IQSS logo

IRC log for #dataverse, 2018-10-12

Connect via chat.dataverse.org to discuss Dataverse (dataverse.org, an open source web application for sharing, citing, analyzing, and preserving research data) with users and developers.

| Channels | #dataverse index | Today | | Search | Google Search | Plain-Text | plain, newest first | summary

All times shown according to UTC.

Time S Nick Message
02:28 jri joined #dataverse
06:20 jri joined #dataverse
07:56 poikilotherm joined #dataverse
08:14 otheus joined #dataverse
08:14 otheus I want to contribute to the Admin/installation documentation, and possibly to making some parts deployable via RPMs
09:08 poikilotherm Hi @otheus, Harvard is not awake yet... 05:08 over there ;-)
09:14 jri joined #dataverse
09:28 otheus i'm patient.
10:17 pdurbin IRC rewards patience.
10:21 poikilotherm The dinosaur grandpa of all chat systems has some patience built in, hasn't it?
10:21 pdurbin otheus: thanks for the pull request yesterday! Yes! Please help with our docs! More RPMs would be great too. We already build our own RPMs for rApache, downloadable from the TwoRavens section of the Installation Guide.
10:22 pdurbin poikilotherm: I've been running this IRC channel since before Slack existed. :)
10:27 otheus pdurbin: Are the Administration/installation docs in a git repo? I didn't see them in the sources repo.
10:28 otheus Also, there's this: https://github.com/mentalblock/solr-rpm (but its 3 yrs old)
10:32 pdurbin otheus: yes, we use Sphinx and .rst files which are in the main git repo. This is explained at http://guides.dataverse.org/en/4.9.4/developers/documentation.html but that page itself could use a refresh. :/
10:37 poikilotherm pdurbin: and maybe a maven plugin to trigger the build instead of a make call? (plus maybe some CI...)
10:39 pdurbin poikilotherm: for the make thing, it would be great for you to leave a comment at https://github.com/IQSS/dataverse/issues/4528
10:50 otheus @pdurbin I'm looking at https://github.com/IQSS/dataverse/tree/develop/doc and I see no files containing documentation (except .md files).
10:50 otheus oh wait, the documenation is under "sphinx-guides". Yeah, that wasn't expected
10:52 poikilotherm pdurbin: I saw http://irclog.iq.harvard.edu/dataverse/2018-10-11#i_75440 earlier... Actually you have different options in Spring. You can embed the container runtime and build a runnable jar, that will start the container and the application or you can leave that aside and build deployable wars
10:53 poikilotherm When you opt for runnable jars, spring boot is of great help.
10:53 poikilotherm This stuff is really easy to dockerize, as it is just a java jar
10:54 poikilotherm (thus the container basically needs java and maybe some other deps)
10:54 poikilotherm But as you mentioned, Spring is not using EE stuff. I am not aware of anything equivalent like Spring Boot for Java EE applications
10:56 poikilotherm The "fat jar" from Spring makes it also a breathe to run it on a laptop/dev env... Especially when you enable things like LiveReload, the development of frontend stuff is quite neat
10:59 poikilotherm But IMHO all this nice stuff seems not be relevant to Dataverse... Porting all of this to Spring could be an effort to big for the project and let its evolution die...
11:08 pdurbin Yeah. Making breakfast. Kids waking up soon.
11:09 poikilotherm :-)
11:14 pdurbin I have to admit that I don't have any real experience with Spring but I've heard a lot about it. Have you heard of MicroProfile?
11:15 poikilotherm Read about it, but no exp. IIRC this was targeted to microservices. Unsure if Dataverse could be expressed as a set of microservices.
11:16 pdurbin Well, maybe you'll turn the PID code into a microservice. :)
11:16 poikilotherm :-D
11:17 poikilotherm Actually I would think of a Spring Boot app for this... ;-)
11:17 poikilotherm This stuff needs no EE, so that might be more lightweight
11:21 pdurbin But people would be able to write their own PID service in whatever language, right?
11:23 donsizemore joined #dataverse
11:24 pdurbin otheus: are you unblocked? Are you thinking of a big rewrite or just small tweaks? Either way, if you would create an issue before making a pull request, that would be great.
11:24 poikilotherm Uh... I don't think so. Ok, I see your point about microservices as a possible approach for that
11:25 donsizemore @pdurbin i still need to fix something in dataverse-ansible, but i'm going to go ahead and submit a pull request for #5162
11:25 pdurbin poikilotherm: I'm confused. Someone can't come along and write a PID service in Python or whatever?
11:25 pdurbin donsizemore: please! The minute someone from the community makes a pull request, I can help. :)
11:25 poikilotherm Sure, as long as it has a REST api or whatever API that could be talked to by the "proxy service" written in Java
11:26 poikilotherm But if you want to have the service _inside_ the proxy, you would have to use java
11:26 donsizemore @pdurbin it's an asadmin command in dataverse-ansible i'm trying to club into submission. i think the ec2-create script is good once i proof-read to make sure i don't leave behind any debugging mess
11:26 poikilotherm (at least DANS people started with java now, maybe stick with that...)
11:30 otheus speaking of the asadmin command, when the install script runs it, it prompts for the admin username, even though $noninteractive is true. Is there any way to provide the java command (asadmin) with an option or variable so that this prompt isn't necessary?
11:36 pdurbin poikilotherm: I'm not sure I understand the architecture. How do you feel about creating a new GitHub issue that explains the idea? Yesterday at standup I mentioned our chat about PID services but having the plan in one document (maybe a Google doc) would be great.
11:37 pdurbin donsizemore: ok, if you need more eyes on in, please just let us know
11:39 pdurbin otheus: I would love to remove that prompt. Have you tried using a "default.config" file with the installer like this? https://github.com/IQSS/dataverse/blob/v4.9.4/scripts/vagrant/install-dataverse.sh#L29
11:41 donsizemore @otheus the install script does that at https://github.com/IQSS/dataverse/blob/develop/scripts/installer/install#L957
11:43 donsizemore @otheus (and i'm checking now to see what you can do in asenv.conf)
11:44 donsizemore @otheus not much on the latter... n/m
11:49 otheus @pdurbin yes, I'm using the default.config file, but there is no asadmin user default.
11:50 otheus @donsizemore but the asadmin login command prompts for a user name.
11:50 pdurbin otheus: you're saying you'd like a way to configure the username and password for asadmin using the default.config file? I may be confused.
11:50 otheus I'm saying if you want to do an automated install, without prompts, it should be automated without prompts.
11:50 pdurbin heh, sure
11:50 otheus OK, but more importantly I have a nother big issue.
11:51 pdurbin In Vagrant it doesn't prompt. That's what I was trying to link you to.
11:51 otheus It's a catch-22. The first install process failed because glassfish couldn't restart (timed out).
11:51 otheus The second run complained that psql database already existed, so the install process aborted
11:51 otheus The seeming work-around was to use --skip_db_setup
11:52 pdurbin Hmm, you're using the installer in the 4.9.4 release?
11:52 donsizemore @otheus i vaguely remember running into this, but the install script runs repeatedly without prompting. at a glance i don't see where in the script. leonid would be the guy to ask
11:52 otheus But THAT skips "step 3" in the install script, which incidentally, sets up the "$pg_exec_path" required by "import_reference_data" (step 7)
11:53 pdurbin Why did Glassfish fail to start? Not enough RAM or something?
11:53 otheus No reason. It was stopped and wasnt started again.
11:53 otheus I restarted it manually, no issue
11:54 pdurbin Huh.
11:54 otheus I downloaded the dvinstall zip file per the website instructions
11:54 pdurbin Ok, great.
11:55 pdurbin otheus: we just rewrote the installer so maybe a bug or two slipped in. If you would open a GitHub issue about the trouble you're having, it would be much appreciated.
11:55 otheus Sure. :)
11:55 otheus By the way, I know how much pain-in-the-ass-work it is to create an installer. Especially in unix.
11:56 otheus Going to the dentist is by far a more pleasant experience
11:56 pdurbin Yeah, that old Perl script is from 10 or 12 years ago. We just keep hacking on it.
11:56 * pdurbin added "use strict"
11:58 pdurbin donsizemore doesn't even use the Perl script. His dataverse-ansible stuff does the equivalent. That's my understanding anyway.
11:59 otheus I can't hack ansible yet, so I stay away.
11:59 donsizemore i kind of cobbled ansible around the installer. it does some stuff the install script would do, then calls the scripts it calls.
11:59 pdurbin Sure, the same underlying scripts are called, which is good.
12:00 donsizemore @pdurbin when you get time look over https://github.com/OdumInstitute/dataverse/blob/5162_cant_publish_using_EC2_script/scripts/installer/ec2-create-instance.sh ?
12:00 pdurbin But yeah, I agree, writing installers isn't much fun. We are definitely open to ideas on how to do this better. "The script is to a large degree a derivative of the old installer from DVN 3.x. It is written in Perl. If someone in the community is eager to rewrite it, perhaps in a different language, please get in touch. :)" http://guides.dataverse.org/en/4.9.4/installation/installation-main.html
12:00 donsizemore @pdurbin i spent too much time yesterday banging my head on the delete-jvm-option in https://github.com/IQSS/dataverse-ansible/blob/master/templates/configure_doi.sh.j2
12:01 * pdurbin looks
12:01 pdurbin I'm glad to see my friends -b and -r are back. :)
12:02 donsizemore @pdurbin if your branch testing will only run against mds.test.datacite.org you're good to go. but ansible is putting the baseurl variable in template then firing that as a shell script, and i didn't hit on the right combination of escape syntax to make it remove the dataverse-provided mds.test.datacite.org jvm-option
12:02 pdurbin For now I think we'll be using mds.test so that's fine.
12:02 pdurbin \
12:03 pdurbin sorry, cat on my keyboard
12:03 donsizemore (and i wanted to get home because we lost a tree in the hurricane and i wanted to make sure the house was okay)
12:03 pdurbin donsizemore: this isn't your technical debt, it's mine but any thoughts on "TODO: Add some error checking for this ssh command"? Should we run each command individually?
12:04 donsizemore @pdurbin i can take that on if you want
12:04 pdurbin for this pull request or the next one?
12:05 pdurbin I don't want to keep piling it on.
12:05 donsizemore (or we could simply set -e )
12:05 pdurbin hmm? -e?
12:06 donsizemore that may be shell only, but it's an easy test
12:06 poikilotherm donsizemore I have some exp in Puppet and try to come up with Ansible. I saw you working on the ansible module and actually was beginning to invent my own because of granularity etc. But the thing that bothers me most about the installer script and the ansible module calling it is the lack of idempotency... Any chances we maybe could work together to improve that?
12:06 donsizemore that makes a shell script exit on any error
12:06 donsizemore @poikilotherm different people want the installer (and dataverse-ansible) for different reasons
12:07 pdurbin donsizemore: let's defer until the next pull request. Ready to make one?
12:07 donsizemore @poikilotherm so far the highest demand has been for development, which requires no idempotency
12:07 donsizemore @poikilotherm the dataverse installer itself isn't idempotent, and dataverse-ansible is just a wrapper.
12:08 pdurbin donsizemore: sure but you've got interest in production installs such as Jamie from UCLA
12:08 donsizemore @poikilotherm but i'm all for idempotency and i'd love to work together on that
12:08 donsizemore @pdurbin pull requests welcome =)
12:08 pdurbin :)
12:08 pdurbin I think I made one once. :)
12:09 cdsp-rmo joined #dataverse
12:09 pdurbin Oh, all I did was add a Travis CI button. Lame.
12:10 poikilotherm Actually I think most of the stuff in the installer is easy to get inside ansible and is idempotent. The biggest blocker for me seemed to be all of the `asadmin` calls.
12:10 poikilotherm (err - would be idempotent when used properly in ansible)
12:11 pdurbin How are the asadmin calls a blocker? I'm confused.
12:11 donsizemore @poikilotherm dataverse-ansible itself doesn't do much by way of asadmin calls(?)
12:11 poikilotherm Aye. But the installer scripts could be replaced by ansible playbooks that ideally are idempotent
12:12 poikilotherm (and most certainly easier to maintain than a perl script)
12:12 pdurbin cdsp-rmo: hi! Just a quick reminder about docs. :)
12:12 donsizemore @poikilotherm then i'm tracking the dataverse installer... i intentionally tried to avoid that
12:12 donsizemore @pdurbin bash supports '-e': ./ec2-create-instance.sh: line 3: bad: command not found
12:12 pdurbin poikilotherm: that would force people like me and otheus to learn Ansible. Which may not be terrible. :)
12:12 cdsp-rmo pdurbin: hi ! Yep, doing this till I go to my meeting
12:13 poikilotherm asadmin does not check if e.g. a JVM option is already present, it will just add it again and again
12:13 otheus *throws up in his mouth*
12:13 pdurbin cdsp-rmo: thanks!
12:13 cdsp-rmo and hi everyone
12:13 pdurbin poikilotherm: checks could be added. Good idea.
12:13 pdurbin otheus: you'd rather stick with Perl? :)
12:13 poikilotherm Well, maybe its better to edit a config file if you can ;-)
12:14 donsizemore @poikilotherm correct, i ran into that with the datacite provider url
12:14 poikilotherm Basically the asadmin calls don't do other things than write into the domain config
12:14 donsizemore @poikilotherm editing domain.xml was plan A but ansible's XML module wasn't cooperating
12:15 poikilotherm pdurbin: adding checks with asadmin is pretty ugly stuff todo inside a perl script. clumbsy... better use dedicated software like puppet or ansible for such things
12:15 otheus pdurbin: heh, i love perl. But I recognize it's becoming obsolete. I'd stick with shell scripting for an install, but my recent work on MacOS/BSD taught me just how backwards the pure POSIX/BSD world is w.r.t Gnu tools.
12:15 poikilotherm Yeah, i read about that
12:15 pdurbin poikilotherm: in DVN 3.x the installer made a lot of hacks to domain.xml but it was terrible and from my perspective the asadmin approach is much cleaner. I'm happy to link you to the old hacks if you'd like. :)
12:15 donsizemore @otheus what made you throw up in your mouth?
12:15 otheus Oh, the suggestion of having to learn Ansible. :)
12:16 otheus I won't be happy with anything related to Python until 2105, when the last system running 2.7 is finally shut off
12:16 poikilotherm otheus: I favor Puppet, but a lot of people tell me that the learning curve of ansbile is much lower...
12:16 donsizemore well, i wrote dataverse-ansible because i was asked to, and am maintaining it on a volunteer basis
12:16 pdurbin I don't mind Ansible but I'm not sure I want developers with Macs coming to me asking "why is step 1 of getting a dev environment set up to install Ansible and can you help me?"
12:17 poikilotherm pdurbin: sure. but setting up a dev environment with a perl and shell script is not much better... IMHO your dev env should be maven, java, an IDE and docker.
12:17 pdurbin donsizemore: we really appreciate all of your dataverse-ansible effort! We lean on it heavily in the new ec2 script, obviously! Thank you!
12:18 otheus At least perl already exists on MacOS. *shrug*
12:18 donsizemore @otheus as i often point out, i have 20 year-old perl which still runs on modern systems
12:18 pdurbin poikilotherm: some sort of target like `mvn setup-dev-environment`?
12:18 poikilotherm donsizemore: your ansbile playbook is good! I just miss the idempotency, so I can use it together with the rest of the modules
12:19 donsizemore @poikilotherm i admire idempotency as well, and would happily work together to achieve it
12:19 otheus Yes don, and that's why -- despite its lack of readability -- perl is still a sysadmin's choice
12:20 otheus Where as ruby, python, and puppet make incompatible language choices 3 times a year it seems.
12:20 otheus *changes
12:20 donsizemore @pdurbin so, install question.
12:21 pdurbin hit me
12:21 donsizemore @pdurbin the install script creates a datacite baseurl jvm-option for mds.test.datacite.org, but users are left to add their own username/password options
12:21 pdurbin yeah
12:22 donsizemore @pdurbin should the installer do that? it's not really a convenience, as datacite doesn't (yet) allow publishing without credentials
12:22 pdurbin Maybe not. What should it do instead?
12:24 donsizemore @pdurbin config.html doesn't mention the baseurl, either. might be cleaner to pull the baseurl jvm-option from the install script and add it in the docs?
12:25 otheus There's no reason the installer can't be designed to wrap idempotent directives which are invoked by ansible/puppet/whatever
12:25 donsizemore @otheus on my part, that reason is available time =)
12:25 otheus oh right. :)
12:25 otheus there's no reason with inifinite resources...
12:26 otheus ANYWAY, I'm retrying the install step and omg, this is not going well.
12:26 otheus I dropped the database to get around the existing-database problem.
12:26 otheus Once again, the install script stopped glassfish and failed to restart it.
12:27 poikilotherm Oh there is a reason why the Gitlab guys bake rpm/debs and let Chef do the heavy lifting inside...
12:27 otheus More importantly, "remote failure: Error occurred during deployment: Application with name dataverse is already registered. Either specify that redeployment must be forced, or redeply the application". Can someone please tlel me how to do that?
12:29 pdurbin donsizemore: if you're saying we don't allow the doi baseurl to be configured by the Perl script, we should. If you're saying something else, I'm sorry that I'm confused. :) Also, yes docs. Always docs.
12:30 pdurbin otheus: you may or may not get something out of https://github.com/IQSS/dataverse/blob/v4.9.4/scripts/deploy/phoenix.dataverse.org/rebuild for rebuilding
12:30 pdurbin otheus: try `./asadmin deploy --force path/to/dataverse.war`
12:31 donsizemore @pdurbin the test url is hard-coded at https://github.com/OdumInstitute/dataverse/blob/5162_cant_publish_using_EC2_script/scripts/installer/glassfish-setup.sh#L69 — would be great to make all three configurable
12:31 pdurbin If it helps, I run stuff like this a lot: git pull && mvn clean && scripts/installer/custom-build-number && mvn package && /usr/local/glassfish4/glassfish/bin/asadmin deploy --force target/dataverse-4.9.4.war
12:32 pdurbin donsizemore: ah, yes, that shouldn't be hard coded
12:33 donsizemore @pdurbin want me to open an issue to make all three config_variables like the others?
12:34 pdurbin donsizemore: please!
12:34 otheus Thanks people. I added '--force' to my copy of the install script.
12:34 pdurbin otheus: awesome
12:34 otheus I believe the error traces back to >>>
12:34 otheus remote failure: Application dataverse-4.0 is not deployed on this target [server]
12:34 otheus Command undeploy failed.
12:35 otheus Ugh, sigh.
12:35 otheus remote failure: Error occurred during deployment: Exception while loading the app : Error in linking security policy for dataverse -- Inconsistent Module State. Please see server.log for more details.
12:38 otheus @pdurbin the problem with restarting glassfish seems to be related to a conflict with systemd and asadmin
12:38 otheus so if systemd started glassfish, asadmin can stop it and start it, but it doesnt behave well enough for a proper restart
12:38 donsizemore @otheus for inconsistent module state, stop glassfish, rm -rf $GLASSFISH_HOME/glassfish/​domains/domain1/generated and start
12:40 otheus donsizemore: thanks. on the next run I got "EJB Timer Service is not available. Please see server.log for more details."
12:40 donsizemore @otheus that typically means glassfish didn't clean its entries from the EJB postgres table when it shutdown and/or crashed
12:42 otheus ok, so "rm -rf generated" and dropping database *should* fix that?
12:43 poikilotherm pdurbin: watching a talk by thilo frotscher about using the microservices approach similar to spring boot for Java EE
12:43 poikilotherm https://jaxenter.de/java-ee-microservices-frotscher-60969 (sorry, it's in german)
12:44 otheus oje scheiße!
12:44 poikilotherm I think you dipped your nose in the right bucket - this could definitly be interesting...
12:46 otheus @donsizemore and @pdurbin success! (?)
12:46 otheus installer then tries "Executing curl -X PUT -d ADMIN_EMAIL http://localhost:8080/api/admin/settings/:SystemEmail
12:46 otheus " which outputs an HTML page with an error "ERR_CANNOT_FORWARD"
12:48 otheus <li id="permission-denied">The administrator may not allow this cache to make direct connections to origin servers.</li>
12:49 otheus But I do see "Your Dataverse has been configured to use DataCite, to register DOI global identifiers in the
12:49 otheus test name space "10.5072" with the "shoulder" "FK2"
12:49 otheus "
12:49 otheus in the same output.
12:55 pdurbin otheus: that's a new one for me... that cache stuff
12:56 otheus *snap* No dice. I visit the service on port 8080 (from another host) and I get 500 error, with tons of stack traces in server.log.
12:56 pdurbin poikilotherm: isn't DropWizard the Java EE equivalent of Spring Boot? Not sure.
12:56 otheus A lot of complaints about EJB service bean.
12:56 poikilotherm pdurbin: better look at WildFly Swarm and Payara Micro, etc
12:57 pdurbin otheus: so have you had a successful install of Dataverse at all? :/
12:57 poikilotherm pdurbin: this leads to a java -jar ability
12:57 otheus I get no login, so... no. :(
12:57 pdurbin bah
12:58 pdurbin otheus: you *could* try `vagrant up` and see if that works. It should.
12:58 otheus Part of my task is to make this process reproduicable for other universities.
12:58 otheus So vagrant solution only gets me to say 'it works'
12:59 pdurbin otheus: ah, well we'd love to help you with that
12:59 otheus but my employer wants something more significant than that
12:59 otheus And I'm working at a uni, and they see it as their mission to contribute back to this thing, so I will do that
12:59 pdurbin great!
12:59 otheus Provided it's selected among the 3 alternatives
12:59 * pdurbin crosses fingers
13:00 pdurbin Please save us from ourselves. :)
13:00 otheus lol
13:00 otheus direct chat isn't working. dont know why.
13:00 otheus I'm not using a proxy to connect and maybe our firewall is blocking it
13:00 pdurbin Lots of spam on freenode lately. They added +R on everyone's account.
13:01 otheus https://gist.github.com/otheus/c31e67e13f8f47954623afd2a001684f
13:02 otheus ^ that's the first stack trace I see. Does it tell you anything?
13:02 pdurbin otheus: you might be suffering from https://github.com/IQSS/dataverse/issues/3148
13:03 donsizemore joined #dataverse
13:03 otheus I think you hit it on the nose
13:03 pdurbin otheus: you could give our new EC2 script a try if you want: http://guides.dataverse.org/en/4.9.4/developers/deployment.html
13:05 pdurbin The nice thing about Vagrant is that you get a clean environment after a `vagrant destroy`. Maybe you could tell us some more about the server you're using. CentOS 7?
13:08 pdurbin The guy who started the EC2 script said a "small" is too small. Weird errors. Had to bump up to a "medium".
13:12 otheus yes CentOS7
13:13 otheus wait, so does this issue 3148 actually have a general resolution?
13:19 jri joined #dataverse
13:22 otheus @pdurbin: re issue 3148 refers to https://groups.google.com/forum/#!msg/dataverse-community/Esuj7HxZiF4/s3MUjtnzBgAJ . I'm getting the exact (AFAIK) same error
13:26 poikilotherm pdurbin: as we really want to dockerize our setup, this microprofile stuff is very interesting for us... using this for dev could be interesting too, as the installation of glassfish could be skipped. maybe I can prioritize this before working on the external pid and draft doi stuff to get this going for us (test deployments etc)
13:26 poikilotherm what do you guys think is more important?
13:28 poikilotherm shall I open an issue first and maybe discuss there?
13:29 poikilotherm I think this is related to the "update to glassfish 5" stuff, too
13:32 poikilotherm Thorntail (successor to Wildfly Swarm) sound pretty promising
13:32 donsizemore joined #dataverse
13:39 poikilotherm Payara Micro seems to be a good alternative approach more close of course to the existing Glassfish usage. Although the maven plugins of Thorntail seem to make life easier...
13:40 poikilotherm And it has some neat other stuff onboard, like clustering etc
13:42 cdsp-rmo joined #dataverse
13:43 otheus My $0.02 I'm not sure dockerization is a good fit for this software based on deployment use-case. But, I am generally critical of puppet deployments of software for a number of reasons, and find them suitable for environments where hard resources are outsourced. Again, my $0.02.
13:44 poikilotherm otheus: using a fatjar/uberjar has nothing to do with puppet/ansible/... It is entirely possible to instead just use another maven profile to receive the oldschool WAR file.
13:45 poikilotherm For everyone going to docker, kubernetes, automated testing, etc. fatjars make life much easier
13:46 otheus sorry, it's "Unadulterated opinion Friday"
13:47 poikilotherm *G*
13:47 pdurbin poikilotherm: how do you feel about the time and bandwidth spent pushing fat jars to Docker Hub?
13:48 poikilotherm Actually you can provide hollow jars
13:48 otheus @pdurbin I think I see something which may be the source of my problem, and why setup-all.sh fails to setup the root dataverse...
13:48 poikilotherm By that deps and the surrounding server jars are in one layer and on top of that is the application war layer
13:49 otheus The glassfish sever complains about not being able to handle requests to 'localhost.uibk.ac.at:8080'
13:49 poikilotherm Thus you only send the small changed application layer to the hub
13:49 poikilotherm At least this is possible with Thorntail, haven't looked that up for Payara Micro
13:49 pdurbin otheus: that's... weird... I think
13:49 otheus which has a 'permission denied' << >The administrator may not allow this cache to make direct connections to origin servers.< >>
13:50 otheus And it's trying to go through our proxy server to get to the cache server. i think.
13:51 poikilotherm Ah, the same seems to apply for Payara Micro, you don't need to generate a Uber jar with it.
13:51 poikilotherm Sry guys, read you all on monday. Gotta go, pick up kids
13:51 pdurbin poikilotherm: bye!
13:51 poikilotherm Will open an issue - maybe today or on Monday
13:52 pdurbin otheus: the only proxy issue I remember is this one: https://github.com/IQSS/dataverse/issues/4707
13:52 otheus i think i have the opposite problem.
13:53 pdurbin otheus: so the creation of the root dataverse is the first thing that fails? I thought the installer hit some Dataverse APIs before that.
13:53 otheus well, AFAIK, setup-all fails on its first url call
13:54 pdurbin ah, ok, that makes more sense
13:55 pdurbin maybe the "Setup the metadata blocks" step
13:55 pdurbin otheus: you can see how the output of "setup all" is supposed to look at https://build.hmdc.harvard.edu:8443/job/phoenix.dataverse.org-deploy-develop/279/console
13:56 otheus right, it fails on the first step
13:56 pdurbin ok, and it's something with your proxy maybe?
13:57 otheus it does "deleting all data from Solr" and a URL
13:57 otheus + curl 'http://localhost:8983/solr/collection1/update/json?commit=true' -H 'Content-type: application/json' -X POST -d '{"delete": { "query":"*:*"}}'
13:57 pdurbin Does deleting data from Solr work?
13:57 otheus And then I get HTML output
13:58 otheus i think from my proxy server
13:58 otheus omg.
13:58 pdurbin Huh. When I delete data from Solr I see this: {   "responseHeader":{     "status":0,     "QTime":1}}
13:59 pdurbin (on multiple lines)
13:59 otheus no_proxy is not set
13:59 otheus *shaking my head*
14:00 otheus I'm the guy who wrote the deployment script for setting up the proxy env variables. So damn, it's my fault
14:00 pdurbin What's the proxy for anyway?
14:03 otheus running inside a non-routable ip address
14:04 otheus NOW IT WORKS
14:05 otheus *fist pump* *hangs head in shame* *kicks self in disgust* *time for a beer*
14:05 pdurbin heh, good job
14:06 pdurbin otheus: oh, before you grab your beer, since you're comparing Dataverse to two other platforms, you may be interested in our blog post on this. One sec.
14:06 pdurbin https://dataverse.org/blog/comparative-review-various-data-repositories
14:06 pdurbin It's a bit oriented toward *services* rather than *software* but I hope it's helpful.
14:18 cdsp-rmo pdurbin: I found how to fix my french logging problem, so I can work without changing the dataverse log system
14:18 pdurbin cdsp-rmo: great! What should we do with that issue?
14:19 cdsp-rmo the fix is: do a full reinstall of the dataverse (undeploy, stop, delete generated, start, deploy, stop, start, stop solr, download schema.xml, start solr, curl -X DELETE http://localhost:8080/api/admin/index/timestamps, curl http://localhost:8080/api/admin/index/continue)
14:19 cdsp-rmo so, it's not a REAL fix
14:19 cdsp-rmo and, if I ever stop and start glassfish
14:19 cdsp-rmo the problem reappears
14:20 cdsp-rmo I don't understand what's happening
14:20 cdsp-rmo if I don't do all the steps of the reinstall
14:20 cdsp-rmo the problem keeps going
14:21 cdsp-rmo there is really something strange here ...
14:21 pdurbin ok, I agree... not a real fix... bummer
14:22 pdurbin I wonder if anyone else is suffering from this.
14:23 otheus @pdurbin I'm tasked also with evaluating Dspace and Invenio
14:24 pdurbin Huh, I don't think I've heard of Invenio.
14:25 pdurbin Ah. Yes I have. A friend at CERN uses it.
14:26 pameyer joined #dataverse
14:28 cdsp-rmo pdurbin: I add the full procedure to "fix" the problem to the ticket ? In case someone has the mysterious problem too, and waiting for the real fix
14:29 otheus FYI: https://github.com/IQSS/dataverse/issues/3148#issuecomment-429344986
14:30 pdurbin cdsp-rmo: yes please!
14:31 pdurbin otheus: ah, thanks. How do you feel about opening an issue to document this or tweak the shell scripts?
14:35 pameyer otheus: last I checked, non-interactive installation with default.config was non-interactive (aka - no `asadmin login` required. but it's been a week or 2 since I checked; can't rule out the possibility that it's broken since then
14:36 otheus Should I put the issues on the https://github.com/IQSS/dataverse/issues project or elsewhere?
14:36 pdurbin pameyer: 4.9.4 has only been out 9 days. :) That's what otheus is using.
14:36 pameyer and for what it's worth, my experience with the dcm lead me to think that using ansible as a user-facing (vs developer facing) installation process is a bad idea
14:37 pdurbin otheus: yes, please create the issue there. Thanks!
14:37 pdurbin otheus: oh, pameyer has been building RPMs too
14:37 pameyer pdurbin: people aren't using develop? ;)
14:37 pdurbin A tagged release should be safer. Should be. :)
14:39 pameyer yeah - as much as I'd like to see dataverse and it's dependencies RPM-ized; I think there's a big chunk of tech debt that would need to be paid down before that becomes feasible
14:41 pameyer ... and with more lagged unsolicited opinions about stuff from irc logs: docker's been a big help for development, but I'm not sold on it for production dataverse yet
14:41 pameyer ... and fortran > perl ;)
14:42 pdurbin no installers in fortran. sorry
14:43 pameyer good point.  they're good at different things (although I was impressed to find what was effectively a scripting language with the interpreter written in fortran)
14:47 pameyer otheus: non-interactive installation just worked for me with no prompts (on c95e53d805579b1f6718c7b8f49b6d4126db52ba; so I'd assume 4.9.4 should be ok)
14:48 otheus Is there some special variable used for the admin account?
14:49 pameyer not in conf/docker-aio/install.bash
14:49 pdurbin I don't really understand the Glassfish admin account stuff, to be honest. I run Glassfish on my Mac and as far as I know I don't need a password to start and stop Glassfish. Maybe the password is put in a file somewhere by the installer? I don't know.
14:50 pameyer pdurbin: if you want glassfish "secure remote admin", you need to do some asadmin login magic to enable a password
14:50 pdurbin Yeah, do not want. Not on my laptop anyway.
14:51 pameyer I'm not sure when I'd want it - if there are advantages over connecting to the server and using asadmin commands, I'm not sure what they are
14:51 otheus ./install runs "asadmin login" and THAT requests a password (but empty is accepted)
14:51 pdurbin yeah, I'm not sure why it runs "asadmin login"
14:51 otheus or?
14:53 otheus or it happens at some point in glassfish-setup.sh, but again, via adadmin
14:57 otheus FYI: other issue: https://github.com/IQSS/dataverse/issues/5175
15:05 pdurbin otheus: thanks! I don't suppose that https://github.com/IQSS/dataverse/pull/5167 magically fixes this
15:06 otheus probably not since I was the one who made those changes :P
15:06 pdurbin bummer :)
15:06 otheus But the fact that it never worked in the first palce indicates that this install process was never well-tested with it
15:06 otheus It may be necessary to simply say in installation docs that the systemd service must be stopped before running install
15:07 pdurbin otheus: I think the problem is that our test servers are still on CentOS 6.
15:07 pameyer I'm still using init for glassfish on cent7.  may be suboptimal, but has been working for me so far
15:08 otheus Anyway, 2 other issues created which we already discussed. I would love to work on improving the install process once my other fires are put out. Feel free to contact me: otheus.shelling@uibk.ac.at or otheus@gmail.com . Good luck!
15:08 * pdurbin puts out a couple fires
15:14 pdurbin cdsp-rmo: thanks for leaving a comment on that login issue
15:15 cdsp-rmo yup
15:15 cdsp-rmo I answer to the multiple depositors issue aswell
15:16 pdurbin Ah, great. Yes. Very helpful. Thanks.
15:17 pdurbin cdsp-rmo: time for your weekend, right?
15:18 cdsp-rmo I still has some time to finish my doc and push it
15:18 cdsp-rmo ;)
15:18 pdurbin nice!
15:18 pdurbin donsizemore: "Switched to a new branch '5162_cant_publish_using_EC2_script'" :) Thanks!
15:19 donsizemore @pdurbin i'll tackle the installer changes once i knock a couple of other things out of the way
15:19 donsizemore @pameyer thanks for testing that. i was going to grab the 4.9 or so release and test
15:20 pameyer @donsizemore it's usually a quick and easy thing for me to test
15:20 pameyer hasn't yet been worth the trade off of 100% automating it
15:21 cdsp-rmo pdurbin : about merging the json import and the xml import as one api, a fast answer may be useful, till I still has an idea of what I've done :D
15:21 cdsp-rmo *ave
15:21 cdsp-rmo **have
15:24 pdurbin cdsp-rmo: unfortunately, I don't think you'll get an answer until next week. Sorry.
15:24 cdsp-rmo okay
15:47 isullivan joined #dataverse
15:54 cdsp-rmo doc updated
15:54 cdsp-rmo now I disappear
15:54 cdsp-rmo have a nice weekend everyone ;)
15:56 pdurbin Thanks! You too!
15:59 pdurbin donsizemore: this worked just fine: ec2-create-instance.sh -b 5115-move-dataset
15:59 pdurbin gonna try -r next
16:01 pdurbin donsizemore: how do you feel about me tweaking the dev guide to mention the new "env" file?
16:30 pdurbin donsizemore: -r seems to work fine too. I think I'm gonna tweak the docs.
17:11 CaseyLF joined #dataverse
17:12 CaseyLF left #dataverse
17:18 donsizemore @pdurbin you do boss! (sry i forgot to document it)
17:19 donsizemore ^da
17:31 pameyer joined #dataverse
17:48 pdurbin donsizemore: no worries! I hope I did ok.
17:49 isullivan joined #dataverse
17:56 isullivan joined #dataverse
18:13 andrewSC joined #dataverse
18:50 donsizemore @pdurbin ok? you're great to work with.
18:51 pdurbin Ah, well I meant what I wrote in that doc but thanks. :)
18:51 pdurbin Did you ever figure out if there's another mailing list? I'm not sure what Jon meant.
18:56 donsizemore i think there may have been an e-mail chain in addition to the github issues
18:59 pdurbin Ah. Ok. Plenty of into in the GitHub issue. Looking forward to hearing about CoRe2.
20:11 donsizemore joined #dataverse
20:12 donsizemore joined #dataverse
20:13 donsizemore @pdurbin my turn to work from a bar! (they have blood orange bold rock on tap and i'll be working on #5173 until my friends show up)
20:15 pdurbin way to be
20:33 pdurbin have a good weekend, all!
20:33 pdurbin left #dataverse
21:47 pameyer left #dataverse
23:52 isullivan joined #dataverse

| Channels | #dataverse index | Today | | Search | Google Search | Plain-Text | plain, newest first | summary

Connect via chat.dataverse.org to discuss Dataverse (dataverse.org, an open source web application for sharing, citing, analyzing, and preserving research data) with users and developers.