Thursday, 2014-10-23

*** tsg has joined #openstack-swift00:03
*** kyles_ne has quit IRC00:15
*** kyles_ne has joined #openstack-swift00:16
*** tsg has quit IRC00:17
*** svetmy has joined #openstack-swift00:18
svetmyHi, i'm configuring swift on icehouse - added a storage node. Now when i'm creating a container - it reported created successfully, but not showing. Has anyone seen this?00:18
*** kyles_ne has quit IRC00:20
*** dmorita has joined #openstack-swift00:23
*** rmcall_ has joined #openstack-swift00:34
*** rmcall has quit IRC00:35
*** rmcall_ is now known as rmcall00:35
*** rmcall has quit IRC00:47
*** jergerber has quit IRC01:16
*** dmsimard_away is now known as dmsimard02:01
*** nosnos has joined #openstack-swift02:02
*** kyles_ne has joined #openstack-swift02:17
*** haomaiwang has joined #openstack-swift02:29
*** 6A4AA5E3S has quit IRC02:33
*** haomaiwang has quit IRC02:42
*** SkyRocknRoll has joined #openstack-swift02:50
*** rmcall has joined #openstack-swift02:51
*** mrsnivvel has joined #openstack-swift03:13
*** tsg has joined #openstack-swift03:15
zaitcevsvetmy: check if updaters run well. Maybe just forgot to start, forgot to add into permanent boot with systemctl enable or chkconfig on.03:15
*** haomaiwang has joined #openstack-swift03:18
*** tsg has quit IRC03:21
*** kyles_ne has quit IRC03:26
*** kyles_ne has joined #openstack-swift03:26
*** haomaiwang has quit IRC03:33
*** haomaiwang has joined #openstack-swift03:33
*** exploreshaifali has joined #openstack-swift03:45
*** hhuang has joined #openstack-swift03:48
*** kyles_ne has quit IRC03:49
*** nosnos has quit IRC04:00
*** nosnos has joined #openstack-swift04:01
*** nosnos has quit IRC04:05
*** addnull has joined #openstack-swift04:08
*** delatte has joined #openstack-swift04:11
*** jergerber has joined #openstack-swift04:11
*** delattec has quit IRC04:12
*** dmsimard is now known as dmsimard_away04:18
*** exploreshaifali has quit IRC04:23
notmynamehello04:33
elambert_notmyname: hello :-)04:34
notmynameelambert_: I'm working on fine-tuning a presentation. what's your excuse for being online at 9:30pm? ;-)04:34
elambert_its either that .. or do the dishes04:35
notmynameheh04:35
notmynameclearly you've made the wise choice :-)04:35
elambert_was an easy one to make04:36
*** nosnos has joined #openstack-swift04:40
*** dmorita has quit IRC04:46
*** zaitcev has quit IRC04:49
*** mrmoje has joined #openstack-swift04:52
*** hhuang has quit IRC04:59
*** tsg has joined #openstack-swift05:19
*** exploreshaifali has joined #openstack-swift05:33
*** jergerber has quit IRC05:44
*** SkyRocknRoll has quit IRC05:50
*** haomaiwa_ has joined #openstack-swift05:50
mattoliveraunotmyname: how did the tuesday presentation go? Much tweaking for tomorrows?05:53
notmynamemattoliverau: went' pretty well, thanks05:53
notmynamenot too many edits, but there are a few05:53
*** haomaiwang has quit IRC05:54
*** hhuang has joined #openstack-swift05:55
mattoliveraunotmyname: well I may or may not be goofing off from normal work to attempt a half arsed messaging middleware using oslo messaging as a POC, just to see if it'll work... stealing/borrowing much of the data_migration framework, ie drivers for rabbitmq, qpid, etc. (thanks Gill et el).  Damn swift meetings that make you think :P05:58
*** mrmoje has quit IRC06:01
mattoliverauWow, it's already 5 pm.. I'm supposed to have been to the shops so I have food to cook for dinner... opps.. bbl06:01
*** SkyRocknRoll has joined #openstack-swift06:03
*** bkopilov has quit IRC06:05
*** addnull has quit IRC06:09
*** dmsimard_away has quit IRC06:11
*** openstack has joined #openstack-swift06:13
*** exploreshaifali has quit IRC06:15
*** classics1 has joined #openstack-swift06:15
*** dmsimardz has joined #openstack-swift06:15
*** dmsimardz is now known as dmsimard06:15
*** clds_ has joined #openstack-swift06:15
*** acoles_away has quit IRC06:15
*** mandarin1 has joined #openstack-swift06:15
*** classicsnail has quit IRC06:15
*** j_king has quit IRC06:15
*** mandarine has quit IRC06:15
*** clds has quit IRC06:15
*** theanalyst has quit IRC06:15
*** tsg has quit IRC06:21
*** bkopilov has joined #openstack-swift06:22
*** hhuang has quit IRC06:27
*** dmsimard has quit IRC06:43
*** DisneyRicky has quit IRC06:44
*** dmsimard_away has joined #openstack-swift06:44
*** dmsimard_away is now known as dmsimard06:44
*** fbo has quit IRC06:45
*** fbo has joined #openstack-swift06:45
*** elambert_ has quit IRC06:49
*** fbo has quit IRC06:50
*** fbo has joined #openstack-swift06:50
*** openstack has quit IRC06:59
*** openstack has joined #openstack-swift14:18
-cameron.freenode.net- [freenode-info] channel trolls and no channel staff around to help? please check with freenode support: http://freenode.net/faq.shtml#gettinghelp14:18
*** openstackstatus has joined #openstack-swift14:19
*** ChanServ sets mode: +v openstackstatus14:19
*** foexle has quit IRC14:25
openstackgerritChristian Schwede proposed a change to openstack/swift: Return verbose message if account quota exceeded  https://review.openstack.org/13055314:28
*** jergerber has joined #openstack-swift14:29
*** tdasilva has joined #openstack-swift14:41
*** elambert has quit IRC14:43
*** elambert_ has quit IRC14:43
*** hhuang has joined #openstack-swift14:46
notmynamegood morning14:46
openstackgerritSimon Lorenz proposed a change to openstack/swift: Added flag to create unique tmp dir per strg node  https://review.openstack.org/12408814:48
acolesnotmyname: this is fun... http://15.126.204.92:9000/dashboard/?did=414:54
notmynameacoles: that is cool14:55
acolesclick on swift to drill down14:55
*** lpabon has joined #openstack-swift14:56
acolesnotmyname: from here http://15.126.204.92:9000/dashboard/index/1394 if you click on coverage (87.7%) you get to browse code annotated with coverage issues15:00
notmynameyeah, that's cool15:00
notmynameis this your tool or ...?15:00
acolesa colleague here in bristol just set it up - he's wondering how to integrate it with jenkins15:01
*** hhuang has quit IRC15:02
tdasilvaacoles: that's really neat!15:02
acolesapparently you can train it to ignore certain issues (e.g. false negatives)15:04
*** openstackgerrit has quit IRC15:07
acolespeluse: you about?15:08
notmynamegotta run. going to LA today and tomorrow to talk about swift15:15
notmynameI'll be on my cell if anyone needs me15:15
tdasilvanotmyname: have a good trip, thanks for the help15:16
*** SkyRocknRoll has quit IRC15:27
thurloatcrap, missed notmyname15:29
*** tsg has joined #openstack-swift15:32
*** kyles_ne has joined #openstack-swift15:37
*** annegent_ has joined #openstack-swift15:48
*** lcurtis has quit IRC15:56
*** gyee has joined #openstack-swift15:58
*** openstack has joined #openstack-swift16:02
*** chrisnelson_ is now known as chrisnelson16:04
*** annegent_ has quit IRC16:10
*** haomaiw__ has quit IRC16:33
*** haomaiwang has joined #openstack-swift16:33
*** haomaiw__ has joined #openstack-swift16:36
*** exploreshaifali has joined #openstack-swift16:37
*** lpabon has quit IRC16:38
*** haomaiwang has quit IRC16:40
*** jordanP has quit IRC16:46
*** delattec has joined #openstack-swift16:48
*** rmcall has joined #openstack-swift16:49
*** delatte has quit IRC16:50
*** delatte has joined #openstack-swift16:50
*** openstackgerrit has joined #openstack-swift16:50
*** cdelatte has quit IRC16:51
*** rmcall has quit IRC16:52
*** rmcall has joined #openstack-swift16:56
*** jistr has quit IRC16:57
*** lpabon has joined #openstack-swift16:59
*** rmcall_ has joined #openstack-swift17:07
*** rmcall has quit IRC17:08
*** rmcall_ is now known as rmcall17:08
*** elambert has joined #openstack-swift17:09
*** addnull has quit IRC17:10
*** davidhadas has joined #openstack-swift17:12
*** tsg has quit IRC17:18
*** zul has quit IRC17:18
*** zhill has joined #openstack-swift17:22
peluseacoles, I'm here now for another 20 min or so17:25
acolespeluse: hi, quick question hopefully...17:26
acolesi was looking at this review earlier...17:26
pelusewill give it a go...17:26
peluseBTW, cool stats page earlier17:26
acoleshttps://review.openstack.org/#/c/112449/417:27
acolesjust wondering what the status of this is now w.r.t. the .durable file etc.?17:27
*** rmcall has quit IRC17:27
acolesi.e. is this still current, cos it seems like the decision to delete or not is now not so much per-policy but when the .durable arrives17:28
peluseacoles, this is jsut the "don't delete stuff" portion so its ready to go as as.  Handling of the other aspects of the multi-phase PUT, cleanup, etc., are separate tasks17:28
*** tsg has joined #openstack-swift17:28
*** rmcall has joined #openstack-swift17:29
peluseacoles, well its still per policy as there's no concept of .durable other than per policy17:29
acolesso, will a .durable be in a different policy - if not then the policy.auto_cleanup will prevent deleting old stuff even when the .durable is written?17:30
* acoles is confused17:30
pelusenow I am too :)17:30
acolessorry :/17:30
peluselet me circle back and give it another once over in light of the latest discussions and I'll update it this afternoon with either a tweak or a WIP -1 if it looks like it needs real work.  thanks!17:31
acolesok, i'm at end of day here so won't get back to it til tomorrow.17:31
peluseno problem, thanks for taking a look17:32
acolespeluse: one thought i had was to make auto_cleanup a parameter of the diskfile object, then based on policy set it false for an EC .data, leave it true for an EC .durable17:32
acolesanyway, i'll wait for your thoughts :)17:32
*** zul has joined #openstack-swift17:34
*** annegent_ has joined #openstack-swift17:44
*** openstackgerrit has quit IRC17:48
*** openstackgerrit has joined #openstack-swift17:50
*** zhill has quit IRC17:54
*** geaaru has quit IRC17:56
*** lpabon has quit IRC17:58
*** aix has quit IRC17:59
*** lpabon has joined #openstack-swift18:18
*** zaitcev has joined #openstack-swift18:20
*** ChanServ sets mode: +v zaitcev18:20
*** kyles_ne has quit IRC18:21
*** acoles is now known as acoles_away18:26
*** tsg has quit IRC18:29
*** mahatic has joined #openstack-swift18:32
*** bkopilov has quit IRC18:35
*** kyles_ne has joined #openstack-swift18:38
mahaticcschwede, hi! around?18:38
*** annegent_ has quit IRC18:43
*** exploreshaifali has quit IRC18:44
*** nellysmitt has quit IRC18:45
*** nellysmi_ has joined #openstack-swift18:45
*** jamespage_ has joined #openstack-swift18:59
*** portante has quit IRC19:06
*** ndk_ has quit IRC19:06
*** nellysmitt has joined #openstack-swift19:12
*** jamespage_ has quit IRC19:15
*** nellysmi_ has quit IRC19:16
*** ndk has joined #openstack-swift19:17
*** portante has joined #openstack-swift19:19
*** ChanServ sets mode: +v portante19:19
*** Nadeem has joined #openstack-swift19:22
*** mrmoje has joined #openstack-swift19:25
*** mrmoje has quit IRC19:31
*** mrmoje has joined #openstack-swift19:31
*** exploreshaifali has joined #openstack-swift19:36
*** sandywalsh_ has joined #openstack-swift19:37
*** aix has joined #openstack-swift19:37
*** sandywalsh has quit IRC19:39
*** rmcall has quit IRC19:46
*** nomad411 has joined #openstack-swift20:01
*** exploreshaifali has quit IRC20:02
*** mahatic has quit IRC20:08
*** tristanC has quit IRC20:10
*** tristanC has joined #openstack-swift20:15
*** mahatic has joined #openstack-swift20:15
*** gyee has quit IRC20:21
*** Nadeem has quit IRC20:22
*** nellysmitt has quit IRC20:37
notmynamethurloat: what's up. I'm back online for a bit20:47
thurloathey notmyname, wanted to get your opinion on the approach here: https://bitbucket.org/clouda/swift-container-keys20:49
notmynamelooking20:50
thurloatthanks :)20:50
*** CaioBrentano1 has quit IRC20:50
thurloatmain idea, container based long lived auth tokens for read / write access20:51
notmynamebut still a shared secret passed in a header20:51
notmynamealso, no "identity" associated with it, so you don't give creds (authn) before getting access (authz). so this is "if you know the shared secret, you have access" and "you have to coordinate every user if you ever want to change the key"20:53
notmynamethurloat: so if all of that is ok, it seems like a pretty simple system. what's your use case for this?20:54
thurloatthat's the plan, stan. the main use-case is embedding for applications that need to upload data to or read a specific container20:54
*** gyee has joined #openstack-swift20:54
thurloatso no keystone auth data is ever embedded20:54
notmynamethurloat: interesting. I'd probably have used tempurls first. it's a per-container shared secret that you can use to grant time-limited access for a particular verb20:55
thurloattempurl keys are per account, iirc20:55
notmynameah, yeah. that's right20:55
notmyname...maybe that could be changed...20:56
thurloatessentially the same mechanism, with no expiry20:56
thurloatwas actually planning on using tempurls to come up witha solution to the problem, but didnt want to make changes to core middleware20:57
thurloatand it would have ended up hacky20:57
torgomaticyou can get effectively no expiry with tempurls by using a far future time20:58
*** bitblt has joined #openstack-swift20:58
notmynameI'm looking over at the code20:58
thurloatand we have horizon integration for our customers to revoke & regen new keys for a container20:58
torgomaticalthough it can't give you container reads, just object, so no listings20:58
notmynameline 48 should say container metadata instead of account metadata. that confused me at first :-20:59
notmyname)20:59
thurloattorgomatic: yea, i hmmd and hawd about it before i wrote a LOC. this fits the end user use-case a lot nicer.21:00
thurloatdocsbug!21:00
*** NM1 has quit IRC21:00
notmynamethurloat: line 123 should probably not be a blanket set the authorize callback. maybe21:02
notmynameI don't remember right off what the authorize_override does21:02
thurloatnotmyname: based the skipping off of https://github.com/openstack/swift/blob/master/swift/common/middleware/tempurl.py#L324 from tempurl21:03
thurloatopen to a better way21:04
torgomaticyou need authorize override to keep the other auth systems out of your hair21:04
notmynameyeah. see it now21:05
notmynameit's nicely undocumented ;-)21:05
*** gyee has quit IRC21:06
thurloatcan't be sure enough when dealing with auth stuff21:06
mattoliverauMorning21:06
thurloatI really appreciate you having a review notmyname, helps to get feedback outside of the echo chamber ;)21:08
notmynamethurloat: sorry got distracted here21:11
swat30* ecoh *21:11
notmynamethurloat: so about setting swift.authorize21:11
*** HenryG has quit IRC21:11
thurloatwho let swat30 out of his cage?21:12
notmynameall I'm wondering is if your code starts to play better with others (and be longer-lived and easier to maintain) if you have a guard like https://github.com/openstack/swift/blob/master/swift/common/middleware/tempauth.py#L21121:12
notmynamethurloat: IOW, that kind of thing probably makes it easier to use with other auth systems21:13
thurloatmm would there be a clash with another system checking is swift.authorize is set?21:13
notmynamebut mostly I'm just sitting in a hotel room looking at code I didn't write and thinking out loud. so don't take that as The Way It Should Be (tm)21:13
thurloatmaybe i'd be best just setting override boolean, keystone backend checks for that before even mentioning swift.authorize21:14
thurloatboth tempauth and keystone auth play that way21:15
notmynamethurloat: eg if the pipline is "... tempauth containerkey ..." then tempauth will set the authorize callback (because it's empty) then container key might stomp on it21:15
notmynameyeah21:15
notmynames/might/will/21:15
notmynamebut maybe that's not a problem either:-)21:16
thurloatit goes up front before auths in the pipeline (docc'd in readme)21:16
swat30the one thing he doc'd ;)21:16
notmynameheh21:16
notmynameso otherwise you're using keystone?21:16
thurloatyeah21:17
notmynameare you using uuid tokens or the pki tokens?21:17
swat30just uuid21:17
notmynameand how's that working for you?21:17
swat30fine, haven't run into any issues21:17
notmynamecool21:17
swat30are there known problems w/ swift + UUID?21:17
notmynameswat30: no. with keystone+swift, the uuid is pretty much the way to go (because PKI tokens are so big). but people have run in to scale issues with keystone in the past when using it with swift.21:18
swat30yea, we've run into some issues in testing with PKI. what kind of scale are we talking?21:19
notmynamehundreds to thousands of req/sec to swift21:20
swat30hm, interesting21:21
swat30just due to all of the API calls I assume?21:21
notmynameyeah. or maybe with the storage for the tokens21:21
swat30yea, we use DB + memcached21:22
notmynamestill, if it's working well for you, I'm really happy to hear that21:22
swat30yea, been really great so far21:25
thurloatyou can see the fancy read-key in action on our prod swift21:27
thurloatcurl -H "X-Container-Meta-Read-Key:notmynamekeykeykeykeykey" https://swift.ca-ns-1.clouda.ca:8443/v2.0/AUTH_5512b421d6c84760b22ab4422b783c36/test21:27
* notmyname downloads all the filz21:28
thurloata bunch of swift testy files :)21:28
*** lpabon has quit IRC21:29
thurloatthe 3rd party tool integration is spotty at best with keystone backend21:29
thurloatspeaking of the files in that bucket21:29
thurloats/bucket/container/21:29
swat30thurloat, what issues have we seen WRT 3rd party access?21:30
swat30notmyname, have you seen demand for a container based auth middleware?21:33
thurloatspotty support for keystone auth for most of the windows installed ones, usually different reasons. came across hardcoded rackspace auth urls, only keystone v1 support, etc21:33
swat30more than happy to share :)21:33
notmynameswat30: there were some guys asking in IRC from some canadian hoster recently. they had some "containerkey" middleware....21:34
swat30lol21:34
thurloatdamn canadian hosers21:34
swat30eh?21:34
thurloatswat30: speaking of keystone issue, i forked swift-nbd-server to add better keystone support on our GH repo a few months ago https://github.com/CloudA/swift-nbd-server/commits/master21:37
swat30yea I noticed that, haven't seen it in action tho21:37
*** zhill has joined #openstack-swift21:38
thurloatit's pretty meh, literally creates a blocks -> objects mapping21:38
thurloatimagine how slow that'd be on random reads21:39
swat30yea, wouldn't imagine that you would want to use it for anything that requires any sort of performance lol21:39
thurloatnop, saving grace was i formatted the block device with 1 meg blocks.21:40
thurloatand hope it's in vm_cache when you re-read it21:40
thurloatlet's take this back to hipchat swat30, offtopic land.21:42
thurloatthanks again notmyname for helping out21:42
swat30thurloat, seconded sir21:42
*** delattec has quit IRC22:01
*** delatte has quit IRC22:02
*** gyee has joined #openstack-swift22:32
*** bitblt has quit IRC22:36
*** dmsimard is now known as dmsimard_away22:59
*** HenryG has joined #openstack-swift23:04
*** kyles_ne has quit IRC23:05
*** kyles_ne has joined #openstack-swift23:05
*** kyles_ne has quit IRC23:10
*** elambert has quit IRC23:11
*** elambert has joined #openstack-swift23:12
*** kyles_ne has joined #openstack-swift23:13
openstackgerritJay Bryant proposed a change to openstack/swift: Handle os.listdir failures in container-updater  https://review.openstack.org/12586423:15
*** elambert has quit IRC23:16
*** lpabon has joined #openstack-swift23:17
openstackgerritClay Gerrard proposed a change to openstack/python-swiftclient: Add some bash helpers for auth stuff  https://review.openstack.org/8622423:20
*** mahatic has quit IRC23:27
*** mahatic has joined #openstack-swift23:28
*** X019 has joined #openstack-swift23:31
*** mahatic has quit IRC23:37
*** mahatic has joined #openstack-swift23:37
*** oomichi has joined #openstack-swift23:49
*** oomichi__ has joined #openstack-swift23:49
*** kyles_ne has quit IRC23:53

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!