IQSS logo

IRC log for #dataverse, 2018-05-25

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
02:51 jri joined #dataverse
07:10 jri joined #dataverse
07:46 jri joined #dataverse
11:04 danmcp joined #dataverse
13:52 pameyer joined #dataverse
14:25 juancorr joined #dataverse
14:27 juancorr left #dataverse
17:09 pdurbin pameyer: do you have an account at https://dataverse.harvard.edu ?
17:52 pameyer pdurbin: I don't think so
17:53 pdurbin ok, nevermind then. I'm just thinking about the OSF user experience: https://github.com/IQSS/dataverse/issues/2122
17:54 pameyer I stopped looking into OSF when I found ~ don't try to install this, use our infrastructure - but not too much storage
17:54 pdurbin Huh. I think one can install it. But I don't know who else has.
17:55 pameyer maybe I'm getting things mixed up
17:56 pdurbin I thought disk was cheap. ;)
17:58 pameyer took another quick look - lots of "sign-up", no sign of download or install
18:00 pameyer and one of my irrational technical prejudices is sites that show "you need to enable javascript" and nothing else with a sane browser profile
18:00 pdurbin heh
18:02 pameyer of course, the unbiased observer might call my normal browser profile "paranoid" rather than "sane" ;)
18:02 dataverse-user joined #dataverse
18:04 pdurbin :)
18:26 v joined #dataverse
18:27 vsoch joined #dataverse
18:27 v-dinosaur joined #dataverse
18:31 v-dinosaur hey Pete, Dan, Slava, Philip, this is Vanessa (@vsoch) from email conversations... let's chat about how I can help!
18:33 pdurbin v-dinosaur: hi! Thanks for jumping in! pameyer was active a few minutes ago but I don't know if danmcp is around or not. Some other folks in here might be interested in Docker stuff as well.
18:33 pameyer hi v-dinosaur: welcome to dataverse irc
18:33 v-dinosaur :D
18:33 pameyer looks like you're jumping into an area lots of folks have been dipping their toes into (at least)
18:33 v-dinosaur If you guys chat more in real life we can coordinate a virtual meeting time for later today, if that might be better?
18:34 v-dinosaur yeah! It may be that I would just complicate things by starting fresh (and being the 3rd person to work on something)
18:34 pdurbin v-dinosaur: I have no idea if the timeline I posted at https://github.com/IQSS/dataverse/issues/4665#issuecomment-391926503 was helpful or not but pameyer is the one who created the "docker all in one" code.
18:34 pameyer I'm not even sure if all the folks that would be interested live overlapping timezones or not
18:35 pdurbin v-dinosaur: we have a community call every two weeks. It's a bit early for you but we can certainly dedicate as many calls to Docker stuff as we need. Please see https://dataverse.org/community-calls
18:35 pameyer I don't think it would complicate things :)
18:35 v-dinosaur Dockerization is easier than it probably seems, most of the core databases / search already have pretty good solutions and it's just about adding to the docker-compose. Then it's also not too bad to move to kubernetes via kompose, which it looks like is already being done too.
18:36 pdurbin v-dinosaur: even... Glassfish applications? Easy? :)
18:36 pameyer it sounded like slava had a solution to the entrypoint vs installer question - but I'm not sure if I understood it or not
18:36 v-dinosaur It sounds like user @4tikhonov already has done this work?
18:36 v-dinosaur I have no idea what Glassfish is, heh
18:36 pameyer 4tikhonov ( aka slava) 's stuff is definately worth a look
18:36 pameyer glassfish = java application server for dataverse
18:37 v-dinosaur his solution is pinging the server to get a content length, and if the size is not zero this means it's installed
18:37 v-dinosaur this may not be the best approach because if the server is installed (but just not running) you still wouldn't want to try re-install
18:38 v-dinosaur ahh okay that's Slava!
18:38 pameyer I'd think you'd want to have the install run during image creation, rather than on container start
18:38 pameyer of course, I couldn't figure out a good way to do that in a single image; nevermind with the application server in one image and the database in another
18:38 pdurbin v-dinosaur: I think danmcp was under the impression that it would be easy too. But there's a lot of complexity to deal with, unfortunately. I think we can get there. I'm happy that you're interested in helping!
18:39 danmcp v-dinosaur: we started with openshift rather than docker compose.  I don't think we really want to make a transition from 1 to the other
18:39 danmcp v-dinosaur: but adding docker compose as an option is fine
18:39 pdurbin Right. We want both. :)
18:40 danmcp v-dinosaur: the work isn't in the config though, it's what's in the images that's the hard part
18:40 v-dinosaur ah okay, I think I understand
18:40 danmcp at least for a production level service
18:41 v-dinosaur in that case I should probably just do what I was originally planning - start with the install docs (native), and approach it building from the start up to get an understanding of the components, and that should help to design the various containers
18:42 pameyer agreed - images are the tricky part
18:42 v-dinosaur yeah I've been running Singularity Hub / Registry and some other Google Cloud services just via docker compose, and it works fairly okay given a model with builders that are deployed instances, but I'm wanting now the ability to have more automated scaling (beyond simple load balancing) and everyone and their grandmother is going with kubernetes, so unlike when it was new (and it wasn't certain if it would be a good investment) my sentim
18:44 v-dinosaur The risk always with building your infrastructure around some niche technology is that if one cloud vendor decides to spike prices for it, you are doomed. But now every cloud vendor has it, plus open source means local deployment is an option, so the risk is mitigrated.
18:44 pameyer might be worth keeping in my that not everyone is going kubernetes
18:44 pameyer s/in my/in mind/g
18:44 pdurbin v-dinosaur: that sounds fine. Some fresh eyes on all of this would be great. I do have a specific task in mind about Solr. Right now we build our own image and run it out of /tmp which is a horror I introduced in a recent Solr upgrade ( https://github.com/IQSS/dataverse/blob/ce05eb1618cd3c39a2a217eecd6b133ce6c8293e/conf/docker/solr/entrypoint.sh ). I'd like us to try switching to
18:44 v-dinosaur for some things I think it's probably overkill
18:44 pdurbin https://github.com/dudash/openshift-docker-solr or some other image that doesn't need to run as root.
18:45 v-dinosaur the thing with minikube / kubernetes is that it becomes much harder to develop, for example it's non trivial to just spin up a little cluster. Whenever I use minikube my computer crashes fabulously, heh
18:45 v-dinosaur docker-compose I can have multiple things running at once, and just stop the nginx / web server containers that are trying to grab port 80 heh
18:46 v-dinosaur haha, they are *always* running as root, it's Docker! :P
18:46 jri joined #dataverse
18:47 pameyer v-dinosaur: always running as root is part of why singulary exists ;)
18:47 v-dinosaur yes, this I know well :)
18:48 pameyer haven't played with minikube; but minishift needed me to mess with dnsmasq configs before I could connect to any containers
18:48 pameyer but I lean towards starting simple too
18:48 pdurbin v-dinosaur: we don't run our containers as root in OpenShift. Please see http://guides.dataverse.org/en/4.8.6/developers/containers.html#running-containers-to-run-as-root-in-minishift
18:49 pdurbin huh. "Running Containers to Run as Root in Minishift" is a typo. jeez
18:49 pdurbin Anyway, there I try to explain.
18:49 v-dinosaur I'll read this over carefully this weekend
18:49 v-dinosaur So next week / when the RedHat intern comes, or if that is much later we can do before it, let's have another quick chat and I'll tell you how I was destroyed by a GlassFish, or minimally have some questions about how things work.
18:50 v-dinosaur what time zone are you guys in?
18:50 v-dinosaur time zones(s) ?
18:50 v-dinosaur planetary system(s) ... :P
18:50 pameyer EDT at the moment
18:51 pdurbin v-dinosaur: 8:30 until 4:30 EST. Speaking of Glassfish, it would be nice to use an existing image rather than creating our own, if possible: https://blogs.oracle.com/theaquarium/glassfish-images-on-docker-hub
18:51 aculich joined #dataverse
18:52 aculich howdy @pdurbin @v-dinosaur @pameyer i am just going to listen in on your docker/kubernetes discussion here
18:52 v-dinosaur completely agree on that :)
18:52 pdurbin hey there aculich
18:53 v-dinosaur hey @aculich! I actually have a call in 8 minutes so I think we are wrapping up
18:53 v-dinosaur the container sandwich wrap, for lunchbox safekeeping
18:54 pdurbin Ok, I've been inspired to add a "timezone" column for the who's who in chat spreadsheet: https://docs.google.com/spreadsheets/d/16h3jv24usMGq18495C-JA-yNcQCKiKDa65MTraNDd7k/edit?usp=sharing :)
18:54 pameyer pdurbin: do you believe in daylight saving's time?
18:55 pameyer I was once tripped up by EDT/EST when organizing a webinar with someone in Europe
18:55 pdurbin I'm so confused I'm looking at https://en.wikipedia.org/wiki/Eastern_Time_Zone
18:56 pdurbin pameyer: I'll just copy whatever you put in there. :)
18:56 jri joined #dataverse
18:59 v-dinosaur okay I'll chat later! gotta run!
19:00 pdurbin aculich: thanks for opening https://github.com/IQSS/dataverse/issues/4714 about Jupyter/Binder. I showed it to someone this morning and he asked if there's a live demo we can play with.
19:00 pdurbin v-dinosaur: bye! Thanks!
19:54 pameyer left #dataverse
20:34 pdurbin left #dataverse
20:51 danmcp 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.