Time
S
Nick
Message
07:02 juancorr joined #dataverse
12:46 pdurbin joined #dataverse
13:15 donsizemore joined #dataverse
13:36 pdurbin
donsizemore: no school today? https://chapelboro.com/news/unc/unc-classes-canceled-monday-due-hurricane-florence
14:00 donsizemore
@pdurbin i waited out the tornando warning and came in a little late. saving my vacay for vacay!
14:00 pameyer joined #dataverse
14:02 pdurbin
way to be
14:04 pdurbin
donsizemore: anyway, your boss couldn't make it but I gave Odum a shout out in my talk: https://twitter.com/philipdurbin/status/1040267883090202625
14:15 pdurbin
pameyer: buried in an email I just sent is a link to this "Scientific honesty and publicly shared lab notebooks" paper, which you may or may not find interesting: https://doi.org/10.15252/embr.201846866
14:23 donsizemore
@pdurbin congrats on the talk! jon could've made it up there, but probably not back. not sure if his power is back on yet at home
14:23 pdurbin
yikes
14:24 pameyer
pdurbin: walking through a checklist now; will check it out later
14:24 pdurbin
pameyer: no rush
14:25 pdurbin
donsizemore: at least he didn't get munched by a shark, which happened on Saturday on Cape Cod. First time in 80 years.
14:27 donsizemore
@pdurbin i ran into that no-shoulder upgrade snafu on a test box but had blamed me
14:28 pdurbin
Ah. Maybe we should fix it.
14:28 pdurbin
It's probably pretty confusing for someone doing an upgrade until it's fixed. Weird behavior, I assume.
14:36 donsizemore
@pdurbin i can submit a PR with an example of what things "should" look like for many installations, and how to make it so?
14:40 pdurbin
donsizemore: a doc change? Sure! I hate to ask but could you make a new issue as well?
14:49 donsizemore joined #dataverse
15:09 donsizemore joined #dataverse
15:14 donsizemore
@pdurbin actually, looking back, I think it was a "me" problem, as most things are. i'm all about elaborate documentation, tho
15:15 pdurbin
donsizemore: we love the docs you add. And the code. Love it all.
15:15 pdurbin
by the way, we may have some Ansible questions for you shortly
15:23 donsizemore
@pdurbin yeah, i pushed a minor commit this morning for matthew. he encountered some conditions i admittedly hadn't ran into
15:45 pdurbin
donsizemore: thanks! I just let him know.
16:53 jamie-ucla joined #dataverse
16:55 jamie-ucla joined #dataverse
17:15 jamie-ucla88 joined #dataverse
17:15 dataverse-user joined #dataverse
17:25 andrewSC joined #dataverse
17:37 donsizemore joined #dataverse
17:40 dataverse-user left #dataverse
18:02 jamie-ucla joined #dataverse
18:02 jamie-ucla
Last week I was discussing a problem we were having with our AWS/S3 test instance (<http://54.67.118.35:8080 >). We wern't able to upload files. What seemed to fix the problem was to delete the jvm option dataverse.files.directory, which was set to the default location: /usr/local/dvn/etc.. I'm wondering what the dataverse.files.directory should be set to for S3 storage. I'm uncomfortable just deleting something without knowing wh
18:05 pdurbin
jamie-ucla: hi! We were *just* discussing your issues this morning. I think you opened a support ticket but I haven't looked yet.
18:05 pdurbin
Here it is: https://help.hmdc.harvard.edu/Ticket/Display.html?id=267029
18:07 jamie-ucla
It's fixed in that we can upload now but I'm worried about just deleting a jvm option without knowing what should be there or the consequences down the line.
18:09 pdurbin
Yes, it's really weird that deleting that JVM option fixed it.
18:10 pdurbin
jamie-ucla: at minimum, I feel like we need a doc change. Can you please open an issue at https://github.com/IQSS/dataverse/issues about the problem you had?
18:11 jamie-ucla
I was going to open an issue but I'm wondering if I shouldn't be replacing rather than deleting. Or am I worring too much?
18:12 pdurbin
jamie-ucla: ask pameyer if it's good to be paranoid. ;)
18:12 pdurbin
You're doing your due diligence. It's perfect.
18:19 pdurbin
jamie-ucla: but if you just start with a description of what you were trying to do (move from having files on disk to having files in S3 or whatever) in a GitHub issue, that would be great.
18:20 jamie-ucla
ok, working on it
18:21 pdurbin
thank you!
18:22 pdurbin
That JVM option is referenced quite a bit in the code.
18:22 pdurbin
I wouldn't be surprised if you found a bug.
18:32 pameyer
failure points do happen
18:33 pameyer
and of course, it's always the ones that you don't check for or handle that cause problems
18:34 pameyer
my first thought would be that dataverse.files.directory is meanlingless when using s3 driver; but some of the code is likely using it's presense/absence to indicate which storage driver it should use
18:34 pameyer
but I don't run on s3, so I don't have direct info
18:38 jamie-ucla
I've got an issue now (<https://github.com/IQSS/dataverse/issues/5062 >)
18:45 pdurbin
jamie-ucla: looks perfect. Thank you!
18:56 jamie-ucla left #dataverse
19:02 donsizemore joined #dataverse
19:12 pameyer
pdurbin: minio was one of the things I looked at when brainstorming about how to make the dcm jenkins pipelines work for dcm-s3
19:12 pdurbin
pameyer: ok, interesting
19:12 pameyer
seemed to work as expected with aws-cli; which it hadn't the previous time I'd looked at it
19:13 pdurbin
pameyer: that reminds me, can you please take a look at https://github.com/IQSS/dataverse/pull/5061 ?
19:14 pameyer
pdurbin: yeah, but likely not today
19:14 pdurbin
no worries
19:14 pdurbin
we're backing out of your `export LANG="en_US.UTF -8"` fix
19:14 pameyer
I almost started the migration checklist 4:30pm last Fri - this morning I'm glad I didn't
19:15 pameyer
that was more of an ugly hack than a fix
19:15 pdurbin
ooo migration checklist
19:15 pdurbin
if you need another set of eyes to look at it, lemme know
19:16 pameyer
other than needing to talk a walk to push a button for a box that didn't reboot cleanly, it's mostly going according to plan
19:16 pameyer
that, and re-testing migration since more datasets were released since last time I tested it
19:17 * pameyer
knocks on wood
19:17 pdurbin
:)
19:20 pameyer
ok - there was also creating a flood of tickets when I migrated labs over. but RT has a CLI
19:21 pdurbin
yeah, and a REST interface
19:22 pameyer
I'm pretty sure the CLI is just a wrapper over the REST interface - kinda like ovirt's CLI
19:22 pdurbin
probably
20:13 Jim20 joined #dataverse
20:56 dataverse-user joined #dataverse
21:16 donsizemore joined #dataverse
21:36 pameyer left #dataverse
