#metabrainz

To verify what settings are in place, check the <JIRA application home directory>/logs/atlassian-jira.log or catalina.out file. A section in the startup appears like this: JVM Input Arguments : ....

ruaok: no "por favor". "María, te presento a José de Núñez y Navarro. José, te presento a María, del BBVA."

(sorry, was shopping)

reosarevok: thx

I increased max memory manually

2.5gb looks OK (unless it's more than memory available on the machine and it's swapping)

... Why are you CTCP'ing?

see ./bin/setenv.sh

3.5G avail with no swap.

no swapping happening. but, perhaps OOM conditions are happening and retries succeed? and therefore it takes a while?

<ruaok> why would a new one run slower than an old one? that makes no sense.

the extreme next step would be to break out jvisualvm or something and keep an eye on what is happening to the heap and when it's doing GC

I'm not sure if there's a non-graphical version of that. it's been a while since I java'd

logs are at /home/jira/jira/logs/catalina.out

> com.atlassian.jira.service.OfBizServiceConfigStore$ClassNotFoundServiceException: Could not find class: com.atlassian.jira.plugin.ext.subversion.revisions.RevisionIndexService

something is still expecting there to be a svn repo, which we no longer have.

that might be causing issues

where is that?

at the beginning or just happened?

happens periodically.

hmm, I installed the subversion plugin

you might be looking at old logs

nope. I just saw that come across 30 seconds ago.

at this point keep looking at the logs and investigate anything that looks out of place

which file?

sudo tail -f /home/jira/jira/logs/*

Potential new Jira banner logo thing: https://drive.google.com/file/d/0Bz_3cJoQhf4nYm...

chirlu gcilou: ^ any thoughts/objections?

I have an objection.

There's still more chocolate. Your objection is invalid.

why is the "t" of tracker not capitalized?

gcilou: https://chatlogs.metabrainz.org/brainzbot/metab...

"It's been a while since you last logged in. Due to data security requirements, we have deleted your account. If you wish to apply for courses and programmes, you must create a new account at Universityadmissions.se. Your username is your email address but please be aware that you cannot use the same email address you used with your earlier account; you must

use another in order to create your account."

oh. It looks funny to me

"we removed your account but won't allow you to reuse the email. good luck!"

reosarevok: ... -_-

do we actually need this subversion plugin?

It maybe should be titlecased in this case.

BTW, I see that darwin was scrobbling on Last.FM is the last few days.

in the last 72 hours?

chirlu gcilou: Title cased: https://drive.google.com/file/d/0Bz_3cJoQhf4nRX...

ruaok: Yes, e.g. 11 hours ago.

great. phew.

Freso: that's the same as before

/whois'ing darwin says "[darwin] idle: 07 hours 37 minutes 58 seconds, signon at: Thu, 01 Dec 2016 09:27:31"

I personally didn't know anyone, but have lots of friends who do.

So that would also suggest him being active on IRC somewhere 7½ hours ago.

gcilou: Same?

Oh.

yeah, lowercase tracker

Wtf. Google Drive.

G. Drive is giving me the same share link for both files.

#smart

:)

The text doesn’t look Black enough to me. Also, the trailing “r” is outside the box.

Is it Roboto Regular?

That's how the "Community" one is too

(trailing last letter)

Then it’s a bug there, too. :-)

Oh, I might have linked the wrong one.

Normally, what is outside the box shouldn’t even display.

chirlu: https://drive.google.com/file/d/0Bz_3cJoQhf4nTW... ?

that looks better to me

doesn't look like subversion errors show up anymore

nothing out of the ordinary in the logs

Man. New server hardware, new Jira... 2016 might not be *all* bad after all.

dictionary.com declared "xenophobia" to be the word of the year for 2016..

zas: ping

gcilou: ... I'm trying to raise the spirits a bit. -_-

Is the “Community” text gray instead of black, too? Can’t tell from the low-res PNG.

are we using this https://cloud.google.com/compute/ or something else?

https://community.metabrainz.org/uploads/defaul... looks like a dark grey?

As I said, “can’t tell from the low-res PNG”. Could be a dithering artefact.

#424242

Freso: sorrysorry

ruaok: pong

what tool did you use to do that traffic analysis of the slow mb web pages?

gentlecat could use the same tool for his new jira installation

Hm. "Site discussion" on forum is #1F1F1F.

for slow mb web pages bitmap hacked mb server code

you showed me a client side analysis.

was it chrome?

ah yes, dev tools

or, rather do you have any ideas why the new jira instance on google is so slow?

could also try to switch to more CPUs

what is the url ?

requirements say "multicore CPU", not sure how accurate that is

these are probably estimates anyway

I had a multicore machine recently and it was utterly underutilized.

what was the URL again Gentlecat ?

for what?

single cpu instances are very slow, but JIRA shouldn't not be cpu intensive

for new jira

requirements doc is at https://confluence.atlassian.com/adminjiraserve...

well, lets remove that from the list of possible causes.

shut down the instance, Gentlecat. let me know when that is done, ok?

seems like it shut down on it's own

502s atm

r/it's/its

new IP 104.155.150.101

7.5GB ram, 2 cores.

no way to move data?

it is a clone of the the one you've been working on.

just change dns if ip changed

hmm, I don't see jira stuff

ok, can't trust the clone function then. ya.

ruaok: i was looking at curves at http://stats.metabrainz.org/dashboard/db/mb-vs-... and we didn't find yet the reason of this huge load, i will rate limit the ws a bit more during peak hours to see if it impacts load, but still things remain unexplained, bitmap tried to log slow queries, i had a look at logs but well.. slow queries happen mostly because of

the load, so....

;_;

yesterday we had loads > 50 on all webservers, and it isn't directly related to traffic

I was amazed that they still responded to traffic so well at 50

they do, and upstream number of 200s isn't changing much

if you look at curves, we have load drops while traffic (bytes/s and number of requests/s) don't change much

this is totally abnormal

when bitmap will be around, i want us to add a ws instance on another machine to see if the load follows the same profile when no website instance runs on the same machine

website number of requests per second is very low

Gentlecat: new ip 104.155.150.101

ruaok: http://stats.metabrainz.org/dashboard/db/mb-vs-... is very interesting, compare load vs number of requests/s

chirlu: I linked an SVG with "Tracker" (instead of "tracker") before that you didn't comment on :)

seems a bit better

we had a sudden and very significant increase of the load, but the number of req/s is pretty the same

someone else please look

Freso: I did comment on that SVG.

That was the one where the border was fixed, and where I wondered about the colour.

chirlu: Right. So no comment on t vs. T for tracker. :)

"Community" is uppercase? why not make it consistent?

Gentlecat: "Ticket" is uppercased, so it would be consistent either way with the samples we have now.

ruaok: 16:48 to 16:52 shows a load1 bump from ~20 to ~50 while number of requests/s didn't change

"Community" is consistent with both "Ticket tracker" and "Ticket Tracker". But this one might determine whether we use title case going forward or not. :)

ruaok: Spend some Google Money on whips?

Gentlecat: when you ask people to look, you need to remind people of the URL.

what was the URL Gentlecat ?

alright, people, please look at http://jirabrainz.gentle.cat/

👀