Wednesday, 2016-06-22

kota_eranrom: yeah, the reason I did try to change the infra is I saw you ware fighting to gerrit :)01:54
kota_eranrom: the frustrating point of functests is we cannot get all backend logs (e.g. syslog) so that we are not able to know what's happen actually in the gerrit.01:55
kota_eranrom: At first, I have tried to dump log to the area which can be published in the gerrit result (i.e. .tox/<env name>/log normally)01:56
kota_eranrom: (at least, py27 unittest can show the logs in the dir)01:57
kota_eranrom: however, func doesn't :/01:57
kota_eranrom: that is because we don't have the publisher setting for the tox log dir in the infra :D01:58
kota_eranrom: probably aftter the patch 332065 landed, we can see all logs in the .tox/func/log dir01:59
patchbotkota_: - openstack-infra/project-config - Enable storlets to show test_result01:59
kota_and then patch 331520 is able to move swift/storlets/(perhaps ansible?) related log into the tox dir.02:00
patchbotkota_: - storlets - [WIP] Dump functest server logs into gate output02:00
kota_eranrom: finally, we can see those logs as the gate result, which absolutely helps us what's wrong with our patch ;)02:01
kota_eranrom: I got the technique in the swift3 work, likely
*** eranrom_ has quit IRC05:34
openstackgerritKota Tsuyuzaki proposed openstack/storlets: Dump functest server logs into gate output
kota_eranrom: yey, the infra chang was merged so we can see the execution logs in the gerrit gate.07:01
kota_eranrom: like this,
*** eranrom has joined #openstack-storlets08:15
eranromkota_: Gotcha. This is great! Thanks.08:18
eranromkota_: As you can see I am still fighting with the doc test :-) But the problem now is not related to the logs :-)08:19
kota_eranrom: lol08:19
kota_takashi is absent today.12:55
eranromok. so lets start.12:55
eranromI hope to be done with the doc patch soon12:55
eranromI am afraid I pushed too many things in it12:55
eranromtoo many cleanups and improvements...12:56
kota_today, i had small f2f convasation with takashi12:56
kota_and he is too.12:56
eranromanyway, once this is done, I hope to land quickly all the patches that have been queued up12:56
eranromAnything Takashi wanted to bring up??12:58
kota_not so a big one, he wanted to look at what's changed in the patch.12:59
kota_He is now working around copy thing12:59
eranromoh, great news.12:59
kota_he is now drafting which way we have to go.12:59
kota_move the pipline  position or change code for addressing recent swift change.13:00
kota_I think both ones has pros/cons13:01
eranromthe next thing I want to do is to refactor the SBus code, which is far too complex, and then get back to the range read on object node change13:01
eranromRegarding the copy - yes, there are pros and cons to both approaches. Let's see what Takashi will come up with13:02
kota_yup :)13:03
eranromRegarding the Barcelona summit, seems like we have URV and IBM on board.13:05
kota_yeah, i got your e-mail, and sorry, I didn't ack for you.13:05
kota_that's great anyway.13:05
eranromI think I have nothing else at this point13:06
kota_and he suggested we have 2 separate topics?13:06
eranromyes: one topic that will be around the Storlets use cases (which I have suuggested)13:06
kota_he wants to introduce management plane to conrole some jobs execute on swift/storlets/some sort of other component.13:07
eranromright. This is the second topic13:07
eranromThey have done a prototype of such a management system that is driven by policy.13:07
kota_alright, probably i will adress the first one.13:07
kota_yeah, that's intersting13:08
kota_btw, did you send a paper  for storlets to me?13:09
kota_i might be missing that one....13:09
eranromThey have implemented the system for a block storage and object storage systems: for the object storage part they use storlets to enforce the policy.13:09
eranromI did not send a paper. They have written some general paper on their system, and we are now starting to work on a paper which is focused on the pushdown13:10
kota_ah, ok13:10
eranromanything else for today?13:13
kota_nothing :)13:13
eranromok, so I will get back to finally finish the doc patch :-)13:13
kota_I'm looking forward to it :)13:14
openstackgerritEran Rom proposed openstack/storlets: Documentation face-lift
eranromI am finally merging the massive doc patch. Few comments15:39
eranrom1. The most recent changes were aroune:15:39
eranrom  - factory daemon changes suggested by Takashi. I managed to do every possible mistake while refactoring this :-)15:40
eranrom  - Final fixes to the ansible scripts which went through a big change15:40
eranrom  - fixing typos and some minor language improvements in the docs. Specifically, I fixed Takashi's comments regarding the api doc.15:41
eranrom2. The volume of the patch is far too large, and should be refrained in the future. I started off by fixing the docs, focusing on the installation where it was clear that the installation scripts need to be changes as well and from there I lost control15:42
eranrom3. I am sure the docks are far from being perfect, but hope that now it is in a state where incremental improvements are possible.15:43
openstackgerritMerged openstack/storlets: Documentation face-lift
openstackgerritMerged openstack/storlets: Create indiviaul handler modules under storlet_middleware
openstackgerritEran Rom proposed openstack/storlets: Refactor directory structure about gateway modules
openstackgerritMerged openstack/storlets: Refactor directory structure about gateway modules
openstackgerritMerged openstack/storlets: Dump functest server logs into gate output
openstackgerritMerged openstack/storlets: Add -v option for unit/func tests

Generated by 2.14.0 by Marius Gedminas - find it at!