Log of the #duraspace-ff channel on chat.freenode.net

Using timezone: Eastern Standard Time
* jonathangee joins07:43
* jonathangee leaves07:57
* jonathangee joins07:58
* jonathangee leaves08:57
* anusha joins09:37
* jonathangee joins09:45
<cbeer>i'm probably not going to make sprint planning09:50
and i suspect infinispan can provide a decent file-system serialization, and investigating it would probably take a day or so09:54
jonathangee: you reordered the icebox, huh? looking at it, i'm not clear, but pivotal "releases" are supposed to come after the tasks they depend on (because when you reach it, you know you're done)09:56
is that still true?
<jonathangee>that makes sense09:57
but totally isn't the way i reordered the ice box
so i guess i might have to do another reorder
i was looking at it the other way
after doing it i started to wonder if thats how releases should be used09:58
<cbeer>me too, at first. sorry, i think i failed to send note to the list about it
just in irc
<jonathangee>i'll reorder again after the sprint planning meeting
sorry bout that09:59
<cbeer>do you want to send the note i should have sent to the list about that? :)
<jonathangee>will do
<cbeer>back later. hopefully i'll catch the tail end of the call
<eddies>https://wiki.duraspace.org/display/FF/2013-01-23+Iteration+2+Sprint+Planning10:05
<anusha>Is there going to be a Skype call for the sprint planning.
<eddies>anusha, are you joining?
heh =)
<anusha>Yes - I just tried joining and it said no one else is available. I'll try again
* JasonDGI joins10:08
<eddies>https://docs.google.com/document/d/1d61s5K7LHXLLbTC25Bv2I5sRYQQyoAE-nFMQmlf6_48/edit10:11
https://wiki.duraspace.org/display/FF/Hybrid+Use+Case10:19
https://wiki.duraspace.org/display/FF/Mass+Object+Creation
https://wiki.duraspace.org/display/FF/Scale+for+High+Volume+Read+Access
* barmintor joins10:23
* jonathangee leaves10:41
<barmintor>Getting caffeine, but Neil says: "Basically, yes. Mount and command line utils. Therefore extractable without additional software." Does this disqualify Lily?
<JasonDGI>we dropped
were trying to get back in10:42
* jonathangee joins
* jonathangee leaves10:43
<JasonDGI>still having troubles
* jonathangee joins
<eddies>trying to bridge jonathan in via skype
that didnt work
trying you now jason
* ajs6f joins11:10
I'm here in IRC, if there are any questions for me.11:11
<eddies>https://www.pivotaltracker.com/projects/68482511:18
<cbeer>ok, on IRC now too
* github-ff joins11:19
[ff-modeshape-prototype] ajs6f pushed 1 new commit to master: http://git.io/nX1yng
ff-modeshape-prototype/master dbc1e7a ajs6f: Correcting MS/IS config so that tests pass, also altering FedoraRepository resource class to include path segment /fedora
* github-ff leaves
<ajs6f>That last commit was because having a JAXRS resource class with path "/" is not strictly kosher. Apparently.11:20
<barmintor>it's not? is that in the spec? CXF definitely supports it11:21
<ajs6f>I don't honestly know— Jersey beefed on it.11:22
We're using RESTEasy, which is the JBoss product.11:23
But I used Jersey to test.
<barmintor>Oh, well… *Jersey*
<ajs6f>She's such a drama queen.
Chris— I changed the FedoraRepository resource path to start with /fedora and changed the tests.11:24
At some point, we'll want to think about the repo-wide methods anyway.
I.e. how to expose them.
<cbeer>+1
<ajs6f>Now maybe Travis will stop bothering me. He, also, is a drama queen.
<cbeer>https://gist.github.com/26085aeb31db2a8ddd9311:30
ajs6f: modeshape clustering can do simple sharding, right? (distribution with number of copies = 1, right?)11:32
<ajs6f>Yes— that's how we've got it configured now.11:34
In fact, you can adjust the number of copies to make a balance between redundancy and distribution.
See:
https://docs.jboss.org/author/display/ISPN/Clustering+modes#Clusteringmodes-DistributionMode11:35
There's also straight replication:
https://docs.jboss.org/author/display/ISPN/Clustering+modes#Clusteringmodes-ReplicatedMode
and some other more obscure styles that adjust ideas of invalidation— they're more for straight use as a caching system.11:36
<cbeer>proper content distribution makes on-disk, readable-with-unix-tools use case so much less interesting to me
yeah, i really want to play with the multi-tiered caching.11:37
<ajs6f>I don't get it, Chris— I thought the guarantee is about being able to recover without Fedora— what does distribution have to do with it?
I can look at the multi-tier architecture. It's pretty flexible. It even offers local ("L1") facility.
<cbeer>ajs6f: maybe it's come up on the thread, but it makes me more willing to have that non-fedora-recoverable scenario offered as an external service, not dictating core concerns11:38
<ajs6f>Oh, right. I'm 100% on board with that. I thikn Eddie put it very well.
<cbeer>ok, i've lost track of that thread. about when i woke up and there were 3 or 4 different subthreads going on11:39
<eddies>%22
%22
<cbeer>https://gist.github.com/26085aeb31db2a8ddd93
<jonathangee>i like bens suggestion this be brought up in the steering group11:40
re: non-fedora-recoverable
<ajs6f>+1 to JG's +1 of Ben11:45
Am I missing the Databank 5-thread results, or did that not get run?11:46
<cbeer>ajs6f: sorry, i said it on the call
no databank 5-thread results because i couldn't get databank deployed in a way that'd make that fair11:47
<ajs6f>Fair enough.
Wow, Lilly seems to be on meth!
<JasonDGI>???11:48
<ajs6f>If I'm not completely misreading that result (which is very possible) it's really fast.
<jonathangee>Fedora 4: methdora
#methdora11:49
<JasonDGI>methdora for hackfest!! w00t!
<ajs6f>So this is about moving from methadone to methamphetamine.
<cbeer>ok, i'm heading down into the concrete dungeons of UCSD. probably will drop.
<JasonDGI>methamfedoramine11:51
<barmintor>+1 to as6f's +1 of JG's +1 to me: And the circle is complete. Amen.11:55
<jonathangee>+1
<cbeer>jonathangee: i think we should add a new release for "finish making a candidate selection"?11:58
<jonathangee>i agree. although we decided to split this sprint into two smaller sprints11:59
<cbeer>+1
are we doing another sprint planning next week? or trying to plan them both out now?
<jonathangee>so that the end of this sprint, we pick a candidate12:00
then we plan the next sprint next week when we have a candidate
<cbeer>+1
<jonathangee>so i'm not sure if a release makes sense, since its on the sprint boundary anyway12:01
but maybe it does
<cbeer>:/ i just deleted my jgroups config that does jdbc_ping12:02
<ajs6f>Why?
<cbeer>i was pulling your code
<ajs6f>Oh, fooey. Weren't you working in a branch?12:03
<cbeer>doesn't matter, easy enough to recreate
(and, i dropped)
<ajs6f>Let me know if you need/want any help.12:04
<cbeer>(oh, and now i remember, it doesn't really matter.. i had spun up an AWS RDS instance that i've since killed)
<ajs6f>There's a prebuilt AWS EC2 JGroups config in ModeShape. Maybe use that?
<cbeer>that's the S3 ping?12:05
<ajs6f>Yeah.
I think it
's jgroups-ec2.xml
<cbeer>yeah, if you read the docs, they say use JDBC_PING instead so it's more portable12:06
<ajs6f>Oh, yeah. And we'd want to think about our design from our own standpoint, not AWS's. {grin} I just mean to get up a running quickly.
afk12:15
* ajs6f leaves12:18
* ajs6f joins12:19
bak12:20
<jonathangee>https://github.com/futures/RDFDatabank/tree/launchd/docs12:21
<barmintor>I LOVE WORKING ON THE LIBRARY'S STRATEGIC PLAN AND REVIEW FOR 2013-201612:22
<jonathangee>where are we with the docs for this ticket:12:24
https://www.pivotaltracker.com/story/show/41238629
eddie and I are doing a ticket cleanup
<cbeer>those are not the current docs.
i have stuff in my own branch that i think eddies was following12:25
but i got stuck on the background workers, i think
ajs6f: do you know what version of infinispan we're using?12:26
i'm trying to get the copy of jgroups-tcp we're using12:27
but apparently the 5.1.x and master versions in github are not it.
<ajs6f>5.1.2.Final
<jonathangee>can you make some notes on the ticket cbeer? and merge the latest you have in? databank osx is on the back burner for now but we don't want to lose what we have
<ajs6f>You want me to send it to you?
<cbeer>ajs6f: please.
<ajs6f>See email12:28
<cbeer>jonathangee: i think we can merge from https://github.com/cbeer/RDFDatabank/tree/5cd82835accfe9aaa92e91834970d6bf1214d82f12:29
later commits in that branch break the background workers
(of course, the documentation there doesn't tell you to fix those workers either..)
<jonathangee>can you make a pull request?
<cbeer>oh, duh.12:30
<eddies>cbeer: what's the status of Refine Akubra/Glacier to maintain a local copy of the Glacier inventory?12:31
https://www.pivotaltracker.com/story/show/42226567
it's marked finished12:32
<cbeer>eddies: finished, waiting for someone to deliver
<eddies>where's the code?
<cbeer>i think barmintor was the requestor
https://github.com/futures/akubra-glacier/blob/master/src/main/java/org/fcrepo/akubra/glacier/GlacierInventoryManager.java12:33
<barmintor>the code is in the InventoryManager...
^^ right there^^
<eddies>barmintor: any objection to delivering this?12:34
<barmintor>delivered, just wanted to check that the update from glacier inventory was called12:35
<eddies>cbeer: any objection to delivering Relationships Hints Provider for Akubra (https://www.pivotaltracker.com/story/show/41258587)?12:36
cbeer: also, what about "Update databank workers to look at the environment inis, not a bespoke config" (you're both owner & requester)12:37
<cbeer>i haven't looked at it. i'm not sure i could validate it even if i did
incomplete.
that's the thing i mentioned doesn't work, which is why i didn't merge my branch's docs back into futures/RDFDatabank12:38
<barmintor>https://github.com/futures/akubra-glacier/blob/master/src/test/java/org/fcrepo/akubra/glacier/TestGlacierHintsProvider.java
https://github.com/futures/akubra-glacier/blob/master/src/test/java/org/fcrepo/akubra/glacier/TestGlacierBlobStore.java
<cbeer>ok, if there's a test file that claims to test the future, i assume it works :P12:39
<barmintor>https://github.com/futures/akubra-glacier/blob/master/src/test/java/org/fcrepo/akubra/glacier/TestGlacierBlobStore.java#L70
<cbeer>(sorry, i'm working at UCSD all this week, etc,etc)
<barmintor>I'm just saying the hints provider is done. I do have to test the fcrepo side, but that's really another issue12:40
<eddies>barmintor: is Implement FCRepo classes to make use of Glacier Akubra store done?12:46
i thought you were going to do that on our fork of fcrepo
<barmintor>Mmm… no
I mean- yes, I will do it on our fork. No, it is not done. Yes, it is begun.12:47
The LLStore module is done
but not tested
also, the current impl actually blocks while the data is retrieved, so that has to change12:50
<cbeer>ajs6f: i'm having trouble with the ff-modeshape-prototype master code.. getting 404s on:
curl "http://localhost:8080/fedora/objects/fdsa" -X POST
<JasonDGI>the jmeter lily test is 404ing on delete for me, anyone else get that?
<ajs6f>Let me take a look...
<cbeer>thanks. i should turn on password-based auth to this AWS instance12:52
<ajs6f>Oh, wait— I see. I only changed the FedoraRepository resource paths.
Not any others
So you're looking for :8080/objects/dsg
<cbeer>ok, i can live with that
<ajs6f>We should make the paths make sense.
Sorry about that.
Maybe I will rejigger FedoraFRepository again. Factor it into "/describe" and "/objects"12:53
afk13:02
back13:06
<cbeer>eddies: re: http://www.pivotaltracker.com/story/show/4282174713:09
there's other tickets to make fedora/akubra asynchronous
<anusha>Just spoke to my system administrator. The 6 VMs should be ready by end of day tomorrow13:12
Would 40 GB hard disk be sufficient
<cbeer>anusha: more would be better.13:17
but we could probably start there.
<anusha>How much more
<cbeer>if your system is anything like ours, it's pretty easy to add more later
100G?
<anusha>yes, but I'd rather we asked for something reasonable13:18
I just asked, and 100G should be fine.13:20
<cbeer>thanks
* anusha leaves13:27
* ajs6f1 joins14:20
* ajs6f leaves14:21
<ajs6f1>back14:32
* github-ff joins14:44
[ff-modeshape-prototype] ajs6f created ModeShapeRest (+1 new commit): http://git.io/n5jB_w
ff-modeshape-prototype/ModeShapeRest c04de5d ajs6f: ModeShape REST API no up alongside Fedora
* github-ff leaves
<ajs6f1>cbeer?
<cbeer>ajs6f1: pong
<ajs6f1>See that branch?
The idea is:
I brought up the ModeShape REST API alongside Fedora.14:45
So you can use either.
<cbeer>+!
<ajs6f1>It's nice for diagnosis...
and I wanted to start figuring out how to get at repos that your app hasn't built.
Because I was thinking to start a n event publishing module.14:46
Both to "look like" Fedora JMS,
and to start going down the road that I understand us to have chosen:
"archival" persistence as a coupled service, not as part of the xactional core.
And to do that we'll need to know how to get at the repo from another module.14:47
Which we now do.
Do me a favor and sanity check it before I merge it?
Just run it and see if you can use both APIs.
E.g.
curl localhost:8080/objects/foo -X POST
(Fedora)
should make a "foo" item show up in:14:48
curl -H "Accept:application/json" localhost:8080/repo/fedora/items
(ModeShape)
Meanwhile I've got a conference call, aka an opportunity to go fool around with Apache Camel and see how to make it receive JCR events.14:49
That was easy:14:51
http://camel.apache.org/jcr.html
I loooove Camel.
afk14:55
* ajs6f1 leaves
* ajs6f joins15:05
Back
<cbeer>ajs6f: sorry, working with UCSD this week. i'll try to find time, but i'd vote for merging even without my opinion.15:12
it sounds like a good idea to me, at least
<ajs6f>No prob. I'll leave it for a bit.
Here's another quick quesiton, tho': would you be all right with moving _all_ of our current URLs down one level to /rest/
e,g,
:8080/rest/fedora/describe15:13
:8080/rest/objects/
and for ModeShape
:8080/rest/repo/fedora/items
etc.
?
<cbeer>doesn't matter to me. i'm not sure what the use case is, though?15:14
<ajs6f>For this event publishing experiment, I'd like to throw up an RSS feed. But there's nowhere left to stick it in the URL tree.15:23
The URL tree is alreay completely owned by RESTEasy.
So I ant to move RESTEasy down to /rest/, as described.
We;ll need to do that for the SOAP endpoint, of course.;-)15:24
* github-ff joins15:25
[ff-modeshape-prototype] ajs6f pushed 1 new commit to ModeShapeRest: http://git.io/E9NOJw
ff-modeshape-prototype/ModeShapeRest 9cde475 ajs6f: Added some minor logging
* github-ff leaves
* JasonDGI leaves15:27
* github-ff joins15:31
[ff-modeshape-prototype] ajs6f pushed 1 new commit to master: http://git.io/OnJr6A
ff-modeshape-prototype/master 4016084 ajs6f: Moved all services down one level of URL to /rest
* github-ff leaves
* github-ff joins15:50
[ff-modeshape-prototype] ajs6f pushed 1 new commit to master: http://git.io/aofAiA
ff-modeshape-prototype/master 289a576 ajs6f: Updated README
* github-ff leaves
* github-ff joins15:53
[ff-modeshape-prototype] ajs6f pushed 1 new commit to master: http://git.io/RXIYHQ
ff-modeshape-prototype/master 45f922e ajs6f: Updated README
* github-ff leaves
* anusha joins16:18
* jonathangee leaves16:30
<ajs6f>O and O16:33
* ajs6f leaves
<anusha>cbeer: Do you know if the latest jmx file for testing fedora with the govdocs is in github16:42
<cbeer>yes
https://github.com/futures/ff-jmeter-madness16:43
* jonathangee joins16:46
<anusha>Ah Thanks! Just realised that FuturesJmeterAggregation.jmx has the tests for modeshape, fedora and databank
* jonathangee leaves16:52
* github-ff joins16:55
[ff-modeshape-prototype] ajs6f created Spring (+1 new commit): http://git.io/2a2Cqg
ff-modeshape-prototype/Spring a1cc546 ajs6f: Refactored Bootstrap slightly for clarity
* github-ff leaves
* anusha leaves17:45
<cbeer>ajs6f++18:22
i got modeshape going on the ec2 instance I have running18:36
(and have probably cost us a beer or two already, at ~ $0.08/hour)
the eventual consistency piece is pretty weird, but can clearly be tuned using a bunch of cryptic parameter names18:38
(and, i probably shouldn't call it eventual consistency.. just replication delay)18:41
* cbeer leaves18:51
* cbeer joins
* eddies leaves18:59
* eddies joins19:06
* eddies leaves
* eddies joins
* jonathangee joins19:14
* jonathangee leaves
* github-ff joins19:34
[ff-modeshape-prototype] cbeer opened pull request #21: Mode shape rest (master...ModeShapeRest) http://git.io/KpbHJA19:35
* github-ff leaves
* github-ff joins19:36
[ff-modeshape-prototype] cbeer deleted InfinispanClustering at 4bc434c: http://git.io/SrDX2w
* github-ff leaves
* github-ff joins
[ff-modeshape-prototype] cbeer deleted TransMan at 9ff32af: http://git.io/MDRa2g
* github-ff leaves
* github-ff joins
[ff-modeshape-prototype] cbeer deleted context-listener at 2986adc: http://git.io/8GzrKA
* github-ff leaves
* github-ff joins19:38
[ff-modeshape-prototype] cbeer deleted clustering at 8bf590c: http://git.io/TNwQ2w
* github-ff leaves
* github-ff joins
[ff-modeshape-prototype] cbeer pushed 1 new commit to master: http://git.io/3uVong
ff-modeshape-prototype/master c61bc4d Chris Beer: Merge pull request #21 from futures/ModeShapeRest...
* github-ff leaves
* jonathangee joins20:16
* jonathangee_ joins20:31
* jonathangee leaves20:33
* jonathangee joins20:35
* jonathangee_ leaves20:38
* jonathangee leaves20:39
* jonathangee_ joins
* jonathangee joins20:43
* jonathangee_ leaves20:45
* jonathangee_ joins20:50
* jonathangee_ leaves20:51
* jonathangee leaves20:54
* jonathangee joins20:57
* jonathangee leaves21:04