Time
S
Nick
Message
11:26 donsizemore joined #dataverse
12:14 poikilotherm
Morning donsizemore
12:23 donsizemore
@poikilotherm how are you, sir?
13:13 donsizemore joined #dataverse
13:22 poikilotherm
Hi donsizemore :-) Sorry, a meeting hold me back from posting here :-)
13:22 poikilotherm
Sir are you a friend of short and nice URLs?
13:40 pameyer joined #dataverse
14:03 donsizemore
I like to see what a client is calling
14:05 donsizemore
@pdurbin if it weren't dataverse-users I would report Hanna Ballard's SPN note as spam.
14:40 pdurbin joined #dataverse
14:41 pdurbin
donsizemore: another one?
14:42 donsizemore
"By investing $5,000 per year, your organization will join a vibrant community of practice committed to ensuring long-term access to software through community engagement, infrastructure support, and knowledge generation."
14:42 pdurbin
donsizemore: you added the code of conduct. Is it a violation?
14:43 donsizemore
CoC went in the GitHub repo. this is $Goog's free mailing list
14:43 donsizemore
I'm more annoyed by phrases like "community of practice" and "knowledge generation" then the membership sell
14:43 pdurbin
heh
14:44 pdurbin
"Instances of abusive, harassing, or otherwise unacceptable behavior may be reported by contacting the project team at support at dataverse dot org" ... maybe we should do that?
14:45 pdurbin
donsizemore: also, there's some discussion in #dv-general about this.
14:56 poikilotherm
pdurbin donsizemore I like links that are short and nice to look at. dataverse.org is great, but IMHO longer links like https://github.com/GlobalDataverseCommunityConsortium/dataverse-ansible are quite tedious.
14:56 poikilotherm
For K8s I send people to the project to reuse objects from there
14:57 poikilotherm
gdcc.io was cheap. :-P
14:59 poikilotherm
("there": I meant the future dataverse-kubernetes repo at GDCC)
14:59 poikilotherm
(No k8s in ansible - yet)
15:00 pdurbin
"Use lowercase and keep it short" is one of Unix's 10 lesser tenets. :)
15:01 pdurbin
From Mike Gancarz's UNIX Philosophy: https://www.humbug.org.au/Unix101
15:02 pameyer
yeah - but there can be a tradeoff with url shorteners and gratuitous analytics
15:02 pdurbin
touche
15:02 poikilotherm
pameyer: no plan to use a shortener like bit.ly, git.io et al
15:03 poikilotherm
Just shorten some host names and pathes
15:03 poikilotherm
Maybe streamline some things
15:03 pameyer
great :)
15:03 poikilotherm
Used .io because I wanna see it for the technical scope
15:04 pameyer
pdurbin: btw, you may need to bump the memory in docker on mac to get the app server stable enough for ITs
15:04 poikilotherm
Could be sth. like k8s.gdcc.io /personas/demo providing a shortcut for github.com/GlobalDataverseCommunityConso​rtium/dataverse-kubernetes/personas/demo
15:05 pdurbin
pameyer: heh, I was wondering if you noticed that I tried docker-aio yesterday. It looks like Docker is getting 2 GB right now. What do you suggest?
15:06 poikilotherm
And I wanna contribute a community open anchore engine for scanning for sec problems
15:06 pdurbin
poikilotherm: how do you like http://dataversecommunity.global ?
15:07 poikilotherm
pdurbin too long :-D
15:07 pdurbin
I figured. :)
15:07 pameyer
@poikilotherm - fyi, there was pretty much no interest in my owasp scanner issue
15:08 pameyer
pdurbin: I'd go with at least 4 GB
15:08 poikilotherm
pdurbin with 4.20 I experienced a huge boost in memory requirements than with 4.19 for my dev env. Had to use at least 1.5 gig. https://dataverse-k8s.readthedocs.io/en/develop/day1/resources.html#memory
15:09 poikilotherm
(This is dataverse alone, no db nor index)
15:10 poikilotherm
I suspect a memory leak, because I did not get any heap dumps or OOM exceptions before the kernel OOM killer shot the appserver in the head
15:10 poikilotherm
Dug a bit into this, but didn't want to do more investigation as 5.0 is out now. https://dataverse-k8s.readthedocs.io/en/develop/day1/resources.html#fighting-out-of-memory-situations
15:14 poikilotherm
Guys I'm off for a while, appointment waiting. Would love some more chat on this stuff later, share thoughts and experiences!
15:14 donsizemore
@poikilotherm I would welcome a pointer link off gdcc.io - as @pameyer points out I'd only click a bit.ly or other such link with Javascript disabled.
15:14 pameyer
informally, switching to payara made things a lot more stable
15:14 donsizemore
@pameyer say that again at 2pm =)
15:16 pameyer
:) I'm not going to abuse somebody's prod install to test it ....
15:17 pdurbin
fFrom Slack it sounds like Harvard Dataverse will be upgraded soon. The demo site is already on 5.0 and Payara.
15:18 donsizemore
why I wanted the verbal affirmation for 2pm. words have power!
15:20 pdurbin
:)
15:21 pdurbin
pameyer: I bumped it to 4 GB but It's hanging at this step: Step 16/46 : RUN cp /tmp/dv/pg_hba.conf /var/lib/pgsql/data/
15:23 pameyer
huh - I'll jump through the hoops so I can check the UI on my mac desktop
15:23 pameyer
... if you're on 10.15 though, there may be gremlins there I haven't run across
15:25 pdurbin
Yeah, I am on 10.15. Again, I got pretty far yesterday. It's just that they tests didn't pass. Maybe I'll kill this and try it again.
15:31 pdurbin
pameyer: ok, I restarted Docker and got a good deployment with the API showing the version.
15:34 pdurbin
While I'm waiting for the test suite to run in docker-aio, I thought about installing VirtualBox and Vagrant on this new shiny laptop yesterday but decided to wait a while. I'm wondering if VirtualBox will junk up this new machine or if it's not a big deal.
15:36 pameyer
I wouldn't expect it to junk up much - but would expect a relatively high chance if it not working for problematic to diagnose reasons
15:37 pdurbin
ok, that's reassuring
15:38 pdurbin
pameyer: the tests passed: Tests run: 133, Failures: 0, Errors: 0, Skipped: 4
15:39 pameyer
pdurbin: awesome
15:40 pdurbin
pameyer: oh, on a related note, we ( mostly donsizemore ) have done a lot with testing since you opened https://github.com/IQSS/dataverse/issues/4332
15:42 pameyer
yeah - I'm out of the loop, but have still noticed small cool things creeping in :)
16:21 donsizemore
@pdurbin did you see that GitHub CodeSpaces is going to offer Visual Studio Code *and Docker* allegedly in your browser?
16:21 donsizemore
https://www.aristotlemetadata.com/blog/2020/08/first-impressions-github-codespaces.html
16:24 donsizemore
@pdurbin also, what were you suggesting by 'a YAML snippet with both "s3" and "file"'?
16:24 pdurbin
I hadn't seen this but it looks neat. Might be good for drive by contributors. I'll still probly use vi though. :)
16:25 pdurbin
Well, Jim made it sound like "multi-store" is supported.
16:25 pdurbin
And if so, that's cool.
16:26 pdurbin
I guess having two or three "file" stores would work also, to demonstrate multi-store. Does that make sense?
16:26 donsizemore
well, right now i just want to correct the default (it won't choose s3), and stop it from automatically deleting the file storage_id if s3 is enabled
16:26 donsizemore
which will make jim happy
16:27 pdurbin
ah, ok
16:27 pdurbin
If you're both happy, I'm happy.
16:27 donsizemore
then on another rainy day i'll add in a flag to specify deletion of the file storage_id in... s3? new dataverse.storage section?
16:28 pdurbin
Right because you have to delete a JVM option.
16:29 donsizemore
i'm a little edgy about allowing the deletion of the file store (which is why i'm asking) but jim says it's fine
16:30 pdurbin
The default file store is all I need, most of the time, for dev.
16:30 pdurbin
But I'm glad these fancy new options exist. :)
16:31 donsizemore
Jim uses S3... once in a while ;)
16:32 pdurbin
:)
19:53 poikilotherm
Hey folks :-)
19:53 poikilotherm
Anyone still around?
19:54 pdurbin
Yes, but we've been struggling all day with bad news.
19:54 pdurbin
We've lost a four footed friend.
19:55 poikilotherm
Doh!
19:55 poikilotherm
Bad news indeed
19:56 pdurbin
Sounds like we might dedicate the 5.0 release to him.
19:57 poikilotherm
Yes please Sir!
19:57 poikilotherm
So sorry for @djbrooke
19:58 pdurbin
yeah
19:59 poikilotherm
Sad times :-/
19:59 pdurbin
T-T
19:59 pdurbin
Is there something we can help you with?
19:59 donsizemore
@pdurbin I think it should be part of buildNumber.properties
19:59 poikilotherm
Nope, just walked in for a chat about memory.
19:59 poikilotherm
But this is not important
20:00 poikilotherm
Losses are important
20:00 poikilotherm
Good idea donsizemore
20:00 poikilotherm
Please re-tag
20:01 poikilotherm
Do a master hotfix and re-tag
20:01 pdurbin
:)
20:02 poikilotherm
It's a breaking change for you all. Both release and your loss.
20:03 poikilotherm
So sad. Your pal deserves a memorial
