#theforeman

ehelms I see pulp-manage-db, earlier there was a lot of mongo. I don't see any celery at the moment.

From what I remember it's pulp-manage-db+mongo->pulp-manage-db->pulp-manage-db+mongo ;)

I straced that pid and it looks like it is doing a stat on the pulp symlinks

While Katello/Foreman is down, does anyone know how to get the ENC cached yaml files to work? They're present on my servers, they just don't get used when node.rb can't talk to the server.

howdy folks. question about the APIs. When creating a host I get the following: {u'error': {u'message': u'undefined method `each\' for "ptable_id":String'}} but I'm sure that ptable_id is an int (I'm creating the dictionary in python and using requests to interact with foreman's api). Am I doing something obviously wrong?

"each" implies it's expecting an arry

what's the actual api call you're making?

gkoch: hmm, i thought that worked. gimme 5 to finish thi s up and i'll go re-read the enc script

gwmngilfen thanks. I looked at the script and it appears that it should read the cache, but it just complains `Error retrieving node host.example.com: Net::HTTPServiceUnavailable Check Foreman's /var/log/foreman/production.log for more information.`

gwmngilfen: I got one step ahead, now I'm actually converting the dict to json and added the content-type headers

now a nastier one: {u'error': {u'message': u'Environment(#55656080) expected, got String(#3587540)'}}

gwmngilfen I see the cached yaml files in /var/lib/puppet/yaml/foreman. All owned by puppet:puppet.

elisiano: again, thats a type mismatch, so i'd ask to see the call/parameters again :)

shoot, spotted it (I didn't convert environment to environment_id). Thanks gwmngilfen ! :D

happy to be the dog in the corner ;)

is anyone running a stand alone candlepin server?

I don't think I know anyone that masochistic

gkoch: just to make sure we debug the same thing, are you on the latest version of https://github.com/theforeman/puppet-foreman/bl... ?

gwmngilfen Yes, that's the file - from the v5.2.0 of the foreman module on puppet forge. Same md5sum.

sweet, having a dig now

gkoch: ah, you might be on to something here

if i'm reading this right, the enc call is properly rescued with a timeout reading the cache - https://github.com/theforeman/puppet-foreman/bl...

but

the prior call to upload facts - https://github.com/theforeman/puppet-foreman/bl... - is not wrapped

and that will also need to hit foreman to upload

try commenting out the upload section (323-338) and try it then

I now get /etc/puppet/node.rb:362: syntax error, unexpected keyword_rescue, expecting keyword_end rescue => e

oh whoops

Commented out 323-338

ugh, i cannot read

just comment 335-338

Back to the original error now

hmm

paste the error again

Error retrieving node da1haln01pe.stsky.biz: Net::HTTPServiceUnavailable Check Foreman's /var/log/foreman/production.log for more information.

aha

its not a timeout error code, gotcha

so 347 only rescues certain error types

you could try adding Net::HTTPServiceUnavailable to that list

So now I have lines 335-338 commented out and I've added Net::HTTPServiceUnavailable to line 347 and I still get the Net::HTTPServiceUnavailable error

intriguiging

the message comes from line 199, by the way, i'd have expected that raise to be caught on line 347

oh wait, i see

ugh, thats ugly

got it

gkoch: not the final solution, but for testing, add RuntimeError to the list on 347

That did the trick

so we have two problems

I uncommented 335-338 and it still works.

acidrainfall: :( I'm begining to think the same

1) upload_facts will bomb before we ever get to the caching - be good if you could uncomment that and retest though

ah ok

jgarr: What are you trying to do anyway?

so the issue is that rather than atimeout, your apache is still up and returning an error

thats fixable, could you submit a bug?

And I removed the Net::HTTPServiceUnavailable and it still works

The RuntimeError added on 347 was all that was needed

acidrainfall: run candlepin in a container. external db

yeah, i figured it would, line 199 obscures the error in a nasty way

acidrainfall: there's some examples but they run postgres in the container and don't save pgdata. It's just for testing releases as far as I can tell https://github.com/candlepin/candlepin/tree/mas...

gkoch: a bug to http://projects.theforeman.org/projects/puppet-... would be awesome, if you're willing

sorry, they don't run pg in the container. they link them together but I don't want pg in a container. I already have db servers

gwmngilfen always happy to file a bug - you all help us out a lot as it is. http://projects.theforeman.org/issues/15539

gkoch++

gwmngilfen now this is interesting, I've changed the file on both my puppet masters, and for some reason when I run puppet, it still complains that node.rb returns 1. When I run it myself, it works just fine.

are you running that as puppet or root?

jgarr: Ah. Way out of my league. Sorry :(

Bye all!

o/

gwmngilfen good call - I'm running it as root

Seems to work at puppet also though

hmm

is the axit status actually 0?

*exit

No, it's nothing

as in, echo $? == 0

echo $? is just a blank line.

that'll only work if its the very next command

even hitting enter will clear $? iirc

yeah, I did it as the very next command

I just tested it several times, it only returns 0 sometimes.

v odd

i'd at least expect consistency

When I run puppet it is saying it returned 1 though. Not exactly sure what else is different about puppet running the script vs root

other than the obvious of the different user

there isn't much

the script tries to switch user to puppet anyway, we just ask to make sure

i'm not really sure how to debug an inconsistent exit status

That makes 2 of us. I just ran it like 10 times as puppet and got 0 each time.

i can only suggest very basic thing like logging debug statements to a file at various points in the enc to see how far it's getting

More than happy to try that if you have suggestions of where to stick debug log lines

so the line you'd want is

File.open('/tmp/debugenc', 'w') {|f| f.write('here') }

FWIW httpd is not running on foreman - we're waiting for our ridiculously large pulp upgrade to finish.

yeah i saw the discussion with ehelms earlier

you can start from line 362 which is where the exit status is set to 1 if there's an error

put it inside the rescue and see if it logs

Nothing logged

so, not hitting that

I moved the log line to 199, which we know we were hitting before, and still got nothing

chown puppet:puppet for the log file too

interesting, thats called from 344

move it to 314

thats pretty much before anything happens

So weird. When I run it manually as root it logs

at 314?

Yep

And when I run it manually as puppet it does too

but not running from the master?

waiiiiiiiiit

the master might be caching it

can you bounce the service?

I restarted apache (we use passenger)

still nothing?