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

Using timezone: Eastern Standard Time
* kaarefc joins01:27
* ksclarke leaves01:32
* kaarefc leaves02:02
* kaarefc joins04:54
<pivotal-bot____>Mike Daines started "Fix export URL" https://www.pivotaltracker.com/story/show/6088375207:27
* ajs6f joins08:24
* github-ff joins
[fcrepo4] fasseg created jgroups-min-credits (+1 new commit): http://git.io/i9hu8w
fcrepo4/jgroups-min-credits a977771 fasseg: added min_credits to jgroups tcp config, since a warning popped up saying that the fragmentation size was larger than min_credits and this could lead to deadlocks
* github-ff leaves
<ajs6f>awoods: ping08:26
* github-ff joins08:29
[fcrepo4] fasseg force-pushed jgroups-min-credits from a977771 to 40ec3eb: http://git.io/s_9qFA
fcrepo4/jgroups-min-credits 40ec3eb fasseg: added min_credits to jgroups tcp config, since a warning popped up saying that the fragmentation size was larger than min_credits and this could lead to deadlocks
* github-ff leaves
* github-ff joins08:30
[fcrepo4] fasseg opened pull request #231: Small config change to remove a possible issue with JGroups (master...jgroups-min-credits) http://git.io/tqGyqQ
* github-ff leaves
<ajs6f>cbeer: Have you seen this:08:39
Hope that link works.
Marmotta is (internally) debating their impl of LDP. We might want to look over their shoulders.
* travis-ci joins08:41
[travis-ci] futures/fcrepo4#1520 (jgroups-min-credits - a977771 : fasseg): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/commit/a977771d0d5f
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/18651044
* travis-ci leaves
* travis-ci joins08:45
[travis-ci] futures/fcrepo4#1521 (jgroups-min-credits - 40ec3eb : fasseg): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/a977771d0d5f...40ec3eb2d7f7
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/18651286
* travis-ci leaves
* ajs6f leaves08:50
* ajs6f joins08:55
* tecoripa joins09:14
* kaarefc leaves09:15
* tecoripa leaves
* github-ff joins09:17
[fcrepo4] awoods pushed 1 new commit to master: http://git.io/-e_iHQ
fcrepo4/master c1bb4d1 Andrew Woods: Merge pull request #231 from futures/jgroups-min-credits...
* github-ff leaves
* tecoripa joins09:18
* ermadmix joins09:21
* tecoripa leaves
<pivotal-bot____>Andrew Woods edited "Create unit tests for auth-common" https://www.pivotaltracker.com/story/show/6549233809:23
Andrew Woods edited "Create unit tests for auth-roles-basic" https://www.pivotaltracker.com/story/show/6549210209:24
Andrew Woods edited "Provide pluggable strategies for updating resource properties" https://www.pivotaltracker.com/story/show/6521084209:26
* osmandin joins
<pivotal-bot____>Andrew Woods edited "Enable auth-war maven profile to build" https://www.pivotaltracker.com/story/show/6293851609:29
<awoods>ajs6f: I assume we can "unstart" this ticket: https://www.pivotaltracker.com/story/show/6521084209:33
<pivotal-bot____>feature: Provide pluggable strategies for updating resource properties (started) / owner: A. "Flambulacio" Soroka
* kaarefc joins
* gregjansen joins09:34
* travis-ci joins
[travis-ci] futures/fcrepo4#1523 (master - c1bb4d1 : Andrew Woods): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/fbb2520e7923...c1bb4d17788c
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/18654440
* travis-ci leaves
<pivotal-bot____>A. "Flambulacio" Soroka unstarted "Provide pluggable strategies for updating resource properties" https://www.pivotaltracker.com/story/show/65210842
<awoods>fasseg: Would you mind creating a more detailed ticket that describes exactly what you are trying to achieve with the SCAPE clustering work?09:35
fasseg: The following has become a bit "stale": https://www.pivotaltracker.com/story/show/6225715009:36
<pivotal-bot____>feature: Feature Documentation: Clustering (started) / owner: Frank Asseg
<fasseg>awoods: something like: assess performance of different confgurations and API access patterns in a clustered environment?
<awoods>fasseg: or, if you could detail the target performance, that could be helpful.09:37
<awoods>fasseg: Also, is there anything we can do on this ticket? or is it OBE? https://www.pivotaltracker.com/story/show/5850997809:38
<pivotal-bot____>feature: Analyze SCAPE Report (unscheduled) / owner: Frank Asseg
* kaarefc leaves09:40
<fasseg>ok, will do at the momen: requirement it's 15 mill objects / day and I need 783secs to only *create* 1000 objects that's a factor 2 without creating any properties/binaries ...09:41
I guess matthias can start on the reevaluation as soon as I get a satisfiying performance out of the cluster, so the ticket is still valid09:42
and every little screw I try to tune which actually improves performance, leads to error states evetually: e.g. fsyncMode="default" instead of "perWrite" gives you Too Many Open Files errors from the OS :/09:43
* kaarefc joins
<fasseg>setting async replication mode leads to ParentNodeNotFoindException ....I guess because of reordering ...09:44
Oh and I still have to have some strange behaviour in there which does not allow me to ingest two things in parallel using the SCAPE implemtnation on top of fedora, but that's hopefully an impl issue09:45
<gregjansen>fasseg: have you tried setting a higher open file limit? We have to do that here sometimes. The default on many distributions is 102409:49
<fasseg>gregjansen: yeah but that just pushes the problem to a later point in time
<gregjansen>fasseg: oh I see, so there are file handle leaks09:50
<fasseg>I guess not, but tusing async with an object that takes a long time to create, open file handles are put on top of a stack/queue or sth and if the datastrucutre is not emptied faster than requetss are incoming, it's a problem.09:51
so not a classical leak where some file handles remain open unintentionally
i tried 250k and still ran into the problem after a couple of minutes09:52
<gregjansen>fasseg: wow
<ajs6f>fasseg: Re: the problems with ParentNodeNotFoindException, have you tried using transactions, or are we still missing transactions in clustered environments?
<pivotal-bot____>A. "Flambulacio" Soroka added "Attempt to modularize indexer using OSGi" https://www.pivotaltracker.com/story/show/6555737809:54
A. "Flambulacio" Soroka started "Attempt to modularize indexer using OSGi" https://www.pivotaltracker.com/story/show/65557378
<fasseg>ajs6f: nah using a transaction that spans the whole parent/child creation works fine, but it introduces a long wait time for the other nodes in the cluster to get to do something sine they only get the state transfer as soon as the whole thing is ingested n the first node, whic leads to a serial ingest behaviour that leaves nodes without work 1/2 of the time
* kaarefc leaves
<ajs6f>fasseg: Can you parallelize the transactions across nodes, or is everything coming through a single load-balancer?09:55
<fasseg>so I added a couple of session.save() calls at strategic positions, but the of course you lose the possibility to rollback ;)
all from one loadbalancer so 1ingest <-> 1node
<ajs6f>fasseg: It might be worth rethinking that design. I can see the "one load-balancer to rule them all and bind them" as appropriate for access, but what about splitting ingest across all the nodes and using transactions also?09:56
<awoods>fasseg: I do not think session.save() has any visible effect outside of the transaction until you commit the transaction.09:57
<fasseg>ajs6f: but that introduces all sorts of state handling for ingest on the layer on top of fedora something I though infinispan would take care of for us..09:58
<ajs6f>fasseg: I don't see how it would: it's just a queue with a splitter on the end.09:59
<fasseg>awoods: nah session.save() has an effect, I can control when the work gets pushed to other nodes using this call, and the Fedora Transaction does nothing more that saving the session on a commit() call when I rmemeber correctly
ajs6f: parent -> child and references in Sparql form one to the other
ajs6f: that would mean I have to create all the objects fist and then fire Sparql/Property updates against them after ingestion, since I cannot attah properties to non exitent nodes...10:00
<ajs6f>fasseg: Sure, but above Fedora, it's just a couple of messages that assemble a part of the repository that all have to go to the same node. (A transaction)
* mikeAtUVa joins
<ajs6f>fasseg: Can't you build a node and fill its props in the same xaction?
<fasseg>you're right I can, but the spanning the tx over the cluster nodes, would mean all the states needs to be transferred also, right?10:01
* github-ff joins
[fcrepo4] ajs6f pushed 1 new commit to UpdateTactics: http://git.io/32PmZw
fcrepo4/UpdateTactics 38d1b55 ajs6f: Leaving in unfinished state
* github-ff leaves
<ajs6f>fasseg: That's why I'm suggesting _not_ spanning the tx over nodes. Make lots of txes, send each one to a single node. That's why the load balancer would work.10:02
<fasseg>What I can't do is reference node a from node b when they're on diffrerent clusters or somehow the nodes can access the to-be-created node in the tx or session cache10:03
*different nodes
so I can not tell node a to create object a and node b to create object b with a ref to object a since it might not yet exist and I dont know it's URI10:04
...it plonks and gives me a PrentNodeNotFoundException...wouldn't it?
<ajs6f>Right, but if the txes each contain _all_ of the node creations and _all_ of the prop creation for a given partition of repo, each one should fulfill successfully on the single node to which I am imagining you would send it, and after you close the tx the complete delta should propagate. The special sauce is making sure that each tx is responsible for constructing a _partition_ of repo, one with no relationships going in or out.10:05
I assume that to be possible… but maybe SCAPE's graph is so rich it is not?
Also, you may consider ordering the txes… that is, ordering the creation of the repo from top to bottom.10:06
<fasseg>nah the graph is not very dense, it just parent-child relation ships on 3 levels
<ajs6f>Okay, so we could partition it "horizontally", right?10:07
* kaarefc joins
<ajs6f>A given "grandchild" should only require the existence of the parent (which would require the existence of the grandparent).
<fasseg>so you mean creating the IDs in a layer on top of fedora4 and just reference these IDs in the child nodes....
<ajs6f>But not siblings.
fasseg: No, I mean… hang on, this is one of those things that is _much_ easier to draw than put into words. Let me see if I can draw something and put it up somewhere.10:08
<fasseg>should def work using jcr properties
<ajs6f>Does anyone know if we have drawing software on the wiki?
<fasseg>sorry if im a bit slow today I still feel a bit sick :)10:09
<awoods>ajs6f: yes
<ajs6f>awoods: thanks
* ksclarke joins
<ajs6f>fasseg: NP— this is tricky stuff and I don't pretend to have it all figured out.
<awoods>ajs6f: in wiki, edit page, insert->Gliffy Diagram10:10
<ajs6f>awoods: Just found it— thanks
<gregjansen>I find that modeshape 3.7.1.Final depends upon infinispan version 5.2.7.Final and jgroups 3.2.10.Final. However in fcrepo/pom.xml we specify infinispan 5.3.0.Final and jgroups 3.4.0.Final.. Does anyone know if there is a reason why modeshape hasn't upgraded?10:11
<ajs6f>I think I'll actually use an org chart template. :)
<awoods>gregjansen: nope, it may be worth asking on the modeshape irc.
<gregjansen>awoods: good idea10:12
* kaarefc leaves
* travis-ci joins10:13
[travis-ci] futures/fcrepo4#1524 (UpdateTactics - 38d1b55 : ajs6f): The build has errored.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/1ba154b9b367...38d1b55180a1
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/18657468
* travis-ci leaves
* kaarefc joins10:14
<pivotal-bot____>Andrew Woods edited "Provide pluggable strategies for updating resource properties" https://www.pivotaltracker.com/story/show/65210842
<fasseg>awoods: btw what should our default cluster config be? The only one I can get to run consitently is using UDP Multicasting for discovery and TCP unicasting for state transfer...shall I just add a PR for that?
<pivotal-bot____>Andrew Woods added comment: "https://github.com/futures/fcrepo4/compare/1ba154b9b367...38d1b55180a1" https://www.pivotaltracker.com/story/show/6521084210:15
<fasseg>and the one in fcrepo4 does use TCPPING for discovery which requires you to know the IPs of all the nodes beforehand
<pivotal-bot____>Andrew Woods edited "Provide pluggable strategies for updating resource properties" https://www.pivotaltracker.com/story/show/65210842
<awoods>fasseg: gregjansen and tecoripa are starting to try clustering10:16
fasseg: maybe we wait a moment to see which configs they are able to get working.
<gregjansen>fasseg: that UDP advice is interesting. I remember that I had trouble with TCPPING back in Nov.10:17
<fasseg>gregjansen: you can check out this one: https://github.com/fasseg/fiz-fcrepo-cluster-config
there's a jgroups-fedora-tcp.xml that does discovery using MPING instead10:18
* kaarefc leaves10:19
<pivotal-bot____>Scott Prater started "Set up fcrepo-test-profiles on UW cluster" https://www.pivotaltracker.com/story/show/65507426
<gregjansen>fasseg: so it sounds like the #modeshape team does not systematically test against more recent versions of infinispan than you see in their pom. So infinispan may be backward compatible, but they haven't tested against it systematically. (We are a few versions ahead of them currently)10:22
"rhauch 10:21:49 AM10:23
gregjansen: I don't think it's low-risk, considering we've done basically nothing with 5.3.0.final"
<pivotal-bot____>Frank Asseg added "Check perfomance according to the SCAPE requirements from the ONB" https://www.pivotaltracker.com/story/show/6555990610:24
Frank Asseg started "Check perfomance according to the SCAPE requirements from the ONB" https://www.pivotaltracker.com/story/show/65559906
<gregjansen>fasseg: Perhaps using the latest infinispan is adding some complications?
<ajs6f>fasseg: https://wiki.duraspace.org/display/FF/Scratch+-+Partitioning+a+Repository+into+Transactions+for+Scaled+Ingest10:25
Each colored box is a transaction (sent to a different node).
The transactions run from dark to light.
<fasseg>gregjansen: I can certainly try but I had the same issues on the AWS cluster and on my small home cluster before the switch to the new ispn version I think...
<ajs6f>(All transactions of one color run before any from a lighter color.)
<awoods>gregjansen/fasseg: it seems safest to use the same ispn version that modeshape suggests.
<ajs6f>Blue boxes, that is. The green and white boxes, obviously, are nodes.10:26
<gregjansen>awoods: yeah, not spending cycles on a stack that modeshape hasn't tested yet makes sense
awoods: did the latest infinispan advertise any features we particularly needed?10:27
<fasseg>ok I think I understand correctly then, you create a partition plan beforehand and then start creating the objects from top to button in a queue like fashion
ajs6f: ^^
<ajs6f>fasseg: yeah, that's the soul of it. The other parts are:
The other part, rather, is, parallelization.10:28
Transactions encompass the creation of a given partition.
<awoods>gregjansen: I am not sure. If we are lucky, the git commit comment for that version change in our repo will say.
<ajs6f>And can therefore run in parallel.
(Obviously not through a load-balancer, preferably, unless you are sure to get a perfect round-robin and your partitions are strongly balanced.)10:29
(Although maybe you could do it through a l-b without much loss of troughput. Hard to say.)
<fasseg>redirection would make the most sense I guess
<ajs6f>Sure. Up to you. The important thing is to spread the partition-creation.10:30
<fasseg>but you can not have a single endpoint then and have to have access to all nodes
<ajs6f>Once the full ingest is done, you should be able to.
<gregjansen>awoods: cbeer may have bumped it to improve leveldb config somehow? It seems tied to a leveldb config
* mikeAtUVa leaves
<ajs6f>Not _during_ ingest, though.
<ajs6f>I'm assuming INSPN clustering in play.
So the completion of a tx results in the broadcast of a delta.
1 partition -> 1 tx -> 1 node, but then all nodes, once the tx completes.10:31
And no node should ever be waiting.
<fasseg>that's a non issue since the top layer has to keep track of the ingestion state any way for an async ingestion endpoint we have in SCAPE
<ajs6f>A non-repo endpoint?10:32
<fasseg>a non fcrepo scape endpoint
* mikeAtUVa joins
<ajs6f>So, it needs to know when a node is "fully" existent (exists on all nodes)? That's true some time after tx completion, but if you're running async, you can't know exactly when.10:33
If you're will to let that workflow lag the main ingest, you could just make sure if waits some really long time, like a couple of seconds.
<fasseg>just wait for all the txs to complete and update a "lifecyclestate" property on the parent node
then you know when the node is complete...10:34
<ajs6f>Sure, absolutely. Then the other workflow can attack all the nodes underneath that one.
I'm not sure what your additional workflow needs to do or know, so I'm not sure what would be best, but it sounds like you know.10:35
afk bbs
<fasseg>almost nothing I have all the info on the top layer since all the relevant data is inside the POSTed XML file
as long I can be sure that a node has a certain ID and Path I can cmpletely parallelize that10:36
or rather *will have*
btw: Is there an API call to get the current node hosts?10:38
<ajs6f>fasseg: You mean hostname?10:39
<ajs6f>Not from MODE, I wouldn't think.
* tecoripa joins10:40
<ajs6f>Maybe it's in the root RDF? There's a huge amount of stuff in there.
<fasseg>Ill have a look where the cluster info on the main page comes from...
or is this just the node names -> like host-28281 or sth?
<ajs6f>I think from https://wiki.duraspace.org/display/FF/Scratch+-+Partitioning+a+Repository+into+Transactions+for+Scaled+Ingest10:41
That's what I meant
<pivotal-bot____>Eric James added comment: "one finding: SPARQL query support is very limited: ""10:52
JQLQueryVisitor.visit(final ElementPathBlock el)
-if su..." https://www.pivotaltracker.com/story/show/64499514
<fasseg>ajs6f: i actually ctrl-t'd it in eclipse so found the class but it seems no ips or hostnames in there, but I will have to check after the call10:54
awoods: omw, might be three minutes late10:57
afk bbs
* ajs6f leaves
<pivotal-bot____>Andrew Woods added comment: "@ericjames, for this ticket, if you can verify that @stefanoc's query works on an external triplestore, the..." https://www.pivotaltracker.com/story/show/6449951410:59
* mikeAtUVa leaves
* mikeAtUVa joins11:02
<pivotal-bot____>Eric James added comment: "Current hypothosis on why Stefano's query doesn't is is that the converter changes fcrepo:hasParent to jcr:ha..." https://www.pivotaltracker.com/story/show/64499514
<pivotal-bot____>Frank Asseg added comment: "Setting async in infinispan config leads to ParentNotFoundException" https://www.pivotaltracker.com/story/show/6555990611:05
<mikeAtUVa>Having technical problems... no audio (mic or speakers)... my report is that I've started going through the use cases for beta features to see where we stand.11:06
<pivotal-bot____>Frank Asseg added comment: "Setting "perWrite" for infinispan's fsyncMode leads to "Too Many Open Files" errors form the OS, even if set..." https://www.pivotaltracker.com/story/show/65559906
<mikeAtUVa>I also spent time researching and making suggestions for Kai regarding the transaction issue.
* ajs6f joins11:08
* ajs6f1 joins11:12
* ajs6f leaves
<pivotal-bot____>Scott Prater started "Provision UW cluster nodes" https://www.pivotaltracker.com/story/show/5994376811:14
Scott Prater added "Create deployment tool for clusters" https://www.pivotaltracker.com/story/show/6556592611:21
* mikeAtUVa leaves11:22
* mikeAtUVa joins11:23
<pivotal-bot____>Mike Daines unstarted "Create unit tests for auth-roles-common" https://www.pivotaltracker.com/story/show/6549211011:25
Gregory Jansen added "Consolidate war configuration files" https://www.pivotaltracker.com/story/show/6556663411:27
Gregory Jansen edited "Consolidate war configuration files" https://www.pivotaltracker.com/story/show/65566634
Scott Prater unstarted "Set up fcrepo-test-profiles on UW cluster" https://www.pivotaltracker.com/story/show/6550742611:28
<ajs6f1>Dropped connection— I was done anyway.11:31
<tecoripa>gregjansen: I'm reading your wiki page... looks good11:34
gregjansen: based on today's conversations, do we want to break it up into three parts, maybe three different apps?11:35
* ajs6f1 leaves
* ajs6f joins
<tecoripa>gregjansen: I have some ideas that we could talk about quickly, if you're around
<gregjansen>tecoripa: do you want to meet in hangout?11:36
<tecoripa>sure, same one?
* ajs6f leaves11:47
* ajs6f joins11:54
* github-ff joins12:09
[fcrepo4] ajs6f created DiamondOp (+1 new commit): http://git.io/1qkUrg
fcrepo4/DiamondOp f4daad5 ajs6f: Code cleanup
* github-ff leaves
<pivotal-bot____>A. "Flambulacio" Soroka added "Code cleanup" https://www.pivotaltracker.com/story/show/65570296
A. "Flambulacio" Soroka started "Code cleanup" https://www.pivotaltracker.com/story/show/65570296
A. "Flambulacio" Soroka added comment: "https://github.com/futures/fcrepo4/pull/232" https://www.pivotaltracker.com/story/show/65570296
A. "Flambulacio" Soroka finished "Code cleanup" https://www.pivotaltracker.com/story/show/65570296
* github-ff joins
[fcrepo4] ajs6f opened pull request #232: Code cleanup (master...DiamondOp) http://git.io/h7zL2A
* github-ff leaves
<pivotal-bot____>A. "Flambulacio" Soroka edited "Code cleanup" https://www.pivotaltracker.com/story/show/65570296
<ajs6f>awoods: ^^^ that PR is totally non-functional. Just using the "diamond operator" <> and removing declared exceptions we don't actually throw.12:11
<awoods>ajs6f: on a call
<osmandin>awoods: let me know when you have time to chat12:13
* ajs6f leaves12:18
* ajs6f joins12:20
* travis-ci joins12:29
[travis-ci] futures/fcrepo4#1525 (DiamondOp - f4daad5 : ajs6f): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/commit/f4daad5f65fb
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/18666424
* travis-ci leaves
do you have time now? or would after 2pm ET be better?12:33
<osmandin>now is better
<awoods>standard hangout?
<pivotal-bot____>Andrew Woods edited "Code cleanup" https://www.pivotaltracker.com/story/show/6557029612:34
* scossu joins12:53
<osmandin>afk 3013:00
<pivotal-bot____>Mike Durbin added "Determine authentication challenge strategy for environment with open and protected content." https://www.pivotaltracker.com/story/show/6557466613:02
Mike Durbin edited "Determine authentication challenge strategy for environment with open and protected content." https://www.pivotaltracker.com/story/show/65574666
* ermadmix leaves13:04
* ermadmix joins13:07
<pivotal-bot____>Stefano Cossu added comment: "@ericjames: Yes, I have the properties jcr:hasParent and jcr:hasChild explicitly defined in my JCR. But I ..." https://www.pivotaltracker.com/story/show/6449951413:14
* ermadmix leaves13:39
* edInCo joins13:52
* ermadmix joins13:53
<tecoripa>anyone interested in clustering: we've been eyeing Docker over here for a while...14:11
seems like it might be a good path for setting up test environments, including in a cluster14:12
anyone have any experience with it?
my worry is that it might be pretty green
<pivotal-bot____>Mike Durbin added "Batch Modify (unified search and update)" https://www.pivotaltracker.com/story/show/65580850
<tecoripa>and could introduce performance problems that wouldn't necessarily be bottlenecks in Fedora, ModeShape, etc.14:13
<pivotal-bot____>Mike Durbin added "Download to shared location" https://www.pivotaltracker.com/story/show/6558149814:19
Mike Durbin added "Test for appropriate behavior for concurrent updates" https://www.pivotaltracker.com/story/show/6558212814:25
Mike Durbin added "Ensure that use case querying is supported" https://www.pivotaltracker.com/story/show/6558288414:33
Gregory Jansen added comment: "Found the bug in PropertiesRdfContext in kernel. MS used to throw AccessControlException, which was incor..." https://www.pivotaltracker.com/story/show/6418377614:34
* ajs6f leaves14:35
* ajs6f joins14:37
<pivotal-bot____>Mike Durbin added comment: "I've thoroughly perused the use cases linked for the beta feature set and made note of those that seemed to ..." https://www.pivotaltracker.com/story/show/6548803214:39
Mike Durbin accepted "Evaluate the level of completion for beta features" https://www.pivotaltracker.com/story/show/6548803214:40
* github-ff joins14:46
[fcrepo4] gregjan opened pull request #233: fixed fcrepo for modeshape 3.7.1 (mode-3.7.1...mode-3.7.1) http://git.io/A9Y8Aw
* github-ff leaves
<pivotal-bot____>Gregory Jansen added comment: "https://github.com/futures/fcrepo4/pull/233" https://www.pivotaltracker.com/story/show/64183776
Gregory Jansen finished "Investigate cause of test failures in fcrepo-auth-roles-basic when using Modeshape 3.7.1" https://www.pivotaltracker.com/story/show/6418377614:47
Gregory Jansen started "Upgrade to ModeShape version: 3.7.1.Final" https://www.pivotaltracker.com/story/show/6341588414:59
* ajs6f1 joins15:00
* ajs6f2 joins15:01
* ajs6f leaves
<pivotal-bot____>Gregory Jansen added comment: "I have fixed the IT issue related to permissions. Now doing a quick build and verify of modeshape 3.7.1 c..." https://www.pivotaltracker.com/story/show/63415884
Gregory Jansen added comment: "As an aside, note that Infinispan has released 6.0.0.Final in November. It is probably not backward compa..." https://www.pivotaltracker.com/story/show/6341588415:02
* ajs6f1 leaves15:04
<pivotal-bot____>Mike Durbin added "fcrepo should not reference versions in the system workspace." https://www.pivotaltracker.com/story/show/6558581615:06
Gregory Jansen added comment: "Okay, I am leaving 5.3.0.Final infinispan configured as is. It tests out okay for modeshape repository AP..." https://www.pivotaltracker.com/story/show/6341588415:15
Gregory Jansen added comment: "mode-3.7.1 can be merged to master after it incorporates https://github.com/futures/fcrepo4/pull/233" https://www.pivotaltracker.com/story/show/6341588415:18
* ajs6f joins15:31
* ajs6f2 leaves
<pivotal-bot____>Mike Daines edited "Declare exportAs for root without using double slashes" https://www.pivotaltracker.com/story/show/6088375215:41
* ajs6f leaves15:44
* ajs6f joins15:46
* ajs6f leaves
<pivotal-bot____>Gregory Jansen started "Consolidate war configuration files" https://www.pivotaltracker.com/story/show/6556663416:22
Gregory Jansen added comment: "Fedora's default config files will all be under WEB-INF/classes/config: ""16:30
infinispan.xml" https://www.pivotaltracker.com/story/show/65566634
Gregory Jansen added comment: "activemq.xml as well (already there)" https://www.pivotaltracker.com/story/show/65566634
Gregory Jansen added comment: "The defaults on this page will be adjusted once the relevant PR is accepted: https://wiki.duraspace.org/d..." https://www.pivotaltracker.com/story/show/6556663416:31
Eric James added comment: "Updated documentation: ""16:34
Added test to verify an exter..." https://www.pivotaltracker.com/story/show/64499514
* github-ff joins16:37
[fcrepo4] awoods closed pull request #232: Code cleanup (master...DiamondOp) http://git.io/h7zL2A
* github-ff leaves
<pivotal-bot____>Eric James finished "SPARQL search does not find results for fcrepo:hasParent and fcrepo:hasChild" https://www.pivotaltracker.com/story/show/64499514
Scott Prater started "Create deployment tool for clusters" https://www.pivotaltracker.com/story/show/65565926
Andrew Woods delivered "Code cleanup" https://www.pivotaltracker.com/story/show/65570296
Scott Prater added comment: "I'm going to use Puppet to deploy Fedora to a cluster. The deliverables will consist of a wiki page, expla..." https://www.pivotaltracker.com/story/show/6556592616:39
Scott Prater estimated "Create deployment tool for clusters" as 5 points https://www.pivotaltracker.com/story/show/65565926
Andrew Woods edited "Download to shared location" https://www.pivotaltracker.com/story/show/6558149816:40
Andrew Woods edited "Test for appropriate behavior for concurrent updates" https://www.pivotaltracker.com/story/show/6558212816:41
Andrew Woods edited "Test for appropriate behavior for concurrent updates" https://www.pivotaltracker.com/story/show/65582128
* github-ff joins16:43
[fcrepo4] awoods closed pull request #233: fixed fcrepo for modeshape 3.7.1 (mode-3.7.1...mode-3.7.1) http://git.io/A9Y8Aw
* github-ff leaves
<pivotal-bot____>Andrew Woods added comment: "@gregoryjansen, can you please submit a PR with this update against "master". Your current PR is against a ..." https://www.pivotaltracker.com/story/show/6418377616:45
Andrew Woods rejected "Investigate cause of test failures in fcrepo-auth-roles-basic when using Modeshape 3.7.1" https://www.pivotaltracker.com/story/show/64183776
Eric James started "Inject triples for supertypes into RDF" https://www.pivotaltracker.com/story/show/6521027216:47
* github-ff joins
[fcrepo4] gregjan opened pull request #234: consolidates default configuration files for modeshape in the WAR under ... (master...master) http://git.io/T9OsxQ
* github-ff leaves
<pivotal-bot____>Gregory Jansen added comment: "https://github.com/futures/fcrepo4/pull/234" https://www.pivotaltracker.com/story/show/65566634
* travis-ci joins16:52
[travis-ci] futures/fcrepo4#1528 (master - 5d999eb : Andrew Woods): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/c1bb4d17788c...5d999eb27ce2
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/18684671
* travis-ci leaves
<pivotal-bot____>Andrew Woods added comment: "@gregoryjansen, is PR-234 ready to go? If so, please "FINISH" this ticket (or more likely, create a new tic..." https://www.pivotaltracker.com/story/show/6556663416:53
<awoods>mikeAtUVa: are you around much longer today? do you have time to chat?16:54
<mikeAtUVa>awoods: yeah, a few more hours.16:55
<awoods>mikeAtUVa: do you have time for a google-hangout sometime within the next few hours?16:56
<mikeAtUVa>awoods: sure. let me know when you're free.16:57
awoods: now's fine, later's probably also fine.
<pivotal-bot____>Andrew Woods finished "Investigate cause of test failures in fcrepo-auth-roles-basic when using Modeshape 3.7.1" https://www.pivotaltracker.com/story/show/6418377616:58
<awoods>gregjansen: nevermind about my comment on https://www.pivotaltracker.com/story/show/64183776
<pivotal-bot____>bug: Investigate cause of test failures in fcrepo-auth-roles-basic when using Modeshape 3.7.1 (finished) / owner: Gregory Jansen
<awoods>gregjansen: I will do the merging, etc.
mikeAtUVa: ready now? I am heading to the hangout.16:59
* daines leaves17:00
* gregjansen leaves17:14
* ermadmix leaves17:56
<pivotal-bot____>Mike Durbin edited "fcrepo should not reference versions in the system workspace." https://www.pivotaltracker.com/story/show/6558581617:59
Mike Durbin started "fcrepo should not reference versions in the system workspace." https://www.pivotaltracker.com/story/show/6558581618:12
Andrew Woods added "Copy between Federation and Repo" https://www.pivotaltracker.com/story/show/6560015218:23
* ksclarke leaves18:28
* edInCo leaves18:48
* tecoripa leaves19:02
* mikeAtUVa leaves19:03
<pivotal-bot____>Andrew Woods added comment: "Resolved with: ""19:06
" https://www.pivotaltracker.com/story/show/63415884
Andrew Woods delivered "Upgrade to ModeShape version: 3.7.1.Final" https://www.pivotaltracker.com/story/show/63415884
Andrew Woods delivered "Investigate cause of test failures in fcrepo-auth-roles-basic when using Modeshape 3.7.1" https://www.pivotaltracker.com/story/show/64183776
Mike Durbin added "Versioning of container nodes should be more efficient." https://www.pivotaltracker.com/story/show/6560232019:08
Mike Durbin edited "Versioning of container nodes should be more efficient." https://www.pivotaltracker.com/story/show/65602320
Andrew Woods accepted "Code cleanup" https://www.pivotaltracker.com/story/show/65570296
* travis-ci joins19:22
[travis-ci] futures/fcrepo4#1531 (master - e2773e4 : Gregory Jansen): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/5d999eb27ce2...e2773e4a0fd1
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/18694404
* travis-ci leaves
* scossu leaves19:37
<pivotal-bot____>Scott Prater added comment: "Done." https://www.pivotaltracker.com/story/show/5994376819:48
Scott Prater delivered "Provision UW cluster nodes" https://www.pivotaltracker.com/story/show/5994376819:49
* osmandin leaves19:58
<pivotal-bot____>Andrew Woods accepted "Investigate cause of test failures in fcrepo-auth-roles-basic when using Modeshape 3.7.1" https://www.pivotaltracker.com/story/show/6418377620:20
Andrew Woods accepted "Upgrade to ModeShape version: 3.7.1.Final" https://www.pivotaltracker.com/story/show/63415884
Andrew Woods added comment: "Resolved with: https://github.com/futures/fcrepo-jms-indexer-pluggable/commit/575457aa3acd6bf3d5eefedcb0e13..." https://www.pivotaltracker.com/story/show/6449951421:03
Andrew Woods delivered "SPARQL search does not find results for fcrepo:hasParent and fcrepo:hasChild" https://www.pivotaltracker.com/story/show/64499514
* ksclarke joins21:04
* travis-ci joins21:14
[travis-ci] futures/fcrepo-jms-indexer-pluggable#83 (master - 575457a : Andrew Woods): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo-jms-indexer-pluggable/compare/3fe042586cf1...575457aa3acd
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo-jms-indexer-pluggable/builds/18699849
* travis-ci leaves