Time
S
Nick
Message
07:22 poikilotherm joined #dataverse
08:37 jri joined #dataverse
11:52 pdurbin
poikilotherm: morning. I was thinking... if you have other people on your branch, they'll be able to just run "git pull" if you avoid using "git --force push".
11:53 poikilotherm
Yeah
11:54 poikilotherm
Morning
11:54 poikilotherm
From now on this should be possible
11:55 pdurbin
Ok, good. It also helps in the QA scenario we were talking about earlier. To help answer "What changed in the branch since I last tested it?"
11:57 poikilotherm
It was quite hard to get this deployed to the app server and needed some very ugly changes
11:57 poikilotherm
But those should be all set now
11:58 pdurbin
Ok, yesterday the war was deploying but setup-all and friend were failing.
11:58 pdurbin
friends*
11:59 poikilotherm
Oh did you try to apply the script to the docker image based server?
12:01 pdurbin
No. I'm at "Your branch and 'poikilotherm/5292-small-container' have diverged" ... conflicts because I did a pull.
12:01 poikilotherm
Aye ;-)
12:01 poikilotherm
Please use git reset --hard <remote name here>/5292-small-container instead
12:02 poikilotherm
Oh and maybe a git fetch before
12:02 pdurbin
done, thanks
12:02 poikilotherm
What is the git hash of the commit?
12:03 poikilotherm
Should be 20c25ee5
12:05 pdurbin
yeah. 20c25ee Support Dataverse system properties as ENV vars.
12:05 poikilotherm
Great :-)
12:07 pdurbin
javax.persistence.NoResultException: getSingleResult() did not retrieve any entities.
12:07 poikilotherm
Yeah - the setup-all.sh stuff is missing
12:07 pdurbin
ok, so expected behavior
12:07 poikilotherm
Or in other words: we need a fixture
12:08 pdurbin
that's what they're called in Django
12:08 poikilotherm
Yeah, but it is a more general term, too
12:09 pdurbin
I wouldn't be surprised.
12:10 pdurbin
scripts/database/reference_data.sql could be considered a fixture, right?
12:10 poikilotherm
Totally
12:11 pdurbin
Ok. Without it, SWORD won't work.
12:11 poikilotherm
Altough it mixed
12:11 poikilotherm
It has fixtures inside, but also some migrations (indexes being created)
12:12 pdurbin
yeah, there are some indexes that can't be expressed in JPA
12:20 poikilotherm
Maybe this splited and any DDL plus sequence.sql can be added to Flyway migrations
12:21 poikilotherm
Your current SQL scripts made for migrations could be reused with Flyway, too IIRC
12:21 poikilotherm
But still - I need a fixture for demo/dev purposes
12:21 poikilotherm
That was what I was talking about with pameyer yesterday
12:24 pdurbin
Have you considered jumping on the flyway branch and playing around with it?
12:24 poikilotherm
Not yet
12:25 poikilotherm
First looking at the setup-all.sh stuff - this cannot be fixed with Flyway
13:17 poikilotherm
Oi this is driving me nuts
13:18 poikilotherm
pdurbin do you think it would be OK to have some startup bean that will load initial data from JSON file?
13:18 poikilotherm
(of course based on a env var and possibly more)
13:19 poikilotherm
Building a database snapshot everytime you need to rebuild the image (e. g. because you are developing) is pretty slow
13:19 poikilotherm
And cumbersome
13:19 poikilotherm
Using the existing JSON files seems to be pretty cool
13:20 poikilotherm
Those are a definite reference
13:20 poikilotherm
I am thinking about a bean at startup time that will load those and call the internal Java API without REST to store this initial config in the database
13:28 poikilotherm
I would really like to avoid hacking a script that waits in background for the app to be deployed and then call the REST APIs for initial config
13:48 pdurbin
poikilotherm: have you ever installed Wordpress?
13:48 poikilotherm
Sure
13:48 poikilotherm
Multiple times
13:49 pdurbin
The user experience is pretty nice. Web based.
13:49 pdurbin
"5 minute install" I think they say
13:49 pdurbin
You get a GUI that asks for a database server (could be localhost) and a db username and db password.
13:49 pdurbin
And that's it.
13:49 pdurbin
You're up and running.
13:50 pdurbin
Could what you're talking about tie into a user experience like this?
13:50 poikilotherm
Well perhaps not with a GUI asking for DB credentials
13:50 pdurbin
Oh. Why not?
13:50 poikilotherm
Because it is not PHP :-D
13:51 poikilotherm
You need a persistance framework and JDBC resources
13:51 pdurbin
Even with the magic microprofile config api?
13:51 poikilotherm
I don't think so. I am a bit nearer to this with my JDBC resources living in glassfish-resources.xml now
13:52 poikilotherm
You just need to provide credentials as an ENV var
13:52 pdurbin
Bummer. Weird that this isn't possible with Java. Kudos to PHP I guess.
13:53 pdurbin
Why is it that PHP can do this but Java can't?
13:53 poikilotherm
I dunno if this is impossible, I just don't know if there is an option for this. And from what I know about Java EE and app servers, this is at least not easy...
13:53 poikilotherm
Well you could do that by having an installer based on Java that will write some sort of config
13:53 poikilotherm
That's how PHP does it
13:54 poikilotherm
There could be workaround for this
13:54 poikilotherm
That config could be usable via MicroProfile API
13:54 poikilotherm
But you will need an installer app
13:54 poikilotherm
Java is a compiled language in contrast to PHP ...
13:55 poikilotherm
That definitly makes a difference
13:55 poikilotherm
You can't "deploy" a PHP app like you do with Java EE apps
13:56 pdurbin
Sure. I'm just wondering if we could have various "modes" or whatever when installing Dataverse:
13:56 pdurbin
- default.config for fully unattended installations
13:57 pdurbin
- run the installer at the command line and get prompted for answers, as now
13:57 pdurbin
- the first visitor to the Dataverse web GUI gets a Wordpress-like experience
13:58 pdurbin
- other modes I can't even imagine without more coffee
13:58 poikilotherm
Well, having a default config deployed from within the app would already help, wouldn't it?
13:59 pdurbin
Absolutely.
13:59 poikilotherm
A good example might be JIRA and the like
13:59 poikilotherm
They definitly have a wordpress like installer
13:59 pdurbin
I've never installed JIRA. Or Jenkins.
13:59 poikilotherm
You can even configure your database connection from the browser
13:59 poikilotherm
I don't know how they did that
14:00 pdurbin
So it is possible.
14:00 pdurbin
We don't need PHP for this.
14:00 poikilotherm
Most certainly yes - it might be difficult to achieve, but there seem to be options
14:01 pdurbin
We may never know. JIRA is not open source.
14:03 pdurbin
Is there anything written in Java that's open source that has a Wordpress-like installation experience? Where you put in your database server, username, and password and go?
14:04 poikilotherm
I am not aware of sth.
14:05 poikilotherm
Most certainly the atlassian guys setup a web app that kicks in when no config is present
14:05 poikilotherm
That one writes the config to disk
14:05 poikilotherm
And then they restart the app
14:05 poikilotherm
That last part is a real puzzle for me, dunno how to achieve that
14:06 poikilotherm
At least for all database-stored configuration stuff this is relatively easy to achieve for Dataverse, too
14:07 poikilotherm
One would need to change the startup, if no root dataverse is present
14:08 poikilotherm
The initial config for database resources etc could stay elsewhere
14:08 pdurbin
Right. If no root dataverse is present, it would mean you didn't use default.config or the installer that prompts you with questions.
14:08 poikilotherm
ACtually you could change the default.config to something inside the app
14:08 poikilotherm
At least for everything database related
14:09 poikilotherm
errr - stored
14:09 pdurbin
No objections.
14:09 poikilotherm
Most of the system properties in use right now could be changed to database IMHO, so the external app stuff would be reduced to setting up things like DB
14:10 pdurbin
Sure.
14:10 poikilotherm
Or be replaced by ConfigAPI and use a JDBC source to add these options
14:11 poikilotherm
However...
14:11 poikilotherm
Do you see a chance that people might like the idea of a Startup job or separate module deployed that will pipe in the setup-all.sh absolute basics?
14:12 poikilotherm
(Of course with a switch to turn it on/off)
14:12 poikilotherm
This could either enhance or replace the current setup-all.sh...
14:14 pdurbin
Would it deliver a Wordpress-like experience of being able to complete the installation of Dataverse in a browser? And not break the currently supported installation modes?
14:14 poikilotherm
It could offer one step towards that
14:15 poikilotherm
About breaking: that's what the switch is used for... It defaults to off
14:18 pdurbin
You know I like small steps.
14:18 poikilotherm
Yes
14:18 pdurbin
Three points of contact while rock climbing, right? Not sure if you saw that.
14:19 poikilotherm
Errr - pardon?
14:20 pdurbin
"reminds me of the rock climbing principle that you want to have 3 points in contact at all times" http://irclog.iq.harvard.edu/dataverse/2018-11-28#i_80227
14:21 pdurbin
Hopefully you see what he's getting at.
14:22 poikilotherm
I'm not sure
14:22 poikilotherm
pameyer seems to like some of it and sees potential, but I am totally unsure if he is with me
14:23 poikilotherm
Or if this is just overwhellming
14:23 pdurbin
We should talk it out with him sometime. I have feedback for you too.
14:23 poikilotherm
Name place, time and topic :-D
14:24 pdurbin
The calendar doesn't say he's out today. How long until your weekend starts?
14:25 poikilotherm
~25 minutes
14:26 pdurbin
!
14:26 jri joined #dataverse
14:26 pdurbin
Do you want my feedback now?
14:26 poikilotherm
As I don't know if this is good or bad, I am not sure if this should be a public recorded channel...
14:27 poikilotherm
+on
14:27 pdurbin
It's mostly good! But I can correct some misconceptions.
14:28 poikilotherm
Misconceptions sounds technical, so at least I haven't upset anyway, I hope?
14:28 poikilotherm
-anyway + anyone
14:28 pdurbin
I doubt anyone is upset.
14:29 pdurbin
The first challenge is to get people to read what you wrote. :)
14:29 poikilotherm
Sorry...
14:30 poikilotherm
I just had to braindump
14:30 poikilotherm
This whole stuff is moving around in my head and there are so many things to keep in mind...
14:30 pdurbin
No need to apologize. Pleas see my post about "documentation driven development" at https://groups.google.com/d/msg/dataverse-community/rQWNllAyTu0/RMD0GEFzAgAJ :)
14:32 poikilotherm
Ok, go ahead.
14:33 pdurbin
Ok, for context, I'm giving feedback on your "Future use of Glassfish and/or Payara" post at https://groups.google.com/d/msg/dataverse-dev/L6rNXbbZG0o/UFVfI0_gAQAJ
14:33 pdurbin
poikilotherm: first, thanks for making this post in the first place and calling Dataverse a wonderful open source software project!
14:34 pdurbin
Thanks also to you and your institution for being willing to put effort into improving Dataverse! Time is money!
14:36 pdurbin
I'm with you on up-to-date software. Especially as Dataverse starts to host sensitive data (see https://dataverse.org/goals-roadmap-and-releases ), having patched software will be more important than ever. I feel terrrible that we are holding the 35 production installtion back on an ancient version of Glassfish.
14:37 pdurbin
Yes, a stable platform is very important. In a recent GitHub survey, stability and security are the most important things to users by far. I mentioned this in a talk at JavaOne: http://wiki.greptilian.com/talks/2017/javaone-how-to-run-an-open-source-project/
14:39 pdurbin
I don't object to a microservice architecture but I'm also comfortable calling Dataverse a monolith. Our monolith's first ability to support plugins/add-ons/components in our "external tool" framework. We could and should do much more in terms of modularity. I don't think Java makes this particularly easy but it's possible.
14:41 pdurbin
"more people are opting for running Dataverse in Docker / Kubernetes / OpenShift" is not true for production installations. Of the 35 installations on the map, none of these are using these technologies.
14:42 pdurbin
"others might still prefer the classic installation for good reasons and that should continue to work."
14:43 pdurbin
that should say "must" instead of "should" because, again, everyone in production is using the classic installation. new installations ask how established installations do things
14:44 pdurbin
Yes, Glassfish 4.x is EOL. All "Oracle Glassfish" is EOL. The future for Glassifhs is under the name "Eclipse Glassfish".
14:44 pdurbin
Yes, devs are more interested in a small chunk of moving to a non EOL Glassfish-compatible server before even thinking about open liberty or wildfly.
14:45 pdurbin
Yes, we want the app server to be secure and not EOL.
14:47 pdurbin
It's too strong to say that some people at IQSS don't like the options of MicroProfile ConfigAPI. It sounds interesting to all, I'd say, but some seem willing to wait for it to land in Java EE 9 or whatever the next version is called. This would probably be Eclipse Glassfish 6.
14:47 pdurbin
Yes, devs mostly want to think about code, not app servers.
14:47 pdurbin
features, I'd say. bugs
14:47 pdurbin
grant requirements
14:48 pdurbin
"building an image" is not something that devs think about. not on this floor anyway.
14:48 pdurbin
Yes, it absolutely destroying me that it would take so long to upload changes to DockerHub when I was hacking on openshift stuff. Please deliver me from this agony.
14:49 pdurbin
I don't know what a builder is. "Dockerfiles are for builders".
14:50 poikilotherm
pdurbin: please carry on - I will read up the logs later. I'm sorry, but I have to pickup kids now. Will try to come back later and give some feedback on your feedback :-)
14:50 pdurbin
poikilotherm: this seems like a natural place to pause before I launch into "philosophy behind container images"
14:50 pdurbin
more to come, next week
15:14 drew-jhu joined #dataverse
15:31 * pdurbin
waves at drew-jhu
15:34 drew-jhu
waves back
15:35 pdurbin
drew-jhu: I wanted to point you to a discussion about Google Analytics at https://groups.google.com/d/msg/dataverse-community/gCFpuhShecY/RQINkYYBBAAJ
15:36 pdurbin
The question was "how do I put more than one Google Analytics code in Dataverse?"
15:36 pdurbin
And the workaround is to use a custom footer instead of the ":GoogleAnalyticsCode" database setting.
15:37 pdurbin
You can read about custom footers at http://guides.dataverse.org/en/4.9.4/installation/config.html#custom-footer
15:37 pdurbin
The idea is that you can put whatever Javascript you want in a custom footer.
15:37 drew-jhu
good to know. thanks
15:39 pdurbin
Sure, https://github.com/IQSS/dataverse/issues/4660 is related. Basically, the Google Analytics javascript is frozen in time and we've been talking about maybe having users grab the latest from Google and put it into the footer or a new dedicated file for this purpose.
15:40 drew-jhu
we haven't been using our GA data at all. personally, I'm interested in exploring Matomo as an alternative, rather than committing further to GA
15:40 drew-jhu
but it is good to know about this option
15:40 pdurbin
Ah, we have support for Matomo too but in our guides we still call it by the old name, which is Piwik.
15:43 pdurbin
Please see http://guides.dataverse.org/en/4.9.4/installation/config.html#piwikanalyticsid
15:43 drew-jhu
i did note that with appreciation. at the moment, i am working hard to get our dataverse upgraded and our DOIs migrated to DataCite. hoping to move on to analytics sometime in the new year
15:44 pdurbin
Yeah, I'm not sure when EZID will stop working.
15:45 drew-jhu
my understanding is that they'd like everyone off by the new year. idk whether that is viable or not, but we are striving to meet that goal
15:46 pdurbin
Ah. ok. Do you know there's a new consortium that save you some money on DataCite DOIs? Please see http://dataversecommunity.global/blog/introducing-global-dataverse-community-consortium
15:48 drew-jhu
yes. the manager of our data folks is plugged in there. thx for pointing it out
15:49 pdurbin
Sure. And on the technical site, you might get something out of "Harvard Dataverse Move from EZID to DataCite" at https://groups.google.com/d/msg/dataverse-community/s-6_7cjCs7w/S0gkpmzhAQAJ
15:50 pdurbin
side*
15:50 drew-jhu
that is useful to me. thx!
15:50 pdurbin
sure, I hear it was pretty smooth
15:50 pdurbin
but it did take some coordination with DataCite on the timing
15:51 pdurbin
from what I understand
15:52 drew-jhu
that's what Mara (said manager) was saying. so far, my focus has been on our upgrade, but this info will be very useful for our next step
15:52 pdurbin
cool
15:53 pdurbin
drew-jhu: speaking of upgrades, we just merged a pull request for https://github.com/IQSS/dataverse/issues/4980
16:03 drew-jhu
glad to hear you all are working on this. i have ansibilized the process of upgrading each version one-by-one, but it's not ideal. once DSpace incorporated Flyway, it made upgrading *much* easier, so i'd echo the comments in support of that approach. i'm very close to going live with what we have, so i may not have a good oppotunity to try out your script for a bit.
16:06 drew-jhu
my repo doesn't have my latest changes yet, and may be drifting out-of-sync with IQSS's ansible, but it can be found here: https://github.com/dheles/dataverse-upgrade-ansible-role
16:06 pdurbin
drew-jhu: can you please leave that comment on https://github.com/IQSS/dataverse/issues/5344 ? That flyway issue is in code review right now so there's a good chance your comment will have an impact.
16:11 drew-jhu
https://github.com/IQSS/dataverse/issues/5344#issuecomment-443253269
16:12 pdurbin
Thanks! And your Ansible comment reminds me to ask if you plan to make a pull request for https://github.com/jhu-sheridan-libraries/dataverse-ansible-role/blob/v1.1/roles/dataverse/files/dv-mdblocks/jhuSoftware.tsv
16:17 pdurbin
No pressure but we've been discussing that custom metadata block of yours in the context of "Code Deposit - Github Integration": https://github.com/IQSS/dataverse/issues/2739 :)
16:17 drew-jhu
well, that was one of the last things Tim worked on before he moved on from supporting our Dataverse. We are keeping those metadata entries on our instance, but I'm afraid I don't know much about them, how frequently they are used, and how strongly we recommend them for wider adoption. I'll ask about and see what my better-informed colleagues think about the matter.
16:20 pdurbin
drew-jhu: sounds fine. We are also wondering how strongly you recommend what you've come up with. Thanks. :)
16:23 drew-jhu
i will def get back to you about that, then
16:25 pdurbin
drew-jhu: thanks, no rush. We have plenty to work on. :)
16:28 drew-jhu
heh. word.
16:34 pdurbin
drew-jhu: 15 of your 54 published datasets use "swTitle": https://archive.data.jhu.edu/dataverse/root?q=swTitle:*
16:40 donsizemore joined #dataverse
17:00 drew-jhu joined #dataverse
17:34 pdurbin
donsizemore: another Ansible user. Meet drew-jhu :)
17:40 pdurbin
donsizemore: oh, and I'm still not sure what you meant by "mapping" the other day.
17:52 drew-jhu
waves at donsizemore
17:54 pdurbin
drew-jhu: I added a comment to https://github.com/IQSS/dataverse/issues/5357 in case you'd like to make a pull request. :)
18:15 pameyer joined #dataverse
18:16 pameyer
looks like I missed poikilotherm - which is good; because I didn't want him waiting to start his weekend on me getting caught up in other stuff
18:19 pdurbin
He doesn't sound blocked. "This whole stuff is moving around in my head."
18:19 pdurbin
He may not be a rock climber. :)
18:30 poiki-mobile joined #dataverse
18:31 poiki-mobile
Ding dong. just a short message: thx pdurbin for your elaborated feedback :-)
18:31 poiki-mobile
Really appreciate that. Hope there is more from others coming, too.
18:31 pdurbin
poiki-mobile: sure, but I'm only halfway through your email. :)
18:32 pdurbin
Any feedback on my feedback? What's a builder?
18:32 poiki-mobile
a builder is someone that builds an image
18:32 poiki-mobile
just my wording, dunno if there is some official stuff for this.
18:33 poiki-mobile
maybe use devop instead of builder
18:34 poiki-mobile
anyway: someone that knows the knobs and turns of Dataverse well and likes Docker
18:35 poiki-mobile
admins and users often prefer to be on the "just use upstream image, don't rebuild" side of Docker life
18:36 pdurbin
Ok. I don't think there are any builders on this floor.
18:37 pdurbin
But I definitely appreciate builders. :)
18:37 poiki-mobile joined #dataverse
18:37 poiki-mobile
Maybe pameyer?
18:38 pdurbin
He's only on our floor two days a week. :)
18:39 poiki-mobile
LOL
18:39 poiki-mobile
That should be sufficient... ;-)
18:39 poiki-mobile
Alright guys gotta go... more on Monday. Have a good night!
18:40 pdurbin
pameyer: being into Docker and being into Kubernetes are two very different things.
19:08 donsizemore joined #dataverse
19:22 pameyer
pdurbin: yup
19:24 pdurbin
Like being into floating pool chair vs. being into sailing.
19:25 pameyer
feels more like "do you need a canoe or a submersable aircraft carrier?"
19:26 pdurbin
either way you could drown
19:29 pdurbin
and go places you can't on foot
19:31 Sherry joined #dataverse
19:32 Sherry
Hey, looks like dataverse.harvard is down?
19:32 Sherry
FYI UVa dataverse has gone down three times in three days. Something up with attacks on dataverse sites?
19:32 pdurbin
bloops! really?
19:32 pameyer
harvard dataverse looks up to me
19:33 pdurbin
phew
19:33 pameyer
at least, I get the expected response to https://dataverse.harvard.edu/api/info/version
19:33 Sherry
Ok, it is up now, but not 5 min. ago.
19:33 pameyer
:(
19:33 pameyer
@Sherry - could you remind us which version UVa is on?
19:34 Sherry
I have a screen shot, but I guess can't send via chat?
19:34 Sherry
UVa is 4.8.6
19:34 pdurbin
Sherry: you could upload it to https://imgur.com
19:34 Sherry
@pdurbin show off ;)
19:34 pdurbin
Sherry: if we're all under attack, what's our strategy? Who is our general?
19:35 Sherry
Not me!!!
19:35 * pdurbin
takes a step back
19:35 pdurbin
Three outages in as many days doesn't sound good. Any pattern?
19:38 pameyer
I'd want to rule out gf instability before assuming there's an ongoing pattern of attacks
19:38 sherry joined #dataverse
19:38 sherry
Must have exited my window.
19:38 pdurbin
pameyer: we might be doing this to ourselves :)
19:38 sherry
Here's a couple of screen shots from around 2:30: https://imgur.com/a/B91Vjze
19:39 pdurbin
sherry: what do you see at https://dataverse.harvard.edu/host.txt ?
19:39 sherry
dvn-cloud-app-2.lib.harvard.edu
19:39 pdurbin
ok, I see dvn-cloud-app-1
19:39 pdurbin
maybe one of the two glassfish servers was down
19:39 pameyer
sherry: if you missed it when you were disconnecting, my first suspicion would be glassfish stability
19:39 sherry
dataverse harvard is up now.
19:39 sherry
broken glass?
19:40 sherry
Friday afternoon punchy-ness
19:40 pameyer
:) application server
19:40 pdurbin
heh, rimshot
19:40 pameyer
but it does feel more glass-like than I'd like sometimes
19:41 pdurbin
we tried ironfish but it sank to the bottom of the sea
19:41 sherry
Oh, my!
19:45 pameyer
sherry: do you know if the system administratorss at UVa are monitoring memory usage, or file descriptor counts?
19:47 sherry
Probably not. What should they be looking for? Will restarting the server help? I think memory usage problem was fixed in one of the 4.9.X releases. We are upgrading next month.
19:49 pameyer
restarting the application server may help
19:50 pameyer
there were some stability related improvements between 4.8.6 and 4.9.something
19:50 pameyer
but if those fixes were the only stability issues, my suspicion is that app servers at dataverse.harvard.edu wouldn't be freezing every so often
19:51 pameyer
for what to look for - possibly memory or swap usage over time, and output of `lsof | wc -l`
19:51 pameyer
pdurbin or donsizemore may have some additional suggestions
19:53 pdurbin
I put my monitoring suggestions in http://guides.dataverse.org/en/4.9.4/admin/monitoring.html
19:54 donsizemore joined #dataverse
19:54 pdurbin
sherry: this is my home server but your sysadmins should be able to retrieve data like this on the health of your server over time: http://munin.greptilian.com/greptilian.com/server3.greptilian.com/index.html
19:56 sherry
Thanks for all the tips. I'll take a look and give to my sysadmin. Now on to put out a Grad Student Fire in submitting to one of our other repositories!
19:56 pameyer
good luck with both
19:57 pdurbin
phew, the heat is off us, for now :)
20:00 pameyer
at some point, the glassfish falling over problem probably will need to get sorted
20:01 pdurbin
yeah, it would be nice if we didn't need a restart script
20:06 pameyer
did the restart script ever make it into the guides?
20:06 pdurbin
nope
20:08 pdurbin
I have a copy of it in my email somewhere, I think.
20:09 pameyer
if enough folks start seeing things like this, it might be worth adding it.
20:09 pdurbin
looks like it kicked off around 2:33 which lines up pretty well with sherry saying that she saw an outage at 2:30
20:09 pameyer
making it unnecessary would probably be better, but would also be more effort
20:10 pameyer
yeah - 5m interval on HU DV always has me thinking about the bounce script
20:10 pameyer
even though I have no idea if it's actually using a 5m interval
20:13 pdurbin
well, the interval comes from nagios, I beleive. Or at least it did in the version of the script that I have from 2015.
20:16 pdurbin
if [ $2 == "HARD" ]; then
20:30 donsizemore
@pdurbin does the restart script simply call a stop-domain / start-domain?
20:30 donsizemore
(if it's file descriptors for example, the stop-domain command should fail)
20:31 donsizemore
ultimately, i'm asking about stderr from the script
20:33 pdurbin
donsizemore: well, it uses the init script. calls a "stop" and does a "sleep 20" followed by a "start".
20:34 donsizemore
so likely not a process limit then. coolies
20:34 pdurbin
Oh, I should add that if there's no life after that "sleep 20" it goes into "kill -9" mode like any good script. :)
20:41 pameyer
I've gotten into the habit of using jps to fingure out if glassfish has actually stopped when told to
20:46 pdurbin
that fish won't behave
20:48 pameyer
see - this is the real reason for glassfish5 vs payara5. I don't know what kind of animal a payara is :)
20:51 pdurbin
heh
20:52 pdurbin
donsizemore: heads up that "Import a Dataset into a Dataverse with a DDI file" just got merged: https://github.com/IQSS/dataverse/pull/5170
21:08 donsizemore
i saw that =) i'm background-writing some python to translate IPUMS' particular structure
21:10 pdurbin
nice
21:29 pameyer
pdurbin: more messages crossing in the interwebs ;) https://github.com/IQSS/dataverse/issues/4714
21:31 pdurbin
pameyer: Craig just gave you a "hooray" emoji
21:32 pdurbin
Seems like a good way to start the weekend. Take care, all. :)
21:34 pdurbin left #dataverse
