#hypothes.is

<nick> ARGH NPM!!!!

<chdorner> morning :wave:

<chdorner> @nick shall we?

<nick> morning!

<nick> certainly :slightly_smiling_face:

<nick> just waiting for el Jenkins

<chdorner> I guess we haven't deployed in a few days. do you have a list of things we need to check?

<nick> https://github.com/hypothesis/h/compare/985e91c...

<nick> not that much, except for the bit where I moved all the route definitions :,)

<nick> and your document_id migration has already been run, right?

<chdorner> I think Sean already did that, I'll check

<chdorner> the one that adds the column, right?

<chdorner> yup

<nick> yep

<nick> ok, cool

<nick> so it looks to me like we should check

<nick> - API exception views (just check that we get the new 404 message)

<nick> - Create user endpoint

<nick> - And the session stuff we're deploying -- i.e. ensure we don't somehow get logged out

<nick> In the meantime do you want to just check over #3900 and rebase it on master?

<chdorner> yup, will prepare 3900

<nick> Oh, and #3940 should be fairly uncontroversial and it would be nice if it could go out next deploy

<chdorner> let's get 3940 out with 3900? which should happen in about an hour anyway :)

<nick> yep

<nick> ok, going to stage

<nick> in unrelated news, there's a BIND9 CVE fix today

Action: nick is looking forward to when we can automate reboots

<chdorner> do we need to reboot for that?

<nick> oh, you're right

<nick> we don't

<nick> yay!

<chdorner> API 404 looks good, I'm not logged out. do you want to check the create user endpoint?

<nick> yep

<nick> looks fine

<chdorner> uhm, prod new relic isn't happy

<chdorner> grafana as well

<chdorner> looks like request queueing is through the roof

<nick> checking

<nick> hmm

<nick> I think that was just a load spike, perhaps a result of the deployment

<nick> overall it looks fine

<nick> to prod?

<chdorner> yup

<chdorner> No problems NPM, we have time, whenever you're ready. Travis is already retrying about 10 times on network errors!

<nick> fwiw the CPU usage spikes on the app machines over the last hour appears to be unattended-upgrades running

<nick> I think it got upset because we managed to run a deploy at the same time as u-u was running on app-3

<chdorner> ah

<nick> prod looks good

<nick> Casey just closed my laptop

<nick> thanks, Case

<nick> ok, just going to review bye-redis-pt-2 again

<nick> @chdorner I reckon we should take 3901 out at the same time

<nick> do you want to rebase it?

<chdorner> done

<nick> jesus this NPM situation is really bad

<nick> hmm, although actually that's not why https://jenkins.hypothes.is/job/h/job/master/59... failed

<nick> it looks like something went wrong with the containerised elasticsearch

<nick> or perhaps there's a race condition in the test setup code

<nick> I'm just waiting for Jenkins...

<nick> :timer_clock:

<nick> NPM :(

<nick> oh

<nick> not npm again

<nick> haha

<chdorner> nope

<chdorner> me :slightly_smiling_face:

<nick> too quick to blame

<nick> blame @chdorner instead

<nick> :bell: blame

<nick> whoah, what happened to prod

<nick> suuuuper unhappy all of a sudden

<nick> looks like we're being hammered by one IP

<nick> details in sr

<chdorner> My internet connection is acting up, I'll try and restart the router

<chdorner> alright, back

<nick> ok, we've blocked our rogue badge requester, so going to deploy 20160928-g86f1bc5

<chdorner> looking pretty good

<nick> looks good to me

<nick> to prod!

<nick> (after ansible)

Action: nick taps fingers impatiently

<chdorner> good bye redis :tada: :tada: :tada: :cake:

<nick> woop :fireworks:

<chdorner> do you want to look over https://github.com/hypothesis/h/pull/3930 again?

<nick> no, if Sean's happy with it that's fine

<nick> and sorry for the misleading information about `.in_`

<nick> s/misleading/incorrect/

<chdorner> Sean is off until next week, so feel free to merge it :slightly_smiling_face:

<chdorner> did you test the migration locally though?