Friday, 2013-09-13

*** NikitaKonovalov has joined #savanna00:48
*** NikitaKonovalov has quit IRC00:53
*** nosnos has joined #savanna01:06
*** fandikurnia01 has joined #savanna01:29
*** NikitaKonovalov has joined #savanna01:48
*** NikitaKonovalov has quit IRC01:53
*** fandikurnia01 has quit IRC01:59
*** tstclair has quit IRC02:02
*** fandikurnia01 has joined #savanna02:04
*** tstclair has joined #savanna02:08
*** kbroughton has joined #savanna02:14
*** fandikurnia01 has quit IRC02:25
openstackgerritChangBo Guo proposed a change to stackforge/savanna: Don't use inside of transaction
*** chandoo has quit IRC02:43
*** NikitaKonovalov has joined #savanna02:49
*** NikitaKonovalov has quit IRC02:54
openstackgerritChangBo Guo proposed a change to stackforge/savanna: Don't use inside of transaction
*** NikitaKonovalov has joined #savanna03:50
*** NikitaKonovalov has quit IRC03:54
*** fandikurnia01 has joined #savanna03:58
kbroughtonhas anyone tried savanna with the docker hypervisor?04:11
openstackgerritChangBo Guo proposed a change to stackforge/savanna: Don't use inside of transaction
*** fandikurnia01 has quit IRC04:15
*** fandikurnia01 has joined #savanna04:16
*** fandikurnia01 has quit IRC04:21
*** akuznetsov has joined #savanna04:48
*** SergeyLukjanov has joined #savanna04:49
*** NikitaKonovalov has joined #savanna04:50
*** NikitaKonovalov has quit IRC04:55
*** nvn77 has joined #savanna05:02
*** sacharya has quit IRC05:06
*** SergeyLukjanov has quit IRC05:08
*** dmitryme has joined #savanna05:14
*** NikitaKonovalov has joined #savanna05:43
*** NikitaKonovalov has quit IRC05:49
*** dmitryme has quit IRC06:03
*** akuznetsov has quit IRC07:03
*** asavu has joined #savanna07:20
*** akuznetsov has joined #savanna07:59
*** nprivalova has joined #savanna08:00
*** dmitryme has joined #savanna08:02
*** dmitryme has quit IRC08:04
*** kbroughton has quit IRC08:37
*** alazarev has joined #savanna08:42
*** nprivalova has quit IRC08:43
*** nprivalova has joined #savanna08:50
*** asavu has quit IRC09:06
*** alazarev has quit IRC09:10
*** alazarev has joined #savanna09:19
*** NikitaKonovalov has joined #savanna09:34
*** SergeyLukjanov has joined #savanna09:41
*** SergeyLukjanov has quit IRC09:54
*** nprivalova has quit IRC10:08
*** SergeyLukjanov has joined #savanna10:10
*** nosnos has quit IRC10:26
*** nosnos has joined #savanna10:27
*** nosnos has quit IRC10:27
*** nosnos has joined #savanna10:28
*** nprivalova has joined #savanna10:35
*** nosnos has quit IRC10:35
*** alazarev has quit IRC10:36
*** nosnos has joined #savanna10:36
*** nosnos has quit IRC10:40
openstackgerritMatthew Farrellee proposed a change to stackforge/savanna-dashboard: Add neutron import available in the RDO release of Havanna
*** nprivalova has quit IRC10:48
*** nprivalova has joined #savanna10:48
*** _mattf is now known as mattf10:55
*** ruhe has joined #savanna10:57
*** alazarev has joined #savanna11:06
*** nprivalova_ has joined #savanna11:12
*** alazarev has quit IRC11:15
*** nprivalova has quit IRC11:15
*** nprivalova_ is now known as nprivalova11:15
*** ruhe has quit IRC11:16
*** alazarev has joined #savanna11:16
*** ruhe has joined #savanna11:19
*** alazarev has quit IRC11:28
*** alazarev has joined #savanna11:29
*** alazarev has quit IRC11:40
*** alazarev has joined #savanna11:52
*** alazarev has quit IRC12:03
*** kbroughton has joined #savanna12:12
*** _crobertsrh is now known as crobertsrh12:14
*** chandoo has joined #savanna12:15
openstackgerritVadim Rovachev proposed a change to stackforge/savanna-dashboard: Added UI tests
*** SergeyLukjanov has quit IRC12:29
*** rakhmerov has joined #savanna12:32
rakhmerovhi, I noticed that we currently have two savanna API urls for job binaries: "/job-binaries" and "/job-binary-internals". The first one is intended only for "read" access in cases we store some of the binaries, i.e. in "swift" or other external storages. The second one is for storing binaries right in savanna tables and using API we can upload whatever we need.12:38
rakhmerovQuestion: aren't these names confusing?12:39
*** asavu has joined #savanna12:39
rakhmerov"job-binaries" and "job-binary-internals"12:39
rakhmerovin my opinion, they don't 100% correspond to what they're really designed for12:40
*** SergeyLukjanov has joined #savanna12:42
rakhmerovI could suggest something like "/job-external-binaries" and "/job-binaries" respectevely12:43
rakhmerovany thoughts on this?12:43
nprivalovajob-binary-internal is just a blob in db12:43
nprivalovamaybe it will be better to call it just internal_blob12:44
rakhmerovI'm trying to think about it from a user perspective12:44
nprivalovauser doesn't work with job-binary-internal  directly12:44
rakhmerovno-no, for a user it doesn't matter if it's a blob in a db or something else. Internal representation is not what they should thing about12:44
nprivalovafrom UI I mean12:44
rakhmerovhm.. I might not have understood correctly then. I thought we could just upload some script via UI using /job-binary-internals, no?12:46
nprivalovait was created to let user upload a file from PC, e.g.12:46
nprivalovano, it will be done in job-binary page12:46
nprivalovatake a look at
nprivalovauser will not work with  /job-binary-internals, only with  /job-binary12:48
nprivalovasavanna will use  /job-binary-internals to create corresponding job-binary12:48
rakhmerovok, give me a second12:49
*** alazarev has joined #savanna12:54
rakhmerovwhat's described in this file is totally fine. However, I'm mostly talking about REST API itself. Imagine, I'm a 3rd party person developing an application working as a client for Savanna. In this case I may not be interested in using anything like SWIFT first to be able to run some jobs. I just want to use some url to immediately upload some script and use it as a binary for my job. So this url would be a little bit confusing to me12:56
rakhmerovso I mean from a REST API standpoint it's a regular, and pretty useful method that I can use to upload scripts I need12:57
rakhmerovdo see my point?12:57
mattfrakhmerov is making a nice point that we need to enhance the way we view consumers of our rest apis12:57
rakhmerovI would just suggest you guys disscuss it once more12:57
mattfwe're already thinking about 3rd parties, but focus is currently on the dashboard user12:58
rakhmerovyes, but I'm trying to extend that vision a little bit further12:58
mattfwe're approaching it in a somewhat organic way, but your point is well received - we need to step it up a little on the non-dashboard user of the rest api12:59
*** alazarev has quit IRC12:59
rakhmerovyep, ok12:59
rakhmerovjust wanted to deliver my message here..12:59
mattfgiven that context, what would you suggest? fyi, rnirmal has also been interested in this topic.13:00
nprivalovato work directly with script from swift you will create bon-binary without job-binary-internal13:00
rakhmerovI would suggest i.e. "job-binary-storage" (for various storages) and "job-binary" (for what is called "job-binary-internal" now)13:03
rakhmerovor may be we could leave "job-binary-internal" as it is but rename "job-binary" so that expresses the storage-oriented binaries13:04
mattfshould the 3rd party client have to care if it s internal or external? i'd think not.13:05
rakhmerovi.e. "job-binary-external" and "job-binary-internal", this way they would look symmetric13:05
mattfeach storage endpoint endpoint and a type. the internal is just a different type13:06
rakhmerovwell, yes, it's another thought that I had, just to have one url for all the operations13:06
rakhmerovyes, I agree13:06
rakhmerovinternal is just one of the protocols we support13:07
rakhmerovthat's it13:07
rakhmerovalong with swift, hdfs and whatever13:07
*** alazarev has joined #savanna13:08
mattfthat seems cleaner to me13:08
nprivalovabut we use it is this way now13:10
nprivalovain job-binary we have several protocols "swift-internal", 'savanna-db' and so on13:10
*** ruhe has quit IRC13:13
mattfnprivalova, so you'd cast it as we have the primary endpoint, the job-binary and then we have a savanna-db end point for "internal" storage13:14
openstackgerritAlexander Ignatov proposed a change to stackforge/savanna: Replacement of Vanilla Hadoop 1.1.2 to Hadoop 1.2.1
mattfso maybe it's just a matter of naming getting in the way. job-binary-internals would be better named savanna-db ?13:15
nprivalovaif we have the following url in job-binary 'savanna-db://bla-bla-bla' we should find bla-bla-bla in our db among jbinternals13:16
rakhmerovbut we can't upload anything to savanna-db in an understandable way through rest API13:17
nprivalovawhy? we have PUT method to upload a file13:18
rakhmerovactually we can, using job-binary-internals but it's not clear it corresponds to savanna-db13:18
nprivalovayou don't like a name?13:18
rakhmerovyes, but we have to use job-binary-internals13:18
rakhmerovyes, I don't like it, exactly13:18
rakhmerovit doesn't correlate to savanna-db, at least at the first glance13:19
nprivalovaplease suggest a better one and contribute :) I don't like it too. it's too long13:19
rakhmerovit's not that it's long for me… It's not expressive enough :)13:19
rakhmerovjob-binary-savanna-db is even longer but it would be more correlating to the protocol "savanna-db://"13:20
*** alazarev has quit IRC13:21
rakhmerovok, think it over  please13:24
nprivalovamaybe it's better to rename only url13:24
nprivalovaif we decided to rename let's create smth better then "job-binary-savanna-db". need to think13:26
mattfnprivalova, we might not even want to have that api as a supported savanna api13:27
*** ruhe has joined #savanna13:28
nprivalovamattf, and how user will upload a file?13:29
mattfmaybe we insist on using swift or some trove managed db13:30
nprivalovaI see your point13:30
mattf...maybe savanna doesn't want to be a data storage end point itself13:31
nprivalovaWe created this level for the case when there is no swift-trove and so on in cluster13:31
mattfbut maybe it should export an hdfs storage endpoint13:31
mattfi'm on board with why we're where we are13:32
nprivalovawe have no hdfs at this point13:32
nprivalovalet me explain13:32
mattfreading the TC commentary, we're going to have increased pressure to consider what our api surface looks like as we incubate13:32
*** SergeyLukjanov has quit IRC13:32
nprivalovawe start savanna with empty db on cluster without any storages. And we'd like to run a script on transient cluster. We need to Create jobBinary at first. Cluster is not created by this moment. So we need to store a script somewhere. Internal db is a solution13:34
*** alazarev has joined #savanna13:34
*** rakhmerov has left #savanna13:38
tmckaylate to the party, I have a sick dog.... just read back13:39
tmckaymattf, there was feedback from the field (to either aignatov or akuznetsov, I forget) that there are users who commonly do not have swift set up at all, and need a place to put files.13:41
tmckaythe internal db support is partly there to serve them, too.13:41
mattfi'm not arguing against having the api13:41
tmckayI agree we can probably create a better name...13:41
aignatovI think that was feedback from Nadya13:42
tmckayaignatov, my mistake :)13:42
tmckayAlternatively, we could have a separate standalone utility for injecting files into the savanna db.  Maybe that would create some separation in minds of the users.13:43
tmckayOr, just write good documentation :)13:43
tmckaynprivalova, the truth is, those blobs can be anything at all.  We could call it "savanna-blob" or "opaque-object" or "savanna-storage" and still have create, list, get, delete functions.13:45
nprivalovayes, I mile savanna-blob :)13:46
nprivalovaI think that this REST-call (create some blob in savanna) is useful for now. We always may remove it in future. But I don't think we should do it right now13:48
mattfcrobertsrh, NikitaKonovalov, please poke
tmckaynprivalova, agreed.  The only other idea I have is to allow JobBinary objects with a url of "file:///blah/blah" which would require binaries to be stored on the savanna server.  Or even "user@host:/blah" and use scp or something to move files to the server.  Once they are on the server, they could be read and "upload_job_files()" would take it from there.13:52
tmckayThat's the only way I can see to not have a db blob interface13:53
tmckaynprivalova, do we want to rename to savanna-blob?  It's fine with me.  Maybe we should vote13:54
nprivalovavoting is ok.  Think it may be moved to weekly meeting13:56
tmckaynprivalova, so you mentioned yesterday a problem of passing parameters to scripts.  Can you elaborate?13:59
nprivalovatmckay, workflow.xml has <configuration> block and <params>, <arguments> block for Pig and Hive job-types14:01
* tmckay nods14:01
nprivalovaso we need to implement this :)14:02
nprivalovafirst of all: there is 'config' param in job and job_execution. I think it should be renamed to 'extra'. extra = {configs={…}, params = {…}, ...}14:04
tmckaynprivalova, sounds like a worthy project.  So where would params and arguments be specified by a user?  In a Job?  JobExecution?  Both, with JobExecution overloading?14:04
tmckayok, that makes sense.14:04
nprivalovatake a look at
tmckaynprivalova, hmm, I wonder looking at that picture if we also need a way to filter the list of job-origins based on job-type.14:08
tmckayIf someone selects "Pig", it seems to me the list of JobOrigins should be pruned.14:08
tmckaydifferent topic14:08
*** alazarev has quit IRC14:09
tmckaymaybe JobOrigin stores the type of the JobBinary in mains14:10
*** asavu has quit IRC14:16
*** kbroughton has quit IRC14:17
*** ruhe has quit IRC14:24
*** ruhe has joined #savanna14:32
*** kbroughton has joined #savanna14:32
*** rnirmal has joined #savanna14:32
*** SergeyLukjanov has joined #savanna14:35
*** sacharya has joined #savanna14:36
*** NikitaKonovalov has quit IRC14:39
*** NikitaKonovalov has joined #savanna14:39
*** sacharya has quit IRC14:44
tmckaynprivalova, the configs field is currently "job_configs".  Can we leave the name, and just make it nested with {"configs": {}, "params": {}, "args": {}}14:45
tmckayor maybe "job_settings"?14:46
tmckayextra sounds too vague to me14:46
nprivalovaActually I'm ok with job_configs. Not sure about other team-members14:47
tmckaynprivalova, okay, for now we'll stick with job_configs.  Thanks.14:49
mattfcrobertsrh, i forgot i'm also blocked by that keypair issue, darn15:13
mattfi may open a review for a fix later today15:14
*** sacharya has joined #savanna15:16
crobertsrhawesome.  I never really dug much.  At the time, I was just in the "get something working" mode15:22
openstackgerritJohn Speidel proposed a change to stackforge/savanna: Configuration token replacement is incorrect for some topologies
*** tmckay has quit IRC15:24
*** asavu has joined #savanna15:26
*** tmckay has joined #savanna15:26
*** tmckay has quit IRC15:30
openstackgerritNikita Konovalov proposed a change to stackforge/savanna: Floating ip assignement support
openstackgerritNikita Konovalov proposed a change to stackforge/savanna: Floating ip assignement support
openstackgerritA change was merged to stackforge/savanna-dashboard: Add neutron import available in the RDO release of Havanna
openstackgerritIlya Tyaptin proposed a change to stackforge/savanna: Added job status update and hook for transient cluster shutdown
openstackgerritIlya Tyaptin proposed a change to stackforge/savanna: Trusts for longrunning tasks
openstackgerritAlexander Kuznetsov proposed a change to stackforge/savanna: Added job status update and hook for transient cluster shutdown
*** NikitaKonovalov has quit IRC16:11
*** asavu has quit IRC16:21
*** ruhe has quit IRC16:25
*** SergeyLukjanov has quit IRC16:26
*** nprivalova has quit IRC16:30
openstackgerritA change was merged to stackforge/savanna: Configuration token replacement is incorrect for some topologies
*** kbroughton has quit IRC16:57
*** kbroughton has joined #savanna16:58
*** tmckay has joined #savanna16:59
openstackgerritAlexander Kuznetsov proposed a change to stackforge/savanna: Added job status update and hook for transient cluster shutdown
*** akuznetsov has quit IRC17:11
*** NikitaKonovalov has joined #savanna17:12
*** tmckay has quit IRC17:14
*** tmckay has joined #savanna17:14
*** kbroughton has quit IRC17:18
*** kbroughton has joined #savanna17:30
*** NikitaKonovalov has quit IRC17:46
*** NikitaKonovalov has joined #savanna17:47
openstackgerritChad Roberts proposed a change to stackforge/savanna-dashboard: Adds EDP support in the UI for job execution
*** kbroughton has quit IRC18:01
*** kbroughton has joined #savanna18:04
*** NikitaKonovalov has quit IRC18:04
openstackgerritChad Roberts proposed a change to stackforge/savanna-dashboard: Adds EDP support in the UI for job execution
*** NikitaKonovalov has joined #savanna18:04
*** NikitaKonovalov has quit IRC18:06
*** NikitaKonovalov has joined #savanna18:07
*** mattf is now known as _mattf18:07
kbroughtonI just installed savanna and savanna ui on a maces vmware vm (ubuntu 12.04) following and .  However, i don't see a savanna tab on the dashboard.  Any suggestions for troubleshooting?18:18
*** akuznetsov has joined #savanna18:20
crobertsrhkbroughton:  did you add "savanna" and "savannadashboard" to HORIZON_CONFIG and INSTALLED_APPS?  I think an apache restart is also required.18:22
kbroughtonyes, I followed those steps.  I only added them to the / file, not the / as per the instructions.  Is that correct?18:24
kbroughtonThere was this warning when starting /savanna-api :  WARNING keystoneclient.middleware.auth_token [-] Configuring auth_uri to point to the public identity endpoint is required; clients may not be able to authenticate against an admin endpoint18:24
kbroughtonalso, those files were not in the advertised location /usr/share//openstack_dashboard/ but rather /opt/stack/horizon/openstack_dashboard/18:26
*** _mattf is now known as mattf18:27
*** akuznetsov has quit IRC18:44
*** akuznetsov has joined #savanna18:46
crobertsrhkbroughton:  I'm thinking that your horizon is somehow just not finding the savanna dashboard plugin.  Any error messages in the horizon logs?18:51
crobertsrhI think the difference in file locations might be due to devstack vs openstack installs.18:51
mattfbtw, you need to add SAVANNA_URL to local_settings and the CONFIG/APPS to settings18:52
kbroughtoncrobertsrh: horizon_errors.log is full of things, but they don't look like errors.    Nothing suspicious.  [Fri Sep 13 18:05:17 2013] [error] REQ: curl -i -X GET -H "User-Agent: python-keystoneclient" -H "Forwarded: for=;by=python-keystoneclient" -H "X-Auth-Token: 85e38d7ccc164bb08acbbdfbd6748535"18:53
kbroughtonI added SAVANNA_URL = 'http://localhost:8386/v1.0' to local_settings.  What is that about CONFIG/APPS to settings?18:54
kbroughtondocs mentioned checking that 8386 port wasn't blocked.  How do i do that?18:54
crobertsrhEven if the savanna api isn't running, the dashboard would still show a savanna tab (but error out when you navigate to it)18:55
kbroughton$ telnet vmware-controller 838618:56
kbroughtonConnected to vmware-controller.18:56
kbroughtonlooks ok.18:56
crobertsrhCONFIG/APPS to settings was a reference to step 2 from the instructions that you followed18:56
kbroughtonok, yes, i did those.18:57
crobertsrhOk, this is just a guess.  Maybe a link from your horizon/openstack_dashboard/dashboards directory to your savannadashboard directory would get it working.18:58
crobertsrhshouldn't be necessary, but I'm still guessing that your horizon just isn't finding your savannadashboard install19:00
kbroughtonok, trying ln -s /usr/local/lib/python2.7/dist-packages/savannadashboard /opt/stack/horizon/openstack_dashboard/dashboards/savannadashboard19:03
crobertsrhan apache restart on top of that19:04
kbroughtondone.  still nothing on horizon.   $ ls /opt/stack/horizon/openstack_dashboard/dashboards/savannadashboard19:08
kbroughtonapi  clusters  cluster_templates  dashboard.pyc  data_sources  image_registry  __init__.pyc  job_origins  jobs  nodegroup_templates  openstack  plugins  templates  utils  version.pyc19:08
kbroughtonUsing the URLconf defined in openstack_dashboard.urls, Django tried these URL patterns, in this order:19:09
kbroughton    ^$ [name='splash']19:09
kbroughton    ^auth/19:09
kbroughton    ^home/$ [name='user_home']19:09
kbroughton    ^i18n/js/(?P<packages>\S+?)/$ [name='jsi18n']19:09
kbroughton    ^i18n/setlang/$ [name='set_language']19:09
kbroughton    ^i18n/19:10
kbroughton    ^qunit/$ [name='qunit_tests']19:10
kbroughton    ^project/19:10
kbroughton    ^settings/19:10
kbroughton    ^admin/19:10
kbroughton    ^static\/(?P<path>.*)$19:10
kbroughton    ^media\/(?P<path>.*)$19:10
kbroughton    ^500/$19:10
kbroughtonThe current URL, savanna/, didn't match any of these.19:10
kbroughtonthats what i get if i go to ip/savanna  in the browser address bar19:10
crobertsrhYeah, for some reason, your horizon denies any knowledge of savanna.19:10
crobertsrhI'm going to see if I can reproduce your situation.  I've never done a devstack + pip install savanna-dashboard install before.19:11
*** akuznetsov has quit IRC19:11
kbroughtonok thanks.  Tried to do exactly as the docs said.  Only changed pwd from nova to admin to be consistent with devstack install.19:12
crobertsrhYeah.  I hear your frustrations there.19:13
kbroughtonand I used the latest Savanna, not the stable, ala savanna-venv/bin/pip install ''19:15
crobertsrhYou could also log a bug report against the docs that lead you astray.
crobertsrhkbroughton:  Hmm.  The install worked for was on fedora though (don't have ubuntu handy to try).19:35
kbroughtonok, thanks.  I'll maybe give a centos version a whirl. and log a bug.19:42
crobertsrhsorry I haven't been more help19:43
*** tstclair has quit IRC19:44
*** NikitaKonovalov has quit IRC19:44
kbroughtonnot at all, it's a step forward for sure.  Thanks for your time.   I take it that doc bugs get logged along with code until savanna becomes an official project?19:47
crobertsrhYeah, just one big lump o' bugs19:47
openstackgerritMatthew Farrellee proposed a change to stackforge/savanna: Add complete paths in
*** tstclair has joined #savanna19:50
*** akuznetsov has joined #savanna19:55
*** akuznetsov has quit IRC19:57
*** rnirmal has quit IRC19:59
*** akuznetsov has joined #savanna20:04
*** NikitaKonovalov has joined #savanna20:11
*** asavu has joined #savanna20:15
mattfcrobertsrh, <- novaclient keypair bug20:28
*** rnirmal has joined #savanna20:30 we're not entirely crazy20:34
mattfi'm working around by commenting out the test in cluster.py20:36
mattfEstimated value of Pi is 3.1415925600000000000020:38
mattfgetting closer20:38
*** NikitaKonovalov has quit IRC20:42
tmckaycrobertsrh, fyi, I will have a CR today (late) that changes what "configs" are for Jobs and JobExecutions20:51
tmckaycrobertsrh, it probably will affect you...20:51
crobertsrhok.  Should be fine.  I haven't really done much with configs yet.20:51
crobertsrhthanks for the heads-up20:52
tmckaycool.  Instead of "job_configs" being a single level dict with just string key/value pairs, it will be a dict of dicts of string key/value pairs.20:52
tmckayjob_configs = {configs, params, args}20:52
tmckayconfigs, params, args = {key: value}20:53
tmckayglad it doesn't mess you up :)20:53
*** SergeyLukjanov has joined #savanna20:53
tmckayby the way, I saw some of the trouble earlier... I'm going to be trying devstack with a dev savanna.  Have you had any trouble there?20:54
*** akuznetsov has quit IRC20:55
crobertsrhNo, it's been good for me.20:56
crobertsrhThere is the nova keypair gotcha right now20:56
crobertsrhEasy workaround is to comment out the check20:56
crobertsrhI've been using devstack + savanna for months now20:57
*** asavu has quit IRC20:58
*** crobertsrh is now known as _crobertsrh21:00
*** akuznetsov has joined #savanna21:01
tmckay_crobertsrh, k, thanks21:01
*** ruhe has joined #savanna21:02
mattfruhe, do you have anyone looking at hadoop 2.x yet?21:08
ruhenot yet21:08
ruhei guess we'll need to implement it in vanilla plugin after 0.3 (or during Icehouse release)21:11
*** openstackgerrit has quit IRC21:14
*** sacharya has quit IRC21:15
*** ruhe has quit IRC21:21
kbroughtonI'm following the macosx option for setting up a dev env and dev dashboard .  When it comes to setting up the dashboard ui, the docs switch from multi envs to ubuntu only. .  Is it possible to run the ui dashboard on the mac to access my ubuntu devstack vm on vmware fusion?21:27
*** nadya has joined #savanna21:29
tmckaynadya, hi!21:36
*** mattf is now known as _mattf21:37
tmckayI believe I discovered an error in the test code.  When we're looking at xml workflow generation, the "assertIn" tests I think aren't good enough.  Because I suspect that dicts in Python under the hood are stored as some type of balanced tree, insertion order matters.21:38
tmckayWorking on the configs changes.... things are no longer done in the same order.  The output is correct but the assertIn string spans multiple properties and the order has changed.21:38
* tmckay may be talking to himself21:39
nadyado we have a bug in workflow creation?21:39
tmckayhi.  No, in the test.  I'll paste something...21:39
tmckay      <configuration>21:39
tmckay        <property>21:39
tmckay          <name>c</name>21:39
tmckay          <value>f</value>21:39
tmckay        </property>21:39
tmckay        <property>21:39
tmckay          <name>mapred.output.dir</name>21:39
tmckay          <value>swift://ex.savanna/o</value>21:39
tmckay        </property>21:39
tmckay        <property>21:39
tmckay          <name>mapred.input.dir</name>21:39
tmckay          <value>swift://ex.savanna/i</value>21:39
tmckaysorry for the throttling..21:40
tmckay        self.assertIn("""21:40
tmckay        <property>21:40
tmckay          <name>c</name>21:40
tmckay          <value>f</value>21:40
tmckay        </property>21:40
tmckay        <property>21:40
tmckay          <name>mapred.input.dir</name>21:40
tmckay          <value>swift://ex.savanna/i</value>21:40
tmckay        </property>""", res)21:40
tmckaySee, mapred.output.dir is now before input dir21:40
*** rnirmal has quit IRC21:40
tmckayIt's because I've changed the order of how things are constructed...21:41
*** tstclair has quit IRC21:42
tmckayso assertIn has to be smarter.  Or, we have to have some way to control order in xml generation.21:42
nadyato change an order of properties in test is ok :) Yes, I agree that assertIn should be smarter. Maybe we should test onlyon section in asserting21:43
tmckayActually, I think I can fix the insertion/update order to be the same, but that's pretty fragile21:43
nadya*only one21:43
tmckayagreed.  I'll think about it.  The good news is, I think I have the job_configs changes almost done.21:43
tmckayI should have something waiting for you Monday21:44
nadyagood news :)21:44
tmckayI may make the job_manager refactor CR dependent on this one...21:44
tmckaysince they are touching the same area in the job manager21:44
nadyait's 2 a.m. here so I need to go to sleep :)21:44
tmckayyes, go!21:44
tmckayI was just happy you popped up21:45
* tmckay time for dinner21:45
*** nadya has quit IRC21:50
*** akuznetsov has quit IRC21:51
*** sacharya has joined #savanna21:54
*** akuznetsov has joined #savanna21:55
*** SergeyLukjanov has quit IRC22:00
*** akuznetsov has quit IRC22:01
*** kbroughton has quit IRC22:26
*** kbroughton has joined #savanna22:48
*** asavu has joined #savanna22:53
*** asavu has quit IRC23:13
*** sacharya has quit IRC23:40
*** NikitaKonovalov has joined #savanna23:42
*** NikitaKonovalov has quit IRC23:46

Generated by 2.14.0 by Marius Gedminas - find it at!