*** jamielennox is now known as jamielennox|away | 00:17 | |
*** jamielennox|away is now known as jamielennox | 00:23 | |
*** isotope has joined #openstack-swift | 00:25 | |
*** gyee has joined #openstack-swift | 00:31 | |
*** dmorita has quit IRC | 00:46 | |
*** dmorita has joined #openstack-swift | 00:46 | |
*** jamielennox is now known as jamielennox|away | 00:46 | |
*** jamielennox|away is now known as jamielennox | 00:53 | |
*** gyee has quit IRC | 01:06 | |
kota_ | good morning | 01:07 |
---|---|---|
timburke | good morning kota_! | 01:08 |
kota_ | hi timburke! | 01:08 |
mattoliverau | kota_: morning | 01:10 |
kota_ | mattoliverau: o/ | 01:10 |
*** openstackgerrit has joined #openstack-swift | 01:20 | |
openstackgerrit | Kota Tsuyuzaki proposed openstack/pyeclib: ISA-L Cauchy support https://review.openstack.org/393276 | 01:20 |
kota_ | just rebased | 01:20 |
clayg | kota_: oh sorry - i think i had a diff on that and got distracted! :\ | 01:25 |
kota_ | clayg: no worries, you had great work to find my miss with not-4 spaces indent :/ | 01:26 |
kota_ | in the patch 393276 | 01:26 |
patchbot | https://review.openstack.org/#/c/393276/ - pyeclib - ISA-L Cauchy support | 01:26 |
clayg | kota_: lol :P | 01:27 |
openstackgerrit | Kota Tsuyuzaki proposed openstack/liberasurecode: ISA-L Cauchy support https://review.openstack.org/393263 | 01:29 |
kota_ | rebased also liberasurecode patch to 1.3.1 because i find the warning log with the newest pyeclib | 01:30 |
kota_ | hmmm, in my environment, test_get_metadata_memory_usage is getting failing... | 01:31 |
kota_ | but gate got succeeded | 01:32 |
clayg | hr.... AssertionError: Memory usage is increased unexpectedly 28864 -> 29156 | 01:32 |
kota_ | clayg: oh you also can see that? | 01:33 |
clayg | oh nice | 01:34 |
clayg | er.. yah omm it does that - dunno? | 01:34 |
kota_ | yeah, i saw that assertion error | 01:35 |
clayg | oh, it's sporadic | 01:36 |
kota_ | lemme chck | 01:37 |
kota_ | check | 01:37 |
clayg | kota_: it's making it through quite a few iterations before it bloats -> https://gist.github.com/clayg/079c06db0f980b478add8fcb90d9ca0c | 01:39 |
*** _JZ_ has quit IRC | 01:40 | |
*** zaitcev has joined #openstack-swift | 01:40 | |
*** ChanServ sets mode: +v zaitcev | 01:40 | |
clayg | i think this might just be python vm allocating something or something? I'm not sure this test is that great, or how baddly exactly it was failing when the known bug was in there? | 01:40 |
kota_ | i'm not sure the gist fixes real issue that is because similar test in test_get_segment_info_memory_usage doesn't get failed | 01:43 |
clayg | kota_: that gist doesn't "fix" anything | 01:43 |
kota_ | clayg: sure | 01:44 |
clayg | i was just observing that it's making it 100's of iterations before the memory goes up | 01:44 |
clayg | in fact - OMM it was pop at either 804 or 932 iterations pretty consistently? | 01:44 |
clayg | what iteration does it break for you? | 01:44 |
kota_ | idk for now but when i changes the count of iteration up to 1000, 10000, the memory resource was increasing so I'm doughting it could be real memory leak | 01:46 |
clayg | zomg what is going on? if i increaes the iteration to 10000 it takes longer to bloat? | 01:47 |
openstackgerrit | Kota Tsuyuzaki proposed openstack/pyeclib: Add -v option to let us know which tests are passed in the gate https://review.openstack.org/405710 | 01:47 |
clayg | was or was *not* increasing? | 01:47 |
*** klrmn has quit IRC | 01:47 | |
*** isotope has quit IRC | 01:47 | |
*** isotope has joined #openstack-swift | 01:48 | |
*** isotope has quit IRC | 01:48 | |
*** isotope has joined #openstack-swift | 01:48 | |
clayg | kota_: I have liberasurecode installed from master and I still get this warning -> WARNING:pyeclib:DEPRECATED WARNING: your liberasurecode 1.2.0 will be deprecated in the near future because of the issue https://bugs.launchpad.net/swift/+bug/1639691; Please upgrade to >=1.3.1 and rebuild pyeclib to suppress this message | 01:49 |
openstack | Launchpad bug 1639691 in PyECLib "EC: Swift can return corrupted Data and be able to go data lost at isa_l_rs_vand policy with >=5 parities" [Undecided,In progress] | 01:49 |
kota_ | AssertionError: Memory usage is increased unexpectedly 56424 -> 87576 # range(1000000) | 01:49 |
*** isotope has quit IRC | 01:49 | |
clayg | well but then that *does* sound like a memory leak then yeah? the get_segment_info test doesn't go up like that? | 01:49 |
clayg | it takes a while to 1M anything... | 01:50 |
kota_ | clayg: you need to rebuild *pyeclib* because even if you update your liberasurecode .so file with 1.3.1>=, non-rebuilt pyeclib regards your liberasurecode as 1.2.0 due to the erasurecode_version.h header when pyeclib built. | 01:51 |
kota_ | clayg: get_segment_info doesn't | 01:51 |
clayg | umm... | 01:52 |
kota_ | clayg: yeah, the version dependency is getting my headacke, i have an idea to resolve that, but sorry not yet start to work on that. | 01:52 |
*** dmorita has quit IRC | 01:52 | |
clayg | so but... how do i rebuild pyeclib? i just `sudo pip install -e .` | 01:52 |
* kota_ is diving into c source | 01:52 | |
kota_ | or likely `python setup.py install`? | 01:53 |
*** dmorita has joined #openstack-swift | 01:53 | |
clayg | how is that... | 01:53 |
kota_ | clayg: maybe you have to `rm build -rf` | 01:53 |
kota_ | before rebuilding | 01:53 |
*** klrmn has joined #openstack-swift | 01:54 | |
kota_ | oh... wait??? has we fixed the issue really??? | 02:00 |
kota_ | the code looks different way I know the memory leak in get_segment_info | 02:00 |
kota_ | and it has landed | 02:00 |
clayg | rm the build dir seemed to do the trick | 02:00 |
kota_ | clayg: you know https://review.openstack.org/#/c/385185 ? | 02:00 |
patchbot | patch 385185 - pyeclib - Fix formatted get_metadata memory leak (MERGED) | 02:00 |
kota_ | clayg: it looks you add +2 +A for that | 02:01 |
clayg | yeah i remmeber looking at that | 02:01 |
clayg | tim had put your name on it and it seemed to wfm - the existing code seemed quite wrong? | 02:02 |
kota_ | sorry, I had another meeting in my office. looking code and irc in parallel though | 02:02 |
kota_ | clayg: my work is only adding the test | 02:02 |
kota_ | clayg: i didn't look at the code yet | 02:02 |
clayg | after `rm -fr build` and reinstall the test is passing more reliably for me (how do computers work!?) | 02:03 |
clayg | the Py_BuildValue stuff looks great IMHO - basically it's idomatic code to do exactly the right thing without all the annoying footguns and boiler plate | 02:04 |
clayg | this is crazy, it's working for me reliably at 1000x but not at 1010x (!?) | 02:06 |
clayg | nm, it's not even that reliably at 1000x | 02:07 |
kota_ | clayg: not sure right now, can it steal ownership? | 02:07 |
kota_ | reference ownership, i mean | 02:08 |
clayg | i have no idea what that means - but i generally support petty crimes in favor of the greater good - steal away! | 02:10 |
timburke | kota_: my (very limited) understanding is yes, but since none of the stuff we're putting into the dict is a python object, there's no reference to steal | 02:12 |
openstackgerrit | Merged openstack/pyeclib: Add -v option to let us know which tests are passed in the gate https://review.openstack.org/405710 | 02:14 |
timburke | https://docs.python.org/2.0/ext/ownershipRules.html seems to agree? and my assumption is that they manage all the DECREFs when there's an error midway through | 02:15 |
kota_ | hmm... https://docs.python.org/2.7/c-api/arg.html#c.Py_BuildValue BuildValue make a new referenc | 02:15 |
kota_ | timburke: yeah I'm talking about around it | 02:15 |
kota_ | but sorry, my head cannot work on both topics (one in irc and the other in f2f in my office) | 02:16 |
timburke | no worries :-) | 02:16 |
kota_ | will track the logs and also check the rule | 02:16 |
kota_ | in my understanding, when we returns the dict to python vm, we should drop the reference in the each item before returning the value. | 02:17 |
*** tqtran has quit IRC | 02:18 | |
clayg | stupid hex_encode_string has an alloc | 02:21 |
kota_ | curious, http://logs.openstack.org/10/405710/1/check/gate-pyeclib-python27-ubuntu-xenial/832406c/console.html#_2016-12-02_01_49_09_289045 | 02:24 |
clayg | https://gist.github.com/clayg/f3055f975d2288e164cba57b824cc579 | 02:25 |
clayg | Buffers provided by the caller are never referenced by the objects created by Py_BuildValue(). In other words, if your code invokes malloc() and passes the allocated memory to Py_BuildValue(), your code is responsible for calling free() for that memory once Py_BuildValue() returns. | 02:26 |
clayg | ^ yeah i think that shit is good | 02:26 |
kota_ | clayg: https://gist.github.com/clayg/f3055f975d2288e164cba57b824cc579 is nice, my assertion error resolved with that. | 02:31 |
kota_ | FWIW, PyBuildValue seems to work well on each item in the dict inside if the gist is a complete fix. | 02:32 |
timburke | given an example like https://github.com/python/cpython/blob/1bfbbbf/Python/sysmodule.c#L149-L160 i think *not* DECREFing the created object before returning it is the right thing? | 02:36 |
kota_ | clayg: make sense, https://github.com/openstack/pyeclib/blob/master/src/c/pyeclib_c/pyeclib_c.c#L1017 hex_encode_string could make definately lost if we dosn't free the alloced memory | 02:39 |
kota_ | timburke: the example looks like for tuple of PyObjects? | 02:41 |
* kota_ is back to desk | 02:41 | |
kota_ | timburke: FWIW for the BuildValue, it looks safe as you said w/o DECREF because the get_metadata_memory_usage passed with no changes about the BuildValue and the references. | 02:43 |
kota_ | i think we should merge clayg's fix immediately because IMHO, current get_metadata has definately lost (alloced memory at hex_encode_string but not free) and it means, we lost swift memory for each audit time in ec object, right? | 02:45 |
kota_ | clayg: timburke:^^ | 02:45 |
kota_ | the last question, is why we couldn't detect that in the gate... | 02:46 |
timburke | i'm all for merging the fix; the cause and solution make sense | 02:46 |
*** Jeffrey4l_ has quit IRC | 02:47 | |
timburke | but i don't think we get formatted metadata anywhere yet? we just get_metadata, yeah? https://github.com/openstack/swift/blob/master/swift/obj/diskfile.py#L2637 | 02:48 |
kota_ | timburke: oh, i was missing it | 02:48 |
kota_ | timburke: yeah, that's good we're safe. | 02:49 |
kota_ | probably, one point we are using formatted dict is in testing. | 02:49 |
kota_ | IIRC | 02:49 |
*** Jeffrey4l_ has joined #openstack-swift | 02:50 | |
kota_ | phew | 02:51 |
* kota_ is going to lunch | 02:55 | |
*** clu_ has quit IRC | 02:57 | |
*** jamielennox is now known as jamielennox|away | 03:11 | |
*** jamielennox|away is now known as jamielennox | 03:20 | |
*** winggundamth has quit IRC | 03:25 | |
*** dmorita has quit IRC | 03:26 | |
*** jamielennox is now known as jamielennox|away | 03:30 | |
*** winggundamth has joined #openstack-swift | 03:31 | |
*** winggundamth has joined #openstack-swift | 03:32 | |
*** winggundamth_ has joined #openstack-swift | 03:35 | |
*** jamielennox|away is now known as jamielennox | 03:37 | |
*** psachin has joined #openstack-swift | 03:46 | |
*** links has joined #openstack-swift | 03:52 | |
clayg | mattoliverau: I agree whatever we do for lp bug #1619408 - we should do it in the object-updater | 03:58 |
openstack | Launchpad bug 1619408 in OpenStack Object Storage (swift) "old async pendings result in ghost listings" [High,New] https://launchpad.net/bugs/1619408 | 03:58 |
mattoliverau | clayg: I thought as much but just wanted to be clear, cause otherwise it would effect sharding | 03:59 |
clayg | good eye ;) | 04:00 |
jrichli | clayg: would it be overkill to be prudent and just change the isdir to a stat, and make sure the creation time is over an hour ago? | 04:03 |
clayg | jrichli: i'm not sure if you're way ahead of me or still looking at timburke's lock_path race thing? | 04:04 |
jrichli | lock_path race thing :-) | 04:04 |
clayg | ok, let me try and re-parse | 04:04 |
jrichli | sorry about that | 04:04 |
clayg | no it's great! | 04:04 |
clayg | jrichli: ok, I think i see where you're going - that *does* sound pretty good | 04:06 |
jrichli | timburke ^^ any thoughts? | 04:07 |
clayg | jrichli: still feels a little LBYL - like there could still be a chance that some guy does try: mkdir; except OSError: pass # it exists and is 1hr old --- meanwhile "oh look, this is an old empty dir!" | 04:08 |
clayg | jrichli: if you look at the third change in the series where I start rm'ing files much to timburke's dismay - in that change I think i get rid of the isdir checks in favor of expecting the errors and dealing with them | 04:09 |
jrichli | ah, ok I didnt realize these would be reused. and yes, haven't looked at that patch yet. will do | 04:10 |
clayg | jrichli: with that attitude that method becomes safe - if lockdir exibits similar problems it should *also* be updated to be safe | 04:10 |
clayg | the suffix dirs or the hashdirs? they could all be re-used in theory | 04:10 |
*** winggundamth_ has quit IRC | 04:10 | |
jrichli | honestly, i just didnt think about the dirs and how they get created when considering that idea | 04:11 |
clayg | if it's an empty suffix (because of e.g. quarantine) a replicator could push a replacement db into the old empty suffix dir just as the local sweep is trying to rmdir | 04:11 |
jrichli | so, is there always a lock done around these things? | 04:12 |
jrichli | i haven't seen the particular lock that timburke talked about | 04:12 |
clayg | jrichli: i went looking and couldn't find a the good example - *somewhere* in utils (or diskfile?) we have a helper function that makes a joke about two processes racing when it does a mkdir; touch; ENOENT # where'd my directory go!? | 04:13 |
clayg | lock_path is in common.utils | 04:13 |
jrichli | yes, i can look for all the usages. just haven't done that yet | 04:13 |
clayg | if you stick a sleep right here https://github.com/openstack/swift/blob/e9749606327fed03bad85abb2160dbf7b2f42d9c/swift/common/utils.py#L2228 | 04:14 |
jrichli | ah, i see | 04:16 |
clayg | there's a good chance IME you can get a traceback, esp. with timburke's race script | 04:16 |
*** isotope has joined #openstack-swift | 04:16 | |
*** klrmn has quit IRC | 04:22 | |
clayg | tdasilva: good merge over here! https://review.openstack.org/#/c/400126/ | 04:50 |
patchbot | patch 400126 - swift-bench - Replace assertEquals with assertEqual (MERGED) | 04:50 |
clayg | i never look at swift-bench patches - there's probably so many repo's I should pay attention to | 04:50 |
clayg | i'm totally feeling accomplished just keeping up with liberasurecode and pyeclib | 04:50 |
clayg | also someone merge this -> https://review.openstack.org/#/c/393263 | 04:51 |
patchbot | patch 393263 - liberasurecode - ISA-L Cauchy support | 04:51 |
clayg | ya'll know that like rs_vand is bad right? like even if you use liberasurecode 1.3.1 you can still get boned if you have a policy with > 4 parities - and it's not like we *say* somewhere you shouldn't use > 4 parity | 04:52 |
clayg | we need an alternative isa-l policy then we can just say "use whatever k+m you want!" - just use rs_cauch instead of rs_vand ;) | 04:52 |
clayg | it's going to be great! | 04:52 |
notmyname | fantastic, even | 04:54 |
clayg | werd | 04:54 |
clayg | and then we'll release the ssync fixes and suffix hashing fixes and ten deprecate everything in favor of hummingbird then *bam* eat it aws we win | 04:55 |
notmyname | and kota_'s global ec and cschwede part power increase :-) | 05:02 |
notmyname | so a while back, I bought a desk chair at ikea. and now when I sit on it, it slowly sinks down | 05:10 |
notmyname | but too slow to notice | 05:10 |
notmyname | just like every 10 or 20 minutes, I need to raise it back up | 05:10 |
notmyname | not often enough to think about it, but so often that it distracts me about ever 15 minutes | 05:10 |
notmyname | what I'm saying is that I think I need a new desk chair | 05:11 |
si1ver | Or thicker oil | 05:11 |
clayg | somedays your first benchmark on the new cloud deployment for your experimental build with the latest tip of master swift looks like this https://cloud.swiftstack.com/v1/AUTH_clayg/junk/bench.jpg | 05:12 |
clayg | <shrug> | 05:12 |
clayg | at least *some* of them 200'd? | 05:12 |
notmyname | me right now https://www.youtube.com/watch?v=x1Fcfr7MP2U | 05:13 |
clayg | what's a good relaxing song? some marley maybe? | 05:15 |
psachin | clayg: If you got time, any thought on this patch? https://review.openstack.org/#/c/349199/ | 05:16 |
patchbot | patch 349199 - swift - Fix swift-get-nodes arg parsing for missing ring | 05:17 |
clayg | ntata: likes it?! let's merge it? | 05:18 |
clayg | hell it even says fixes a bug I reported? what's not to like? | 05:19 |
clayg | kota_: i'm getting so sick of this error message! ERROR: Invalid Storage Policy Configuration in /etc/swift/swift.conf (Wrong ec_type isa_l_rs_cauchy for policy ec, should be one of "flat_xor_hd_3, flat_xor_hd_4, isa_l_rs_vand, liberasurecode_rs_vand") | 05:20 |
clayg | why is isa_l_rs_cauchy not a valid type yet? ;) | 05:20 |
clayg | psachin: why does `swift-get-nodes -P default a c o` say ERROR: Ring file does not exist | 05:23 |
psachin | clayg: let me check | 05:25 |
*** dmorita has joined #openstack-swift | 05:26 | |
*** ppai has joined #openstack-swift | 05:26 | |
jrichli | notmyname: lol. I got an awesome chair from Relax-the-Back. but it was pricey | 05:27 |
clayg | Relax-the-Back does not sound like a reall thing | 05:27 |
jrichli | clayg: oh, but it is :-) | 05:28 |
jrichli | https://www.relaxtheback.com/ | 05:28 |
jrichli | its like Backrite | 05:29 |
clayg | wow | 05:29 |
jrichli | and no, i dont get paid to advertise for them. and i will try not to spam anymore with commercials :-) | 05:29 |
kota_ | clayg: have you had https://review.openstack.org/#/c/393276/? | 05:30 |
patchbot | patch 393276 - pyeclib - ISA-L Cauchy support | 05:30 |
*** dmorita has quit IRC | 05:31 | |
notmyname | jrichli: is that a thing? you work in that leaned back position? | 05:31 |
notmyname | I had always thought that something like http://store.hermanmiller.com/Products/Embody-Chair was the top-end chair | 05:31 |
clayg | kota_: i just keep swithing back to master w/o changing my swift.conf | 05:32 |
clayg | kota_: it was a (bad) joke - i'm trying to emphasize how much I am looking forward to isa_l_rs_cauchy support! | 05:33 |
kota_ | clayg: in current pyeclib master, we don't have isa-l cauchy as supported yet. | 05:33 |
kota_ | yup | 05:33 |
psachin | clayg: I got same error | 05:35 |
jrichli | notmyname: that isn't my desk chair. you can search their web site for desk chairs and see all the cool things. but we do have an early version of that laid back chair called "the perfect chair" | 05:36 |
notmyname | wait. how can you have "the perfect chair" that's also "an early version"?!?! if it's perfect, why would there be a second version? | 05:37 |
openstackgerrit | Kota Tsuyuzaki proposed openstack/liberasurecode: Add get_version functionality to liberasurecode lib https://review.openstack.org/405926 | 05:38 |
jrichli | lol. we bought it years ago. perhaps there have been no changes to the basics - but there prob have been | 05:38 |
clayg | wow, so you know what's fun - a 15+3 policy in a cluster with 18 disks - you know what's less fun? trying to slam a concurrency of 100 at it and not running into a single default timeout because you don't have any handoff nodes. | 05:39 |
zaitcev | even with a node down? | 05:40 |
clayg | write q per disk is much different in (18 * 100 / 18) and (3 * 100 / 18) | 05:40 |
clayg | replicated ftw on swift in the small | 05:41 |
clayg | also don't build 18 disk swift clusters like some sort jerk | 05:41 |
openstackgerrit | Kota Tsuyuzaki proposed openstack/pyeclib: WIP: Change the version reference https://review.openstack.org/405929 | 05:49 |
kota_ | notmyname, clayg: if you consider ISA-L cauchy soon (it may be before version dependency), it would be good to include both patch 405926, patch 405929 | 05:50 |
patchbot | https://review.openstack.org/#/c/405926/ - liberasurecode - Add get_version functionality to liberasurecode lib | 05:50 |
patchbot | https://review.openstack.org/#/c/405929/ - pyeclib - WIP: Change the version reference | 05:50 |
kota_ | that patches enables to refer actual shared lib object instead of c header file and if it's available, we don't have to rebuild pyeclib i.e. you won't see some warning logs or errors. | 05:52 |
kota_ | refactoring merge proccess I supporse right now (which is a bit different from this week's irc meeting), 1. land patch 393263 and patch 405926 | 05:54 |
patchbot | https://review.openstack.org/#/c/393263/ - liberasurecode - ISA-L Cauchy support | 05:54 |
patchbot | https://review.openstack.org/#/c/405926/ - liberasurecode - Add get_version functionality to liberasurecode lib | 05:54 |
kota_ | and bump liberasurecode 1.3.2 | 05:54 |
kota_ | then | 05:55 |
kota_ | 2. land patch 405929 and patch 393276 and patch 395998 | 05:56 |
patchbot | https://review.openstack.org/#/c/405929/ - pyeclib - WIP: Change the version reference | 05:56 |
patchbot | https://review.openstack.org/#/c/393276/ - pyeclib - ISA-L Cauchy support | 05:56 |
patchbot | https://review.openstack.org/#/c/395998/ - pyeclib - Make hard-coded dependency for liberasurecode>=1.3.1 | 05:56 |
kota_ | patch 395998 should be >=1.3.2 and resolved the gate issue though | 05:56 |
patchbot | https://review.openstack.org/#/c/395998/ - pyeclib - Make hard-coded dependency for liberasurecode>=1.3.1 | 05:56 |
kota_ | and then bump pyeclib to 1.3....2?3? | 05:57 |
kota_ | which version for the newest pyeclib... | 05:57 |
kota_ | er 1.3.1 in the newest, so bump to 1.3.2 | 05:58 |
*** sams-gleb has joined #openstack-swift | 05:58 | |
*** sams-gleb has quit IRC | 05:58 | |
*** sams-gleb has joined #openstack-swift | 05:58 | |
kota_ | ocne the get_version fucntionality land to master, we could resolve gate issue for a while until distros supports the newest via sort of test-install.sh?? | 06:01 |
kota_ | i forgot the name of scripts we can use in the gate for running test... | 06:02 |
kota_ | tracking irc meeting logs... | 06:02 |
kota_ | ah, test-setup.sh | 06:02 |
kota_ | Is test-setup.sh available for now? | 06:04 |
notmyname | not yet, as far as i know | 06:06 |
kota_ | notmyname: thx, is there any other tools we can use for root before testing. I think tentative way to build liberasurecode from master for testing purpose resolves to land pyeclib isa-l cauchy support quickly | 06:08 |
kota_ | so the problem is we cannot land pyeclib isa-l cauchy until package managers supports the newest liberasurecode which supports isa-l cauchy in their repos. | 06:09 |
notmyname | right now, I don't know | 06:12 |
*** Jeffrey4l_ has quit IRC | 06:12 | |
kota_ | notmyname: k | 06:12 |
*** Jeffrey4l has joined #openstack-swift | 06:12 | |
kota_ | so i thought the dependency should be managed step-by-step, however, at this time, it's better all fixes (isa-l cauchy, dependency tools) we need included in the liberasurecode 1.3.2 | 06:13 |
kota_ | and then expect distro suppliers set it as default | 06:14 |
clayg | merge everything! | 06:14 |
kota_ | clayg: yey! | 06:14 |
clayg | kota_: but why can't we release pyeclib with cauchy support before distro packages release liberasurecode? | 06:14 |
clayg | that requirement irks me if it's true i'm curious how to remove that restriction | 06:15 |
kota_ | clayg: gate didn't pass unless it's availabele in the distros | 06:15 |
clayg | gate doesn't use isa-l cauchy | 06:15 |
kota_ | ah | 06:15 |
clayg | and the unittests for pyeclib should skip? | 06:15 |
kota_ | I may be missing something? | 06:15 |
kota_ | yeah | 06:15 |
kota_ | clayg: good point, it may be ok | 06:16 |
kota_ | let me check | 06:16 |
patchbot | (let <variable> = <value> in <command>) -- Defines <variable> to be equal to <value> in the <command> and runs the <command>. '=' and 'in' can be omitted. | 06:16 |
clayg | shut UP patchbot | 06:16 |
kota_ | no patch bot | 06:16 |
*** isotope has quit IRC | 06:16 | |
kota_ | ah, yeah, https://review.openstack.org/#/c/393276/ passes | 06:16 |
patchbot | patch 393276 - pyeclib - ISA-L Cauchy support | 06:17 |
kota_ | nice work pachbot | 06:17 |
kota_ | so we can land isa-l cauchy first | 06:17 |
kota_ | that's great | 06:17 |
kota_ | thx clayg | 06:17 |
kota_ | split repo is too hard dependency chains I was doing ever... | 06:18 |
kota_ | so we can land isa-l cauchy in pyeclib whenever but it still requires the master (not 1.3.1) even if isa-l cauchy in liberasureccode landed. | 06:19 |
kota_ | it's time to move to another meeting :/ | 06:20 |
*** ChubYann has quit IRC | 06:31 | |
*** openstackgerrit has quit IRC | 06:33 | |
kota_ | clayg, timburke: I added a little info about the conversation for pyeclib get_metadata to https://bugs.launchpad.net/pyeclib/+bug/1634006 to be track-able | 06:49 |
openstack | Launchpad bug 1634006 in PyECLib "Memory leak on get_metadata(formatted=True)" [Undecided,New] | 06:49 |
clayg | kewl | 06:50 |
*** SkyRocknRoll has joined #openstack-swift | 07:01 | |
cschwede | clayg: acoles_: can we squash patch 405295 and patch 405559 into 402324 ? i’d prefer that to have a single patch to backport, and being able to run tests on stable/newton and stable/mitaka | 07:14 |
patchbot | https://review.openstack.org/#/c/405295/ - swift - Cleanup tests from deterministic suffix updates | 07:14 |
patchbot | https://review.openstack.org/#/c/405559/ - swift - Further improvement to deterministic suffix re-has... | 07:14 |
*** jamielennox is now known as jamielennox|away | 07:18 | |
*** manous has joined #openstack-swift | 07:20 | |
clayg | cschwede: squash all of my patches always | 07:33 |
clayg | cschwede: i try to say so explicitly - i think acoles_ might have also | 07:34 |
cschwede | clayg: alright, will do that - i’ll wait for acoles_ to get online, and if Alistair is ok with squashing his patch too i will squash both | 07:34 |
clayg | cschwede: i agree for the squash for all the reasons you said | 07:34 |
clayg | didn't he say something in his review comment? i'm *pretty* sure he would agree with your reasoning - having a chain is very bad for backport reasons :'( | 07:34 |
clayg | we need to just get better at pushing over each other - i always feel like that's the easiest thing to do - i'm sure like 1 out of 100 times it will end up being more work for the author but if it was done with good intentions i'm sure there's pleanty of forgiveness to go around | 07:36 |
* clayg hugs | 07:36 | |
cschwede | clayg: yeah, i fully agree with you. more squashing, less patches! | 07:38 |
clayg | squash and merge squash and merge squash and merge | 07:38 |
*** rcernin has joined #openstack-swift | 07:40 | |
*** jamielennox|away is now known as jamielennox | 07:44 | |
*** openstackgerrit has joined #openstack-swift | 07:44 | |
openstackgerrit | Christian Schwede proposed openstack/swift: Fix non-deterministic suffix updates in hashes.pkl https://review.openstack.org/402324 | 07:44 |
cschwede | clayg: acoles_ ^^ | 07:44 |
*** openstackgerrit has quit IRC | 08:03 | |
clayg | load average 25 on a two core system is bad right? | 08:04 |
patchbot | Error: You don't have the owner capability. If you think that you should have this capability, be sure that you are identified before trying again. The 'whoami' command can tell you if you're identified. | 08:04 |
zaitcev | I'm kinda idly wondering why Pavel felt it necessary to create a oneliner _do_listdir() where an anonymous function would do the job as well... The indentation is not yet severe at the invocation site. Hmm. Everyone has a style. | 08:07 |
zaitcev | I realy have to drop now, we have a meeting in 6 hours | 08:08 |
zaitcev | No, wait | 08:08 |
clayg | zaitcev: and more importantly everyone *elses* style sucks | 08:08 |
zaitcev | heheh | 08:08 |
*** amoralej|off is now known as amoralej | 08:15 | |
clayg | public service announcement "if you call spawn, pass in your logger thread locals and restore them in the greenthread" | 08:37 |
clayg | i'm getting a bunch of log lines w/o tnxid's | 08:37 |
*** hseipp has joined #openstack-swift | 08:38 | |
clayg | quick audit of proxy.controller.obj has like... two raw in EC and one common in Putter | 08:38 |
clayg | .. that you know .. don't pass their threadlocals threw | 08:39 |
*** geaaru has joined #openstack-swift | 08:44 | |
*** zhengyin has quit IRC | 08:48 | |
clayg | man... we gotta get in that fix for min_part_hours = 0 | 08:51 |
clayg | where is that patch | 08:51 |
*** zhengyin has joined #openstack-swift | 08:59 | |
clayg | zaitcev: oh you ment why did *I* make the _do_listdir a module level oneliner! | 09:00 |
clayg | zaitcev: makes it easier to monkey-patch in tests | 09:00 |
clayg | cschwede: maybe when you're back above water on this rsync mess - patch 346475 | 09:03 |
patchbot | https://review.openstack.org/#/c/346475/ - swift - For any part, only one replica can move in a rebal... | 09:03 |
*** openstackgerrit has joined #openstack-swift | 09:04 | |
openstackgerrit | Sachin Patil proposed openstack/swift: Fix swift-get-nodes arg parsing for missing ring https://review.openstack.org/406012 | 09:04 |
*** zhugaoxiao has joined #openstack-swift | 09:16 | |
*** asettle has joined #openstack-swift | 09:42 | |
*** rcernin has quit IRC | 09:49 | |
*** rcernin has joined #openstack-swift | 09:50 | |
*** zhugaoxiao has quit IRC | 09:54 | |
*** davideagle_ has joined #openstack-swift | 10:01 | |
*** abalfour has quit IRC | 10:03 | |
*** abalfour has joined #openstack-swift | 10:03 | |
openstackgerrit | Kazuhiro MIYAHARA proposed openstack/swift: Rename ObjectUpdater variables in unit test to appropriate names https://review.openstack.org/405281 | 10:07 |
*** davideagle_ has quit IRC | 10:09 | |
*** acoles_ is now known as acoles | 10:24 | |
*** tesseract has joined #openstack-swift | 10:27 | |
*** tesseract is now known as Guest79986 | 10:27 | |
asettle | acoles: quick question if you have a second | 10:35 |
acoles | asettle: go ahead | 10:35 |
asettle | Is swift also changing to the openstack command, or keeping the swift command? (For CLI). I thought it was staying as swift, but I saw some people attempting to update | 10:36 |
acoles | cschwede: sorry, late start today....do squash! | 10:36 |
*** davideagle_ has joined #openstack-swift | 10:37 | |
*** davideagle_ has quit IRC | 10:38 | |
acoles | asettle: my opinion...it would be great to have all of the swiftclient functionality supported under openstack cli. but reality is that there has yet to be resource given to that effort (as far as I am aware), so for time being we still maintain swiftclient. | 10:39 |
*** asettle__ has joined #openstack-swift | 10:39 | |
onovy | notmyname: hi, can you look into patch 402508 and patch 402509 pls? | 10:39 |
patchbot | https://review.openstack.org/#/c/402508/ - swift (stable/mitaka) - Fixed regression in consolidate_hashes | 10:39 |
patchbot | https://review.openstack.org/#/c/402509/ - swift (stable/newton) - Fixed regression in consolidate_hashes | 10:39 |
asettle__ | Damnit I dropped from the channel. | 10:39 |
acoles | asettle: openstack cli does support some swift operations but swiftclient has a lot of other functionality. | 10:40 |
acoles | asettle__: did you miss my response? | 10:40 |
asettle__ | Oof, didn't think that answer would be simple. I think I just got it. | 10:40 |
-asettle__- acoles | 10:40 | |
-asettle__- asettle: openstack cli does support some swift operations but swiftclient has a lot of other functionality. | 10:40 | |
asettle__ | That the response? :) | 10:40 |
acoles | asettle__: asettle: my opinion...it would be great to have all of the swiftclient functionality supported under openstack cli. but reality is that there has yet to be resource given to that effort (as far as I am aware), so for time being we still maintain swiftclient. | 10:41 |
asettle__ | Yep, okay, cool, thank you. I think we might just leave it then if possible. | 10:41 |
acoles | asettle__: notmyname, timburke and joeljwright may have better opinions | 10:42 |
asettle__ | That'll be helpful later :) thank acoles ! | 10:42 |
acoles | joel<tab>, where are you? | 10:42 |
*** asettle has quit IRC | 10:42 | |
asettle__ | No idea, I actually went to message him first but he hasn't been online much lately | 10:42 |
*** asettle__ is now known as asettle | 10:42 | |
acoles | asettle: he has a habit of doing real work ;) | 10:43 |
asettle | That's just outrageous! | 10:43 |
asettle | :p | 10:43 |
*** manous has quit IRC | 10:44 | |
*** 17SAAPBC7 has joined #openstack-swift | 10:45 | |
*** joeljwright has joined #openstack-swift | 10:46 | |
*** ChanServ sets mode: +v joeljwright | 10:46 | |
asettle | joeljwright: thank you for appearing upon command | 10:46 |
joeljwright | :) | 10:46 |
asettle | I was just asking acoles about the swift command in the CLI. Have you guys migrated to the openstack command, or sticking to the swift one? | 10:47 |
asettle | And you're the guy that knows that. | 10:47 |
joeljwright | no, we haven't migrated to the openstack command | 10:47 |
asettle | Okay, cool, thanks. Any plans to do so? | 10:47 |
joeljwright | there's work to be done there, and some serious inter-project negotiations | 10:48 |
joeljwright | the openstack support for swift is very basic atm | 10:48 |
joeljwright | we need support for keystone auth sessions before we can even consider giving them a library they can use | 10:48 |
joeljwright | we are working towards it | 10:48 |
asettle | joeljwright: do you have a list at all of the command that support the openstack command? | 10:49 |
asettle | command command command | 10:49 |
joeljwright | timburke has done a lot of work in that direction | 10:49 |
joeljwright | as far as I'm aware the openstack command can do very basic operations on single objects | 10:49 |
joeljwright | but I don't have a nice list of what it supports | 10:50 |
asettle | Bummer, okay. I think we pretend you guys don't support it at the moment, and stick with the swift command. | 10:50 |
joeljwright | can we define 'you guys' in this context | 10:51 |
joeljwright | ? | 10:51 |
joeljwright | I would say that sticking with the swift command is much better at the moment | 10:51 |
joeljwright | unless your only requirement is to upload/download individual small objects one at a time | 10:52 |
acoles | joeljwright: hi! | 10:53 |
joeljwright | acoles: hi (didn't realise I wasn't logged in!) | 10:54 |
joeljwright | thought irc was quiet :) | 10:54 |
acoles | lol. its the best way to be. blissful ignorance. | 10:55 |
*** joeljwright has quit IRC | 11:10 | |
*** joeljwright has joined #openstack-swift | 11:19 | |
*** ChanServ sets mode: +v joeljwright | 11:19 | |
joeljwright | acoles: well that was a thoroughly inconvenient power cut | 11:19 |
joeljwright | did I miss anything? | 11:19 |
acoles | joeljwright: ummm, no | 11:20 |
joeljwright | sorry, I got a message from you, then everything went dark | 11:20 |
acoles | lol. others have also said I have that effect on them | 11:20 |
joeljwright | :D | 11:21 |
acoles | at least you made it back | 11:21 |
acoles | we seem to have more power cuts since brexit | 11:21 |
joeljwright | I'm going to add a review to the openstack-manuals patch | 11:21 |
joeljwright | I'm a little worried about recommending using openstack commends for swift | 11:22 |
joeljwright | when the support is so limited | 11:22 |
asettle | joeljwright: sorry I missed all this, got distracted. | 11:23 |
asettle | joeljwright: yeah I agree. Thanks for reviewing :) appreciate it | 11:24 |
*** siva_krish has joined #openstack-swift | 11:28 | |
*** zhengyin has quit IRC | 11:31 | |
clayg | psachin: sorry i missed your response - so you were able to duplciate the error - can you fix it? do you want me to leave a note on the review? | 11:39 |
psachin | clayg: Yes. https://review.openstack.org/#/c/406012/ | 11:40 |
patchbot | patch 406012 - swift - Fix swift-get-nodes arg parsing for missing ring | 11:40 |
psachin | clayg: But sorry I have this new environment and new we have new review link | 11:41 |
psachin | my bad | 11:41 |
*** Jeffrey4l has quit IRC | 11:44 | |
*** manous has joined #openstack-swift | 11:46 | |
*** Jeffrey4l has joined #openstack-swift | 11:58 | |
*** siva_krish has quit IRC | 12:01 | |
clayg | psachin: was there a test case missing? | 12:02 |
*** ppai has quit IRC | 12:02 | |
clayg | i didn't really look before, but now i see # -P test -p 1 <- that was the one I haveing problems with right? | 12:02 |
psachin | clayg: Yes.T he relevant test was missing | 12:03 |
clayg | psachin: oh, nope - it's the last one # -P test a c o | 12:03 |
clayg | psachin: hah! that's awesome! go unittest! | 12:03 |
psachin | clayg: :) | 12:04 |
clayg | psachin: nice work, i don't see a test for # -P test a/c/o (sry, have another change checked out right now) | 12:04 |
clayg | if you change again, just `git commit --amend` then when you `git review` it'll keep the same change-id in the commit message (which is all that gerrit cares about for keeping track of change sets) | 12:05 |
psachin | clayg: Yeah. Let me add that one too | 12:05 |
clayg | psachin: right on! | 12:05 |
openstackgerrit | Sachin Patil proposed openstack/swift: Fix swift-get-nodes arg parsing for missing ring https://review.openstack.org/406012 | 12:09 |
*** rcernin has quit IRC | 12:12 | |
psachin | clayg: ^^ | 12:14 |
*** rcernin has joined #openstack-swift | 12:15 | |
clayg | psachin: sweet - is that all of them? | 12:18 |
psachin | clayg: That will be all | 12:19 |
clayg | psachin: i wonder what it does if you give it a policy name *and* a ring | 12:20 |
clayg | it must have a preference or print an error or something... | 12:21 |
psachin | clayg: Yeah! it should be policy requires partition number | 12:23 |
openstackgerrit | Merged openstack/swift: Fix non-deterministic suffix updates in hashes.pkl https://review.openstack.org/402324 | 12:38 |
onovy | notmyname: and patch 406101 pls :] | 12:46 |
patchbot | Error: Spurious "]". You may want to quote your arguments with double quotes in order to prevent extra brackets from being evaluated as nested commands. | 12:46 |
patchbot | https://review.openstack.org/#/c/406101/ - swift (stable/newton) - Fix non-deterministic suffix updates in hashes.pkl | 12:46 |
clayg | psachin: well the -P is like an alternative way to specify the ring - but policy name instead of path? | 12:53 |
clayg | the -p option is like an alternative to a specific a/c/o name | 12:54 |
*** psachin has quit IRC | 12:57 | |
*** geaaru has quit IRC | 12:59 | |
*** nikivi has joined #openstack-swift | 13:02 | |
*** manous has quit IRC | 13:04 | |
openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 13:06 |
*** rcernin has quit IRC | 13:06 | |
*** rcernin has joined #openstack-swift | 13:08 | |
*** geaaru has joined #openstack-swift | 13:12 | |
openstackgerrit | Clay Gerrard proposed openstack/swift: Initial placement gets one go https://review.openstack.org/406111 | 13:13 |
*** manous has joined #openstack-swift | 13:16 | |
*** links has quit IRC | 13:17 | |
clayg | oh snap! 402324 merged! | 13:19 |
*** kei_yama has quit IRC | 13:28 | |
*** ma9 has joined #openstack-swift | 13:33 | |
*** ma9 has left #openstack-swift | 13:33 | |
*** nikivi has quit IRC | 13:37 | |
*** nikivi has joined #openstack-swift | 13:42 | |
*** nikivi has quit IRC | 13:42 | |
*** geaaru has quit IRC | 13:49 | |
*** geaaru has joined #openstack-swift | 14:02 | |
*** nikivi has joined #openstack-swift | 14:11 | |
openstackgerrit | Shashirekha Gundur proposed openstack/swift: Invalidate cached tokens api https://review.openstack.org/370319 | 14:13 |
*** nikivi has quit IRC | 14:14 | |
*** 17SAAPBC7 has quit IRC | 14:16 | |
cschwede | onovy: backport of patch 402324 for mitaka: patch 406133 | 14:16 |
patchbot | https://review.openstack.org/#/c/402324/ - swift - Fix non-deterministic suffix updates in hashes.pkl (MERGED) | 14:16 |
patchbot | https://review.openstack.org/#/c/406133/ - swift (stable/mitaka) - Fix non-deterministic suffix updates in hashes.pkl | 14:16 |
cschwede | notmyname: ^^ | 14:16 |
*** StraubTW has joined #openstack-swift | 14:18 | |
*** StraubTW has quit IRC | 14:18 | |
*** StraubTW has joined #openstack-swift | 14:18 | |
*** geaaru has quit IRC | 14:18 | |
onovy | cschwede: https://review.openstack.org/#/c/406101/ ? | 14:18 |
patchbot | patch 406101 - swift (stable/newton) - Fix non-deterministic suffix updates in hashes.pkl | 14:18 |
onovy | i was first! | 14:18 |
onovy | ah sry, mitaka/newton | 14:19 |
onovy | got it | 14:19 |
cschwede | onovy: thats newton, my backport is for mitaka - i saw yours | 14:19 |
onovy | there was collision in tests for mitaka | 14:19 |
cschwede | onovy: yeah, saw that and fixed them | 14:21 |
onovy | cool! | 14:21 |
*** bkopilov_ has joined #openstack-swift | 14:23 | |
onovy | cschwede: i added notmyname to all stable reviews | 14:24 |
cschwede | thx! | 14:24 |
*** bkopilov has quit IRC | 14:25 | |
*** geaaru has joined #openstack-swift | 14:32 | |
*** rcernin has quit IRC | 14:33 | |
*** nikivi has joined #openstack-swift | 14:34 | |
*** nikivi has quit IRC | 14:35 | |
*** amoralej is now known as amoralej|lunch | 14:36 | |
*** nikivi has joined #openstack-swift | 14:37 | |
*** nikivi has quit IRC | 14:38 | |
*** nikivi has joined #openstack-swift | 14:40 | |
*** rcernin has joined #openstack-swift | 14:41 | |
*** jamielennox is now known as jamielennox|away | 14:45 | |
*** nikivi has quit IRC | 14:45 | |
*** nikivi has joined #openstack-swift | 14:47 | |
*** silor has joined #openstack-swift | 14:52 | |
*** rcernin has quit IRC | 14:55 | |
*** rcernin has joined #openstack-swift | 14:58 | |
*** amoralej|lunch is now known as amoralej | 15:03 | |
*** dmorita has joined #openstack-swift | 15:27 | |
*** dmorita has quit IRC | 15:32 | |
*** openstackgerrit has quit IRC | 15:33 | |
*** klamath has joined #openstack-swift | 15:38 | |
*** klamath has quit IRC | 15:43 | |
*** chlong has joined #openstack-swift | 15:47 | |
*** admin6 has left #openstack-swift | 15:51 | |
*** openstackgerrit has joined #openstack-swift | 15:52 | |
openstackgerrit | Shashirekha Gundur proposed openstack/python-swiftclient: modify 'swift <sub_command> —help' display https://review.openstack.org/390119 | 15:52 |
*** Guest79986 has quit IRC | 15:54 | |
*** isotope has joined #openstack-swift | 15:59 | |
*** rcernin has quit IRC | 16:05 | |
*** isotope has quit IRC | 16:12 | |
*** david-lyle_ has joined #openstack-swift | 16:17 | |
*** david-lyle has quit IRC | 16:19 | |
*** chlong has quit IRC | 16:20 | |
*** chlong has joined #openstack-swift | 16:20 | |
*** jistr is now known as jistr|biab | 16:33 | |
*** tqtran has joined #openstack-swift | 16:38 | |
*** Jeffrey4l has quit IRC | 16:42 | |
*** jlvillal has quit IRC | 16:45 | |
*** rcernin has joined #openstack-swift | 16:45 | |
*** tqtran_ has joined #openstack-swift | 16:50 | |
*** geaaru has quit IRC | 16:51 | |
*** tqtran has quit IRC | 16:53 | |
*** jlvillal has joined #openstack-swift | 16:57 | |
*** jlvillal has quit IRC | 16:58 | |
*** jlvillal has joined #openstack-swift | 16:59 | |
*** chlong has quit IRC | 17:02 | |
*** arch-nemesis has joined #openstack-swift | 17:04 | |
*** mingyu has joined #openstack-swift | 17:04 | |
*** jlvillal has quit IRC | 17:08 | |
*** jlvillal has joined #openstack-swift | 17:08 | |
*** asettle has quit IRC | 17:11 | |
*** nikivi has quit IRC | 17:15 | |
*** nikivi has joined #openstack-swift | 17:19 | |
*** nikivi has quit IRC | 17:19 | |
notmyname | good morning | 17:22 |
openstackgerrit | Christian Schwede proposed openstack/swift: Add support to increase object ring partition power https://review.openstack.org/337297 | 17:27 |
*** dmorita has joined #openstack-swift | 17:35 | |
*** hseipp has quit IRC | 17:36 | |
*** gmmaha has quit IRC | 17:38 | |
*** gmmaha has joined #openstack-swift | 17:40 | |
*** chlong has joined #openstack-swift | 17:41 | |
openstackgerrit | Pavel Kvasnička proposed openstack/swift: Optimize hash calculation when suffix hash invalidated https://review.openstack.org/402043 | 17:42 |
*** rledisez has quit IRC | 17:43 | |
*** rcernin has quit IRC | 17:45 | |
*** rcernin has joined #openstack-swift | 17:46 | |
notmyname | I just heard about http://status.openstack.org/openstack-health/#/?searchProject=swift | 17:46 |
*** sams-gleb has quit IRC | 17:47 | |
*** sams-gleb has joined #openstack-swift | 17:47 | |
*** sams-gleb has quit IRC | 17:52 | |
*** jistr|biab is now known as jistr | 17:53 | |
*** sams-gleb has joined #openstack-swift | 17:54 | |
*** sams-gleb has quit IRC | 17:56 | |
*** sams-gleb has joined #openstack-swift | 17:59 | |
*** sams-gleb has quit IRC | 17:59 | |
*** sams-gleb has joined #openstack-swift | 17:59 | |
*** isotope has joined #openstack-swift | 18:00 | |
*** klrmn has joined #openstack-swift | 18:01 | |
openstackgerrit | Pavel Kvasnička proposed openstack/swift: Optimize hash calculation when suffix hash invalidated https://review.openstack.org/402043 | 18:03 |
*** mingyu has quit IRC | 18:06 | |
*** david-lyle_ is now known as david-lyle | 18:08 | |
openstackgerrit | Pavel Kvasnička proposed openstack/swift: Optimize hash calculation when suffix hash invalidated https://review.openstack.org/402043 | 18:11 |
*** asettle has joined #openstack-swift | 18:11 | |
*** manous has quit IRC | 18:15 | |
openstackgerrit | Pavel Kvasnička proposed openstack/swift: Optimize hash calculation when suffix hash invalidated https://review.openstack.org/402043 | 18:17 |
*** asettle has quit IRC | 18:19 | |
*** asettle has joined #openstack-swift | 18:19 | |
openstackgerrit | Alistair Coles proposed openstack/swift: Refactor recon to use single md5_hash_for_file function https://review.openstack.org/406277 | 18:21 |
*** asettle has quit IRC | 18:22 | |
openstackgerrit | Alistair Coles proposed openstack/swift: Refactor recon to use single md5_hash_for_file function https://review.openstack.org/406277 | 18:23 |
*** acoles is now known as acoles_ | 18:29 | |
*** PavelK has joined #openstack-swift | 18:30 | |
*** daemontool has joined #openstack-swift | 18:32 | |
PavelK | onovy: cschwede: acoles_: Patch 402043 is ready for review and hints are welcome. | 18:34 |
PavelK | One hint for you: do not delete .pkl files in your storage, because new files stored between directory listing and new pkl created are not taken into hashes and not invalidated. | 18:36 |
*** PavelK has quit IRC | 18:37 | |
*** daemontool has quit IRC | 18:38 | |
*** joeljwright has quit IRC | 18:41 | |
*** diogogmt has joined #openstack-swift | 18:50 | |
*** _JZ_ has joined #openstack-swift | 18:58 | |
*** mingyu has joined #openstack-swift | 19:01 | |
*** ChubYann has joined #openstack-swift | 19:02 | |
clayg | notmyname: i don't think i entirely grasp that ci failure thingy | 19:27 |
notmyname | clayg: which one? | 19:27 |
clayg | notmyname: we used to joke that the goal of ci was not to have ci always pass - if ci always passed - you don't need ci | 19:27 |
clayg | notmyname: the health? | 19:27 |
*** silor has quit IRC | 19:30 | |
*** isotope has quit IRC | 19:32 | |
*** serverascode has quit IRC | 19:34 | |
*** nottrobin has quit IRC | 19:34 | |
*** zhiyan has quit IRC | 19:34 | |
*** vint_bra has joined #openstack-swift | 19:43 | |
*** nottrobin has joined #openstack-swift | 19:47 | |
*** serverascode has joined #openstack-swift | 19:50 | |
*** zhiyan has joined #openstack-swift | 19:53 | |
*** isotope has joined #openstack-swift | 20:02 | |
*** chlong has quit IRC | 20:42 | |
*** amoralej is now known as amoralej|off | 20:51 | |
*** chlong has joined #openstack-swift | 20:55 | |
openstackgerrit | Tim Burke proposed openstack/swift: SLO: Make etag and size_bytes fully optional https://review.openstack.org/391090 | 20:56 |
*** sams-gleb has quit IRC | 20:59 | |
clayg | timburke: I think we should find a dependency we like for subUnit and add it to test-requires in Swift (!?) | 21:15 |
clayg | i'm *constantly* wanting it | 21:15 |
*** dmorita has quit IRC | 21:25 | |
*** dmorita has joined #openstack-swift | 21:25 | |
*** dmorita has quit IRC | 21:25 | |
*** asettle has joined #openstack-swift | 21:26 | |
*** asettle has quit IRC | 21:28 | |
*** mingyu has quit IRC | 21:29 | |
*** dmorita has joined #openstack-swift | 21:29 | |
*** dmorita has quit IRC | 21:39 | |
*** nexusz99 has quit IRC | 21:39 | |
*** nexusz99 has joined #openstack-swift | 21:40 | |
*** nexusz99 has quit IRC | 21:40 | |
*** vint_bra has quit IRC | 21:48 | |
*** sams-gleb has joined #openstack-swift | 21:54 | |
*** SkyRocknRoll has quit IRC | 21:56 | |
*** daemontool has joined #openstack-swift | 22:04 | |
*** sams-gleb has quit IRC | 22:07 | |
*** dmorita has joined #openstack-swift | 22:19 | |
*** dmorita has quit IRC | 22:20 | |
*** jamielennox|away is now known as jamielennox | 22:22 | |
*** dmorita has joined #openstack-swift | 22:22 | |
*** npf has joined #openstack-swift | 22:32 | |
*** chlong has quit IRC | 22:44 | |
mattoliverau | morning all | 22:59 |
MooingLemur | mroing | 23:00 |
*** sams-gleb has joined #openstack-swift | 23:05 | |
*** henryn has joined #openstack-swift | 23:08 | |
*** isotope has quit IRC | 23:08 | |
openstackgerrit | Clay Gerrard proposed openstack/swift: Add Status Code tests for Container GET https://review.openstack.org/406392 | 23:09 |
notmyname | mattoliverau: clayg: what am I missing in test_db.py? (looking at patch 405031) | 23:10 |
patchbot | https://review.openstack.org/#/c/405031/ - swift - Quarantine malformed database schema SQLite errors | 23:10 |
notmyname | mattoliverau: clayg: how are the different exception messages ever actually raised? | 23:11 |
notmyname | in test_get(), ~lin743 and following | 23:11 |
notmyname | looks like the exact same setup and calls, just a different db name? | 23:11 |
*** daemontool has quit IRC | 23:13 | |
*** isotope has joined #openstack-swift | 23:13 | |
*** npf has quit IRC | 23:15 | |
mattoliverau | notmyname: there are actually currupted dbs living in our test repo, we copy them, open them in the broker and then runs a select. This then raises the error (currupted db) that gets gets caught and we check the message | 23:16 |
*** sams-gleb has quit IRC | 23:19 | |
*** dmorita has quit IRC | 23:19 | |
notmyname | oh wow. wow. my editor isn't showing those files. at all. but yeah, there they are | 23:20 |
mattoliverau | its magic | 23:20 |
notmyname | the wrong kind of magic | 23:20 |
mattoliverau | yeah, clayg was just as surprised :) | 23:20 |
*** StraubTW has quit IRC | 23:21 | |
*** dmorita has joined #openstack-swift | 23:22 | |
notmyname | look at that. my editor by default excluded *.db | 23:25 |
notmyname | well, I just fixed that | 23:25 |
notmyname | mattoliverau: how'd you create the schema fail? like some partial wirte or something? | 23:26 |
mattoliverau | notmyname: I documented it in the bug :) | 23:27 |
notmyname | kk | 23:27 |
mattoliverau | Because it's a schema fail, you can turn a editable schemas with a pragma | 23:27 |
*** _JZ_ has quit IRC | 23:31 | |
notmyname | clayg: what do you think about adding a few extra lines to patch 406392? https://gist.github.com/notmyname/8933e9e7f9e53c991f122317476f3d37 | 23:52 |
patchbot | https://review.openstack.org/#/c/406392/ - swift - Add Status Code tests for Container GET | 23:52 |
notmyname | I can push that an approve | 23:53 |
notmyname | it's good stuff overall | 23:53 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!