Time
S
Nick
Message
06:45 dataverse-user joined #dataverse
06:47 jri joined #dataverse
06:49 dataverse-user
alow
07:34 pdurbin_m joined #dataverse
07:37 juancorr
Hi @pdurbin, sorry I had not seen your yesterday messages
07:40 pdurbin_m
juancorr: no worries, I just popped in to see if the person who wrote "alow" above is still here.
07:57 pdurbin_m
juancorr: thanks for replying at https://groups.google.com/forum/#!msg/dataverse-community/OALTzINxkX0/v_WwJ4cvAwAJ
08:24 jri joined #dataverse
08:27 jri joined #dataverse
10:19 jri joined #dataverse
11:53 donsizemore joined #dataverse
11:55 donsizemore
@pdurbin i know you all like pull requests, but would you want a problem description without a pull request?
12:41 pdurbin_m joined #dataverse
12:41 pdurbin_m
what's the problem?
12:46 donsizemore joined #dataverse
13:01 pdurbin
donsizemore: bug reports are great. What's the issue?
13:44 donsizemore
@pdurbin i rendered our dataverse inoperable this morning using only a web browser. at a conference for two days but going back and forth with akio over hopeful recommendations.
14:12 pameyer joined #dataverse
14:54 donsizemore joined #dataverse
15:03 pdurbin
donsizemore: yikes! It seems like it's back up now.
15:03 pdurbin
This is the kind of stuff that makes pameyer paranoid. ;)
15:04 pameyer
pdurbin: that implies there's stuff that *doesn't* make me paranoid
15:05 pameyer
but I've been able to kill dataverse from a browser too - so I'm not too shocked
15:05 pdurbin
sadness :(
15:05 pameyer
… stress testing
15:17 pdurbin
At some point we should work on this performance test suite issue: https://github.com/IQSS/dataverse/issues/4201
15:26 pameyer joined #dataverse
15:38 donsizemore
@pameyer @pdurbin I see what I did, and it would be fairly unlikely for a random pedestrian to do it, but... all I had to do to get us back was kill R
15:40 pdurbin
Rserve?
15:41 donsizemore
I pulled a list of id and filesize from the datafile table, by filesize descending. i was checking out our largest datasets for use in another project in sequence, and after I browsed 5 filepages Rserve was no longer responding. kill the children and we're back and zippy. I want to handle that condition more gracefully
15:42 donsizemore
^^ by no longer responding, i mean to additional requests
15:43 pdurbin
Browsing the file page triggers a call to Rserve?
15:45 donsizemore
i wondered about that. i didn't do anything else which should've triggered R. could there have been cached copy not created by an earlier ingest workflow or something?
15:48 pdurbin
I don't know. The behavior you're describing is surprising but I don't know exactly how that page works.
15:49 donsizemore
well, i'm not worried about it because it was definitely a "me" problem but nice to track down which thresholds i need to bump. R is the obvious one but Dataverse couldn't serve its homepage either
15:50 pdurbin
donsizemore: if you can characterize the bug a bit more, please feel free to open a GitHub issue
15:50 pdurbin
or even if you can't :)
15:50 donsizemore
@pdurbin i started to, but wanted to exhaust our local repertoire before whining
15:51 pdurbin
heh. ok
17:29 pameyer joined #dataverse
17:46 donsizemore joined #dataverse
18:49 pameyer joined #dataverse
19:12 jri joined #dataverse
19:17 pdurbin
Dataverse 4.9 is out: https://github.com/IQSS/dataverse/releases/tag/v4.9
19:23 donsizemore
@pdurbin i'm upgrading a test box =)
19:26 pameyer joined #dataverse
19:36 pdurbin
!
19:36 pdurbin
the fastest draw in the east
19:39 donsizemore joined #dataverse
19:39 donsizemore
@pdurbin pew pew
19:41 pdurbin
heh
19:42 pdurbin
I saw "waiting for next dataverse release" on this Solr issue so I left a link to 4.9: https://github.com/IQSS/dataverse-ansible/issues/17 :)
19:45 donsizemore
@pdurbin i'm not even done reindexing my test box yet!
20:18 pdurbin
:)
20:28 donsizemore
p.s. my test upgrade was squeaky-clean so far, i'm doing the file PID bit now
20:29 pameyer
how many files are in your test box?
20:47 donsizemore
65306 (counting harvesting)
20:48 pameyer
I'd been thinking about traffic to the PID provider for giving a big chunk of files PIDs at the same time
20:48 donsizemore
local: very few. and now that i click on the local datasets i get an internal server error
20:49 pameyer
pretty sure that's not supposed to happen
20:49 donsizemore
[2018-06-06T16:48:31.567-0400] [glassfish 4.1] [WARNING] [] [javax.enterprise.web] [tid: _ThreadID=47 _ThreadName=jk-connector(1)] [timeMillis: 1528318111567] [levelValue: 900] [[ StandardWrapperValve[Faces Servlet]: Servlet.service() for servlet Faces Servlet threw exception org.postgresql.util.PSQLException: ERROR: column "reason" does not exist Position: 18
20:49 donsizemore
but i ran the upgrade script
20:50 donsizemore
oh, crap, no i didn't.
20:51 donsizemore
oh, wait. yes i did.
20:51 pameyer
I don't see "reason" in the upgrade scripts
20:52 pameyer
but I think that sometimes java/jsf/bean/other guts can sometimes add stuff to db as needed
20:52 donsizemore
well, i ran the upgrade script twice. checking out those regexs to see what i just broke =)
20:53 donsizemore
going to nuke the index one more time, then try the file PID step again
20:54 donsizemore
the conference is wrapping up for beers anyway. i'll check on this again tomorrow morning
20:54 pameyer
ah - something that just occured to me. is this the test box that you've deployed the patched war to?
20:54 pameyer
because that might be the source of the problem
20:55 donsizemore
no, we're running the patched war in production =)
20:55 pameyer
another possible idea thwarted by reality ;) enjoy the rest of the conference
20:56 pdurbin
There's a "reason" column on the datasetlock table: http://phoenix.dataverse.org/schemaspy/latest/tables/datasetlock.html
20:57 donsizemore
i appreciate the ideas. i did the upgrade during a break when 5 or 6 people were all talking around the table. i still blame me.
20:57 pameyer
test boxes are *for* breaking
20:57 pdurbin
:)
20:58 donsizemore
i think tomorrow i'll start over with a dump of the production db
23:49 jri joined #dataverse
