Time
S
Nick
Message
03:57 dataverse-user joined #dataverse
10:24 juancorr joined #dataverse
12:36 donsizemore joined #dataverse
12:57 donsizemore
@pameyer i would like to speak with the folks who put the "press enter to reboot" logic in the dell idrac firmware update routine.
13:23 pdurbin
donsizemore: morning. I might have an Ansible question.
13:23 donsizemore
@pdurbin yes sir
13:24 pdurbin
donsizemore: please check out this "TODO: allow arbitrary repo, not just IQSS. Will require changing it on the ansible side as well" at https://github.com/IQSS/dataverse/blob/19e2915128f13fee46afe3d63fa4e24d34705a8c/scripts/installer/ec2-create-instance.sh#L4
13:25 pdurbin
You'll see that "IQSS" is hard coded in a couple commands.
13:34 donsizemore
oh, that should be pretty simple. so you just want the github url to become a variable?
13:38 pdurbin
donsizemore: yeah, so that arbitrary branches can be spun up from non-IQSS repos such as https://github.com/OdumInstitute/dataverse
13:42 donsizemore
@pdurbin push first, test later https://github.com/IQSS/dataverse-ansible/commit/7ea1f7a94e203a3c8a21bc11b3da3463e20bd52c
13:43 pdurbin
donsizemore: cool, but now I have another question
13:44 pdurbin
sed -i "s/branch:/branch: $BRANCH_NAME/" dataverse/defaults/main.yml
13:44 pdurbin
The script I linked to above has that sed line in it.
13:45 pdurbin
It seems weird to me that we have to modify that main.yml file on disk. Shouldn't we be able to pass an argument to Ansible?
13:46 donsizemore
@pdurbin that may have been experimentation on matthew's part? i thought he was passing extra variables to ansible-playbook but ansible's inverted variable priority may have bitten him
13:47 pdurbin
The sed line is what he got to work. I'm wondering if there's another way. Passing an argument, setting an environment variable. Something.
13:49 pdurbin
he did say he got bit by something
13:53 donsizemore
the cleanest thing would be to pass --extra-vars "dataverse.git.branch=9999-get-my-margarita" which ansible claims should always win the precedence war https://docs.ansible.com/ansible/2.6/user_guide/playbooks_variables.html#variable-precedence-where-should-i-put-a-variable
14:20 pdurbin
ah "extra vars (always win precedence)"
14:20 pdurbin
yeah, that sounds like what we want, right?
14:22 donsizemore
if ansible does what it says =)
14:27 donsizemore
one other question he had about dataverse-ansible, for which i don't have a good answer due to the dataverse release cycle: dataverse-ansible downloads the dvinstall.zip corresponding to dataverse.version
14:29 pameyer joined #dataverse
14:29 donsizemore
if a branch is specified, it clones that branch, builds the warfile and overwrites the one exploded from dvinstall.zip. however, if he needs to specify say a custom schema.xml or other anciliary file, i suppose i could add in variables for them? (he noticed that he deployed develop but got install files for 4.9.2)
14:34 pdurbin
Buh. yeah, I guess we'll need to handle that stuff in the future. That'll be some work.
14:34 pameyer
this may be a silly question, but why not do a (optional) dataverse-build role; and set that ahead of the dataverse-install role?
14:34 pdurbin
That's why we're calling the current effort "spinning up arbitrary branches, phase one"
14:35 pameyer
assuming that the dataverse-install starts from dvinstall.zip from somewhere; the build role could create that. if it's missing, have the dataverse-deploy role download it
14:35 donsizemore
@pameyer not silly at all. off the top of my head it's because dataverse-ansible doesn't simply use the install script, it dives in to glassfish-setup
14:36 donsizemore
ooh, i do like building custom dvinstall.zips
14:36 pameyer
right - not the script, just use the zip
14:37 pameyer
and yeah; having to physically push buttons isn't the the greatest sometimes
14:38 donsizemore
@pameyer and today i learned you have a Makefile: https://github.com/IQSS/dataverse/blob/cb2d308b17a1a70e9961e009682589ad3d8599d2/scripts/installer/Makefile
14:40 donsizemore
@pameyer might be quicker to just build the custom makefile when a branch is specified. let me poke at that, thank you for the suggestion
14:42 pdurbin
donsizemore: for what it's worth, I recently documented that Makefile at http://guides.dataverse.org/en/4.9.2/developers/making-releases.html#make-artifacts-available-for-download
14:43 donsizemore
the Makefile wants the warfile to exist... so instead i could make the branch tasklist copy out each file specified. only 6 totaly
14:44 pdurbin
whatever is sane to you and Pete. He and I have talked a lot about mvn vs make
14:46 donsizemore
@pdurbin solr is the only service which would require restarting. i can wedge this in pretty easily... but i tend to choose the shortest/simplest path.
14:50 pdurbin
I'm not sure what you're up to exactly but I trust you. Here comes your rain.
14:50 donsizemore
no! no more rain! my mahonia holly can't take it
14:52 pameyer joined #dataverse
14:53 pameyer
@donsizemore - hopefully the rain up here means you folks get a break
14:55 donsizemore
@pameyer i think i can do what you're suggesting pretty easily. will report back this afternoon
14:55 pameyer
great - thanks
14:56 pameyer
two things to keep in mind though 1) you might want to do `make clean` before `make` with that Makefile, it doesn't realize when the war file is newer than the zipfile 2) sometimes I have ideas that seem to make sense, but don't always - so there's always the chance this is a blind alley
14:57 donsizemore
@pdurbin i was responsible for perpetrating maven.ibiblio.org, you can hard-code it during a maven emergency, and yell at me if it falls over
15:06 pdurbin
donsizemore: I'm sure you got way more rain that this but here you go, the view out my window: https://youtu.be/Jj6RC8eC-MQ
15:14 donsizemore
@pdurbin our deputy director measured 11" in two days at his house, just north of chapel hill
15:19 donsizemore
@pdurbin and more than anything i'm envious that your office has a window
15:29 pdurbin
wow, 11 inches
15:31 donsizemore
@pdurbin the river near goldsboro crested at 26' above flood stage, and i believe the cape fear near wilmington rose above the gauge
15:32 pameyer joined #dataverse
15:42 pdurbin
yikes
15:50 pdurbin
andrewSC bjonnh bricas juancorr pameyer: wow, check this out: https://static.scholarsportal.info/static/dataverse/dataverse-dist.png
15:51 pameyer
interesting
15:52 pameyer
has me wondering about things I've heard with harvesting timers and where they run
16:12 pdurbin
via https://groups.google.com/d/msg/dataverse-community/tfd6BZPgk-M/w3QEB8q5CAAJ
16:43 donsizemore joined #dataverse
17:14 donsizemore
@pdurbin who make-a-duh mistake? me?
17:14 donsizemore
@pdurbin will be fixed in my next commit
17:15 pdurbin
awesome. kudos to Matthew for figuring it out
17:15 pdurbin
thanks!
17:21 donsizemore
i shouldn't be allowed to computer before noon. or after, for that matter.
17:27 pdurbin
:)
17:37 pameyer joined #dataverse
17:55 Mahsa joined #dataverse
18:41 * pdurbin
waves at Mahsa
18:43 pameyer joined #dataverse
18:58 qqmyers joined #dataverse
19:00 qqmyers
Does anyone happen to know if DOIDataCiteRegisterCache serves a purpose? I see that it gets populated when a dataset is created and Datacite is the doi provider, but I don't see it being used.
19:01 qqmyers
Which I think is a good thing since there are no entries for datasets created when the provider was ezid and you switch to Datacite.
19:02 pameyer joined #dataverse
19:05 pdurbin
qqmyers: hi! I just checked with sekmiller and it sounds like it might not be used.
19:07 pdurbin
it sure seems to be used a lot in DOIDataCiteRegisterService though
19:15 qqmyers joined #dataverse
19:16 qqmyers69 joined #dataverse
19:18 qqmyers69
Thanks - I think it is a good thing it is not used - like I said, if entries had to be there to reserve DOIs, migrating from ezid would be a problem. As it is, I've checked and you can't reuse a DOI if it exists in the database (in the dvobject table) or at the service, but the cache is not checked.
19:39 Mahsa67 joined #dataverse
19:41 Mahsa joined #dataverse
19:43 Mahsa
Hi everyone, The other day Pameyer pointed that there is an issue with our Glassfish <-> Apache fronting (http://irclog.iq.harvard.edu/dataverse/2018-09-05#i_72053
19:44 Mahsa
I fixed a few things, and I am now seeing this error
19:44 Mahsa
GRIZZLY0013: Exception during FilterChain execution java.lang.IllegalStateException: Unexpected mark=18245 at org.glassfish.grizzly.http.ajp.AjpMessageFilter.handleRead(AjpMessageFilter.java:89)
19:45 Mahsa
Does anyone have any idea what is this error ?
19:51 pdurbin
Mahsa: hi, pameyer and I are in a meeting. Sorry.
19:52 Mahsa
That is fine. Thanks
19:53 donsizemore
@pameyer @pdurbin i have something that May Suck but Should Work(tm), but i want to do more testing
20:20 pameyer
Hi Mahsa - does that exception show up on all requests?
20:21 pameyer
@donsizemore - there's always more testing to do :)
20:36 pameyer joined #dataverse
20:44 Mahsa
Hi, Yes, it shows up when I got to the website , basically, the website does not load on port 8080
20:47 Mahsa joined #dataverse
20:48 pameyer
@Mahsa - I don't see that exception in my server.log; but I'm wondering if it's related to the apache ajp configuration.
20:48 pameyer
do you have a line like `ProxyPass / ajp://localhost:8009/` in your apache configuration?
20:49 Mahsa
Yes
20:49 Mahsa
I have that line
20:49 Mahsa
One quick question,
20:50 Mahsa
the JK connect should be enabled on http listener1 (8080), is that correct ?
20:50 pameyer
not sure what you mean by "JK connector"; but the AJP connector shouldn't be on 8080
20:50 pameyer
unless you've changed some of the default glassfish settings
20:50 Mahsa
I meant JK listener
20:51 Mahsa
Apache mod-jk listener
20:52 pameyer
for connecting apache httpd to glassfish?
20:52 Mahsa
Yes, I have 2 http listeners and and JK connector in my http listeners
20:52 Mahsa
Yes,
20:53 Mahsa
My http listener 1 (on glassfish, port 8080), has a JK listener option which I think should be enabled
20:53 Mahsa
for fronting glassfish with Apache
20:53 Mahsa
is this correct ?
20:54 pameyer
ah - gotcha. I took a quick look at my domain.xml
20:54 Mahsa
I think I have to do some glassfish configs to make things work
20:54 pameyer
for `<network-listener jk-enabled="true" protocol="http-listener-1" port="8009" name="jk-connector" thread-pool="http-thread-pool" transport="tcp"></network-listener>` in domain.xml, I'm using `ProxyPass / ajp://localhost:8009/` in my apache httpd configuration file
20:55 Mahsa
Yes, my setting is identical to this
20:55 pameyer
glassfish is serving web traffic on 8080 in this case (aka - I can get to dataverse on port 8080)
20:56 Mahsa
I get the empty response on 8080
20:56 Mahsa
But things on working on https /port 8181
20:56 Mahsa
*are working
20:56 pameyer
ok - it sounds like you configured glassfish to do ssl; which isn't something I've done
20:56 pameyer
I usually do https/ssl to apache
20:57 Mahsa
I think 8181 is the second http listener on glassfish, it comes with default
20:57 Mahsa
I did not do any configs
20:58 Mahsa
my point was, it seems the problem is not with the dataverse application, but with fronting glassfish on 8080 with apache using the JK connector
20:58 Mahsa
so let me put it this way
20:59 Mahsa
when I set the jk-enabled to "false"
20:59 pameyer
got it. I just checked, and I see dataverse on 8181 from glassfish too (without doing any configuration)
20:59 Mahsa
The website is working with no errors , since JK is disabled
21:00 Mahsa
great
21:01 pameyer
by "the website", you're referring to apache -> glassfish (with JK disabled)?
21:03 Mahsa
I meant dataverse loads on glasssfish successfully, I have not had any luck with Apache -> glassfish yet
21:04 pameyer
ok - I think I understand.
21:05 pameyer
in your case, when you set jk-enabled to true, are you failing to load dataverse on glassfish?
21:05 Mahsa
yes
21:06 Mahsa
and I see the error above
21:06 Mahsa
in glassfish log
21:06 Mahsa
java.lang.IllegalStateException: Unexpected mark=18245 at org.glassfish.grizzly.http.ajp.AjpMessageFilter.handleRead(AjpMessageFilter.java:89)
21:07 pameyer
does your network-listener entry for the jk-connector look different from the one I posted?
21:08 Mahsa
I think it is similiar but will just double check
21:08 pameyer
and just to confirm, when the jk-connector is disabled; 8181 on glassfish works, but 8080 is empty
21:09 Mahsa
correct, 8181 works, 8080 is empty
21:11 pameyer
one thing that might provide more information would be to do `curl -i http://localhost:8080 ` - this will show any headers that the empty response may have
21:11 Mahsa
"empty reply from server
21:12 pameyer
odd - I'd expect a host header at least
21:13 pameyer
server header I mean; not host header
21:14 Mahsa
yes, I see
21:18 Mahsa
<network-listener protocol="http-listener-1" jk-enabled="true" port="8080" name="http-listener-1" thread-pool="http-thread-pool" transport="tcp"></network-listener>
21:18 Mahsa
I just copied/pasted this from domain.xml
21:19 pameyer
could you try changing the port to 8009 and restarting glassfish?
21:19 pameyer
I copy/pasted my example too :)
21:20 Mahsa
Sorry, I have another line like this
21:20 Mahsa
<network-listener protocol="http-listener-1" jk-enabled="true" port="8009" name="jk-connector" thread-pool="http-thread-pool" transport="tcp">
21:20 pameyer
hmm - 2 jk-connector lines?
21:20 Mahsa
I copied the wrong one first, sorry :-)
21:20 pameyer
no problem
21:20 Mahsa
Thst is right, let me check
21:21 pameyer
I'm only seeing one jk-connector line in mine
21:21 pameyer
[root
21:21 pameyer
79: <virtual-server network-listeners="http-listener-1,http-listener-2,jk-connector" id="server"></virtual-server>
21:21 pameyer
251: <network-listener jk-enabled="true" protocol="http-listener-1" port="8009" name="jk-connector" thread-pool="http-thread-pool" transport="tcp"></network-listener>
21:21 pameyer
I'm wondering if the problem could be listening for http and ajp both on port 8080
21:21 Mahsa
can you pleasse copy the http-listener-1 line too ?
21:22 Mahsa
Sorry, You dont have that :D LOL
21:22 pameyer
[root
21:22 pameyer
79: <virtual-server network-listeners="http-listener-1,http-listener-2,jk-connector" id="server"></virtual-server>
21:22 pameyer
230: <protocol name="http-listener-1">
21:22 pameyer
248: <network-listener protocol="http-listener-1" port="8080" name="http-listener-1" thread-pool="http-thread-pool" transport="tcp"></network-listener>
21:22 pameyer
251: <network-listener jk-enabled="true" protocol="http-listener-1" port="8009" name="jk-connector" thread-pool="http-thread-pool" transport="tcp"></network-listener>
21:22 pameyer
273: <virtual-server network-listeners="http-listener-1, http-listener-2" id="server">
21:22 pameyer
403: <protocol name="http-listener-1">
21:22 pameyer
436: <network-listener protocol="http-listener-1" port="${HTTP_LISTENER_PORT}" name="http-listener-1" thread-pool="http-thread-pool" transport="tcp"></network-listener>
21:22 Mahsa
I got confused
21:22 Mahsa
Yes, I see
21:23 Mahsa
I think I know why I have two lines here
21:23 Mahsa
That is probably because I enabled JK on my listener1
21:23 pameyer
that would make sense
21:24 Mahsa
if I remove it, maybe one would disapear, So it seems I have created more problems, to solve Glassfish -> Apache issue :D LOL
21:24 Mahsa
sorry about that
21:24 pameyer
it's worth a try
21:24 pameyer
no problem :)
21:26 Mahsa
I just removed it, let me restart my glassfish
21:27 pameyer
I'll keep my fingers crossed for good luck
21:28 Mahsa
I will be back to step 1 even if it works LOL
21:29 Mahsa
Ok,
21:29 Mahsa
So they are both still here which is a good news
21:29 Mahsa
This might be the reason the fronting is not working
21:30 Mahsa
Can you please do me a favor ?
21:30 pameyer
well, one problem at a time ;)
21:30 pameyer
possibly - what do you have in mind?
21:30 Mahsa
Is it possible that you copy/paste all you have under <network-listeners> </network-listeners?
21:31 pameyer
shouldn't be a problem - let me double-check what's in there first
21:31 Mahsa
thanx
21:33 pameyer
no passwords / credentials in there :)
21:33 pameyer
<network-listeners>
21:33 pameyer
<network-listener protocol="http-listener-1" port="8080" name="http-listener-1" thread-pool="http-thread-pool" transport="tcp"></network-listener>
21:33 pameyer
<network-listener protocol="http-listener-2" port="8181" name="http-listener-2" thread-pool="http-thread-pool" transport="tcp"></network-listener>
21:33 pameyer
<network-listener protocol="admin-listener" port="4848" name="admin-listener" thread-pool="admin-thread-pool" transport="tcp"></network-listener>
21:33 pameyer
<network-listener jk-enabled="true" protocol="http-listener-1" port="8009" name="jk-connector" thread-pool="http-thread-pool" transport="tcp"></network-listener>
21:33 pameyer
</network-listeners>
21:34 Mahsa
Mine is the same,
21:34 Mahsa
so I am back to step 1,
21:35 Mahsa
dataverse is working on 8080 now , but Apache fronting is not working
21:35 Mahsa
Thank you Pameyer
21:35 Mahsa
I will have to look somewhere else now
21:36 Mahsa
:q!
21:36 pameyer
you're welcome - good luck
21:36 Mahsa
Thanks
21:36 pameyer
btw - I'd check that mod_ajp is loaded, and that the ports match; but I usually haven't had much trouble with it
21:37 Mahsa
Sure, I will start there
21:37 Mahsa
One quick Q
21:37 Mahsa
I am running glassfish as root,
21:37 Mahsa
that should not be causing any problems
21:37 Mahsa
right ?
21:37 Mahsa
with regards to Apache fronting
21:38 pameyer
it should not be be. for development, I run glassfish as root with apache in front of it with no problems
21:38 Mahsa
Ok, that is good to know, Thanks you again
21:38 Mahsa
What do you mean by ports match ?
21:38 Mahsa
I know mod_ajp is loaded
21:38 pameyer
it may be more effort than would be helpful, but if you check the `conf/docker-aio` folder of the dataverse github the steps there should get to a working configuration of apache+glassfish
21:39 Mahsa
Great! Thanks! I will check it out
21:39 pameyer
ports matching meaning that the ProxyPass / ajp://localhost:$port/ matches the port for the ajp listener in glassfish
21:40 Mahsa
I see , I will check it. Thanks
21:40 pameyer
you're welcome - good luck
21:40 Mahsa left #dataverse
22:17 pameyer joined #dataverse
