Log of the #fcrepo channel on chat.freenode.net

Using timezone: Eastern Standard Time
<coblej>westgard: I just pulled down the latest changes to the verify tool ... I'm not familiar with python so I haven't been able to find out how to run the latest version15:48
I was just updating the rEADME15:49
<westgard>dbernstein has refactored the entire tool so it's a little different now
<coblej>I can wait till the README is redone
<westgard>try "python3 setup.py install'
<westgard>then you can run with fcrepo-verify from anywhere on your system15:50
that is, with the command "fcrepo-verify"
<escowles>awoods: i'll take a look at that error to see if i can reproduce it
<westgard>setup.py installs it on your PATH
fcrepo-verify --help gives you the options
* github-ff joins
[fcrepo4] awoods pushed 1 new commit to master: https://git.io/vHfAe
fcrepo4/master a07a1a4 Michael Durbin: Fcrepo 2432 (#1193)...
* github-ff leaves
<awoods>westgard: should the following ticket be "In Review"? https://jira.duraspace.org/browse/FCREPO-246016:01
<westgard>awoods: I thought it was ready, but now I'm updating the README16:03
Also I see it failed travis checks
So I'm going to push another commit, but then yes it would be ready for review
<awoods>westgard: no worries, I just noted that you linked a PR (great) but it was still "In Progress" (confusing)
<westgard>awoods apropos travis, do you know how I would get a reason why flake8 failed?16:05
<awoods>westgard: https://travis-ci.org/fcrepo4-labs/fcrepo-import-export-verify/builds/234139990#L21616:06
westgard: does that help?16:08
<westgard>Ahh, thanks -- right in front of me
<awoods>westgard: I appreciate that extra '+'16:11
escowles: fyi, the error is also reproducible with the file-simple configuration16:25
<escowles>awoods: but only with tomcat?16:26
i just tried using the 1-click jar and wasn't able to reproduce the problem yet
just setting up tomcat now
<awoods>escowles: not sure. I have not tried another container. Let me give the one-click a spin
<escowles>awoods: the 1-click is complicated by not being able to set a repository.json (unless you unpack and edit the repo.xml file, which is what i did)16:27
<awoods>escowles: yes, but the default configuration in the one-click should be sufficient
escowles: the default configuration is file-simple
escowles: I also see the error with the one-click. I will document my steps in the testing wiki page.16:32
<escowles>awoods: that is good — i just got done following your steps with tomcat, and i am not seeing any errors...16:33
<coblej>awoods: escowles: I tried to replicate the error using the one-click 4.7.0 and 4.7.2 but didn't see a BOOM so maybe I missed or didn't get it set up exactly right16:34
<awoods>escowles/coblej: are you watching the log?
<escowles>awoods: i am following catalina.out
i wonder if macosx vs. linux is a factor here16:35
<coblej>awoods: I ran the one-click interactively from the command line and was watching that output16:36
I did see this with 4.7.2 ... ERROR 16:26:04.445 (WebApplicationExceptionMapper) WebApplicationException intercepted by WebApplicationExceptionMapper: HTTP 304 Not Modified ... was that the BOOM?16:37
but it seemed to display okay in the Fedora web UI
<awoods>escowles/coblej: no, you should see namespace errors:
Caused by: org.fcrepo.kernel.api.exception.RepositoryRuntimeException: org.modeshape.jcr.value.ValueFormatException: Error converting "ns001:TestSKOS" from String to a Name
escowles/coblej: and Fedora will be unable to start16:38
coblej: what OS are you using?
<coblej>awoods: Mac OSX16:39
<escowles>i just restarted tomcat and now i'm seeing the error
<awoods>escowles: yes, the restart is where it happens
<coblej>awoods: escowles: just got the expected BOOM following awoods one-click directions ... I may have missed the restart part before16:44
<awoods>coblej: that is good to hear16:48
<escowles>ok, so the minimal steps to reproduce are:16:52
1. start 4.7.0: java -jar fcrepo-webapp-4.7.0-jetty-console.jar --headless
2. create object with SKOS namespace: curl -i -X POST -H "Content-type: text/turtle" -d " @prefix skos: <http://www.w3.org/2004/02/skos/core#> . <> a skos:Concept ." http://localhost:8080/rest/
3. stop 4.7.0: ^C
4. start 4.7.2: java -jar fcrepo-webapp-4.7.2-jetty-console.jar --headless16:53
5. stop 4.7.2: ^C
6. start 4.7.2 again: java -jar fcrepo-webapp-4.7.2-jetty-console.jar --headless
=> error: Error converting "ns001:TestSKOS" from String to a Name
<awoods>escowles: that makes sense
* github-ff joins16:57
[fcrepo-import-export] lsitu closed pull request #82: Added support for round-tripping with binaries excluded. (master...feature/filter_binary) https://git.io/v9h3q
* github-ff leaves
<awoods>escowles: thoughts on next steps besides running through the standard release tests and moving forward with a release?17:03
* coblej leaves17:04
* umgrosscol leaves
<escowles>awoods: posting on the mailing list that we have a RC and asking anybody who's hit this problem to try loading their data in the RC?17:05
<awoods>escowles: ok... but once you have already hit the issue,the RC does not help.17:06
<escowles>awoods: right — but if they have an uncorrupted pre-4.7.2 backup they could test, that would be ideal
<awoods>escowles: agreed
<escowles>i think the CHF might be able to do that
<awoods>escowles: I also wonder if Steven A. has a way of getting out of the BOOM state17:07
<escowles>maybe editing the root node directly? that sounds like what he was doing
awoods: on a separate topic: i've been working on https://github.com/fcrepo4/fcrepo4/pull/1195 most of the day and getting ConcurrentModificationExceptions — i just did a clean build of master with concurrency turned on and got the same error17:08
<awoods>escowles: ..something that we could incorporate
escowles: does it work with parallel turned off?
<escowles>awoods: yes, it builds fine with parallel off
<awoods>escowles: travis has also made it clear that either our tests or the code itself around the parallelization is bad17:09
<escowles>i've been seeing three tests in FedoraLdpIT fail (testReplaceGraph:765, testRoundTripReplaceGraph:831, and testRoundTripReplaceGraphForDatastreamDescription:1952) — and all three of those just failed in my master build17:10
<awoods>escowles: We need a ticket for addressing those tests (and potentially others)
<escowles>ok, i'll make a ticket
<awoods>escowles: thanks17:11
i'm heading out — have a great weekend everybody!17:14
<awoods>escowles: thanks. you too17:17
* apb18 joins17:22
* westgard leaves17:27
<apb18>awoods: I wonder how many people are in the state of "unlucky to have installed 4.7.2, but lucky enough to have not re-started it yet"17:37
I tried to do a backup/restore of 4.7.2 while it was in the "works, but hasn't been re-started yet" state. The backup was corrupt (i.e. induced errors when restored into an empty 4.3 repo)17:38
awoods: Maybe we need to send out a warning "If you have 4.7.2, don't shut it down or re-deploy until further notice"17:39
I would guess that fcrepo-import-export could make an export as a stop-gap solution in the absence of any other backup17:40
... but if people shut off their still-working 4.7.2, upgrade to 4.7.3, and turn it on; they could be greeted with a non-working repository17:42
awoods: Just commented on wiki. The underlying problem is that Modeshape removes the ns00* namespaces from the system namespaces node. Putting them back is a potential fix18:31
I did that manually on a backup from a still-working 4.7.2 before I restarted it to see the BOOM18:32
* benpennell leaves18:35
* apb18 leaves18:48
* github-ff joins20:11
[fcrepo-import-export] lsitu reopened pull request #82: Added support for round-tripping with binaries excluded. (master...feature/filter_binary) https://git.io/v9h3q
* github-ff leaves
* benpennell joins20:43
* dbernstein joins21:08
* dbernstein leaves21:23
* benpennell leaves21:39
* mikeAtUVa leaves21:57