nice! | 22:12 |
| *** NM has quit IRC | 22:12 | |
| *** needscoffee is now known as needsmostcoffee | 22:12 | |
| *** needsmostcoffee is now known as morganfainberg | 22:13 | |
| *** abhirc has joined #openstack-swift | 22:13 | |
| *** erlon is now known as erlon_away | 22:21 | |
| openstackgerrit | John Dickinson proposed openstack/swift: fixed ugly code pattern in probe tests https://review.openstack.org/159630 | 22:21 |
| notmyname | clayg: ^^ | 22:21 |
| doxavore | when generating a tempurl with unicode characters, should one url-encode them prior to signing? | 22:22 |
| notmyname | doxavore: yes | 22:23 |
| clayg | notmyname: i'm not so sure about that? | 22:24 |
| clayg | i think the real_path - the thing that the proxy will use to hmac is going to be a utf-8 encode bytestring - not a urlencoded | 22:25 |
| notmyname | ya that | 22:25 |
| notmyname | utf8, urlencoded | 22:25 |
| clayg | oh i'm sure you patch is great | 22:25 |
| notmyname | lol. it's mostly pattern matching. entirely possible it's terrible | 22:25 |
| clayg | i'm just saying there's an if block in bin/swift-temp-url that seems to act like if you give it a urlencoded string you use a --quoted flag to make it turn it into utf-8 before making the hamc | 22:26 |
| clayg | notmyname: you have to get rid of the initalizers too right (exc = None) | 22:27 |
| clayg | in fact just do the assert on the err | 22:27 |
| notmyname | clayg: ya, I realized that after I submitted. that's what I get for dumb pattern matching | 22:27 |
| notmyname | on it | 22:28 |
| *** joeljwright has quit IRC | 22:29 | |
| openstackgerrit | John Dickinson proposed openstack/swift: fixed ugly code pattern in probe tests https://review.openstack.org/159630 | 22:32 |
| notmyname | clayg: more better ^ | 22:32 |
| doxavore | sorry... so the thought is use the normal utf8 string when signing, but obviously use filename=urlencode(str) ? | 22:33 |
| doxavore | err... filename=urlencode(str) for the path when redirecting,etc | 22:33 |
| openstackgerrit | paul luse proposed openstack/swift: Multiple fragment Archive Index Support https://review.openstack.org/159637 | 22:35 |
| notmyname | doxavore: https://gist.github.com/notmyname/10f9dc38f6c1057ad8eb | 22:35 |
| zaitcev | oh boy. I just spent time adding a comment "no longer needed exc=None?" and ... patch #2 | 22:37 |
| notmyname | zaitcev: :-) | 22:37 |
| doxavore | notmyname: got it. thanks! | 22:37 |
| zaitcev | BTW why is it that we get to use assertEquals() but in other projects pep8 bombs | 22:37 |
| notmyname | zaitcev: is that a hacking rule that they enforce? | 22:38 |
| *** gyee has joined #openstack-swift | 22:38 | |
| zaitcev | notmyname: they do in git-review at least | 22:38 |
| notmyname | zaitcev: are you just curious, or do yu think we should add that rule too? | 22:39 |
| zaitcev | Meh. I'm just glad it works | 22:39 |
| zaitcev | https://docs.python.org/2/library/unittest.html: Deprecated alias(es) failUnlessEqual, assertEquals | 22:40 |
| *** gyee has quit IRC | 22:40 | |
| *** gyee has joined #openstack-swift | 22:42 | |
| openstackgerrit | Pete Zaitcev proposed openstack/python-swiftclient: Unindent a chunk of st_list https://review.openstack.org/154653 | 22:46 |
| *** geaaru has joined #openstack-swift | 22:50 | |
| *** rdaly2_ has quit IRC | 23:00 | |
| *** rdaly2 has joined #openstack-swift | 23:05 | |
| *** rdaly2 has quit IRC | 23:07 | |
| *** IRTermite has quit IRC | 23:10 | |
| *** doxavore has quit IRC | 23:16 | |
| clayg | notmyname: zaitcev: FWIW id' be fine enforcing whatever the H is that drops Equal[s] - just because I can't ever remember which one I'm *supposed* to use and end up using them inconsistantly! | 23:22 |
| zaitcev | I don't even know why they started it. Costs them zero to maintain both. | 23:29 |
| zaitcev | But yeah | 23:29 |
| openstackgerrit | Thiago da Silva proposed openstack/swift: WIP: new attempt at single-process https://review.openstack.org/159285 | 23:35 |
| *** panbalag has quit IRC | 23:47 | |
| *** dmsimard is now known as dmsimard_away | 23:48 | |
| *** EmilienM is now known as EmilienM|afk | 23:49 | |
| *** panbalag has joined #openstack-swift | 23:50 | |
| *** panbalag has quit IRC | 23:53 | |
| *** chlong has joined #openstack-swift | 23:53 | |
| *** panbalag has joined #openstack-swift | 23:55 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!