Fork me on GitHub
#datomic
<
2019-08-27
>
Jakub Holý (HolyJak)11:08:33

Any idea if there are any plans to make HTTP Direct available also for Solo topology?

henrik11:08:51

That would necessitate including an NLB with the Solo topology, which would increase the cost of running it. Personally, I think it might be worth it, since it would make Solo more accurate for prototyping things that would ultimately run on Prod.

kenny15:08:18

Though we don't use Ions anymore, I do agree that having the option to include a NLB for more accurate prototyping is quite valuable.

👍 4
Jakub Holý (HolyJak)18:08:42

I wish there was also a cheaper (and necessarily less performant) version of the prod topology...

souenzzo13:08:20

anyone running datomic on-prem inside a ions setup? (I'm worried about deps conflicts)

marshall13:08:29

The peer library will not run in an ion

souenzzo13:08:57

I can't setup my own transactor on AWS and connect on it from inside a ion?

marshall13:08:42

Not at this time

😞 4
eoliphant14:08:10

Hi, i just tried my first ‘from scratch’ install of the separate DC storage and compute stacks

eoliphant14:08:13

I’m getting an ApplicationDoesNotExistException when I try to push. I’ve poked around a bit, and I do see the code deploy app, etc

marshall14:08:57

look at your newly created compute stack in cloudformation dashboard go to outputs and find the name of the codeDeployApplication

marshall14:08:23

make sure that’s the same as what you’re using also check that your AWS creds (and region) you’re using to push are correct for that account

eoliphant14:08:06

dammit lol. had already checked all of those things. I’m running with God creds.. but sorta forgot to set AWS_PROFILE 😉

eoliphant15:08:32

Ok running into another (I think actual) issue. the log is full of streams that appear to be the result of instances cycling due to this msg “:datomic.cluster-node/-main failed: Production Compute Stack not found for query group”. I’ve only installed storage and compute

marshall15:08:32

when you ran the compute stack, are you sure you used the compute template and not the query group template by accident?

eoliphant15:08:56

I’ll retry it but 99.9% certain that I didn’t Is there anything unique in the QG outputs, that I could use as a quick sanity check?

eoliphant16:08:54

well actually, just reviewing the parameters. None of the QG stuff like “max query group instances” is there

marshall16:08:20

what version ar eyou launching?

eoliphant16:08:15

latest 482-8794

marshall16:08:50

and the log errors are showing up in the datomic stream for this system?

eoliphant16:08:01

and from the template

{
	"AWSTemplateFormatVersion": "2010-09-09",
	"Description": "Creates compute resources needed to run Datomic.",
So looks right

eoliphant16:08:09

i tried a deploy

eoliphant16:08:31

and got that ‘failed because too many instances failed deployment’ msg

eoliphant16:08:34

then started digging

eoliphant16:08:20

and the ASG activity history is full of Terminating/Launching messages

marshall16:08:58

from the failed deploy is the system up and running normally prior to doing a deploy?

eoliphant16:08:36

yeah, this is the one I’ve been working on. THe ‘cycling’ doens’t appear to ahve started until I tried the deploy

marshall16:08:19

what is the application name in your ion-config and what is the deploy command you’re running

eoliphant16:08:23

and I was just using ion-started to test the deployment

eoliphant16:08:41

:app-name "ps-cp"}

eoliphant16:08:46

same as the stack

marshall16:08:56

the compute stack?

eoliphant16:08:57

and didn’t override the app name

marshall16:08:12

that’s the issue i think

eoliphant16:08:13

sorry the “system name”

marshall16:08:20

look at the outputs of your compute group stack

eoliphant16:08:32

i have ps-cp and ps-cp-compute

eoliphant16:08:57

SystemName ?

marshall16:08:58

so your CodeDeployApplicationName needs to match

marshall16:08:19

and then when you run the deploy what’s the command? the group needs to match the CodeDeployDeploymentGroup

eoliphant16:08:58

ah.. hmm. i was just using the output of :push

eoliphant16:08:18

but it’s setting the :group to the name of the compute stack

marshall16:08:39

which should match the CodeDeployDeploymentGroup

eoliphant16:08:50

ah yes it does

marshall16:08:59

hrm. guess that’s not the issue

eoliphant16:08:21

CodeDeployApplicationName	ps-cp
CodeDeployDeploymentGroup	ps-cp-compute
SystemName	ps-cp

eoliphant16:08:28

from the outputs

eoliphant16:08:16

Over in CodeDeploy I have a ps-cp app with a ps-cp-compute deployment group

eoliphant16:08:53

and the only thing that’s different (we’ve dozens of solo and prod sized stacks) in the large is that I did the separate stacks from the outset vs the marketplace->first upgrade, etc

marshall16:08:17

that definitely shouldn’t matter

eoliphant16:08:34

yeah I didn’t think it would

eoliphant16:08:09

and theres not much in the logs. each stream is the same 4 lines of what looks like normal startup housekeeping, then the error

marshall16:08:54

can you paste the full error line?

marshall16:08:00

it’s definitely trying to launch a query group

marshall16:08:47

notice: "IndexGroup?": false,

marshall16:08:21

@eoliphant can you file a ticket (<mailto:[email protected]|[email protected]>) and we’ll look further into it

marshall16:08:59

@eoliphant can you try deleting the compute stack and re-creating it?

eoliphant16:08:57

hey sorry stepped away for a sec will retry

marshall16:08:37

@eoliphant also, can you provide the first lines of log from when you originally created the stack (when it succeeded originally)?

marshall16:08:45

in particular, the one with the configuration information in it

eoliphant16:08:03

let me find it

eoliphant16:08:18

there’s quite a few of them now lol

eoliphant16:08:56

well this is peculiar, the oldest one I can find is showing the same thing. ugh.. in any case deleting the compute now

eoliphant16:08:14

will delete all the streams, etc as well

eoliphant17:08:48

ok it’s taking forever, I’d added a VPC endpoint, and forgot to remove it before I started, I thought I caught it before it got there, but I suspect we’re in CF’s 60 min timeout

marshall17:08:51

@eoliphant do you have a lot of ASGs in your account (like more than 50)?

eoliphant17:08:54

and it’s back up and went right to blowing up

eoliphant17:08:43

some sort of AWS API paging issue?

marshall17:08:17

i believe we have a bug when you have > 50 ASGs in an account

marshall17:08:26

we’ll work on a fix for the next release

eoliphant17:08:28

what do I get? 🙂

sharkdance 4
eoliphant17:08:35

ok cool, thx

eoliphant17:08:22

I’m going to see if i can do some cleanup

eoliphant17:08:29

in the meantime