Fork me on GitHub
#datomic
<
2021-10-01
>
pedrorgirardi10:10:12

I’m upgrading Datomic Cloud for the first time, and I managed to ‘break’ it. :man-facepalming: I selected the storage nested stack, and followed the steps, but I’m getting: > The following resource(s) failed to update: [EnsureAdminPolicyLogGroup]. What does that mean exactly?

hden12:10:51

I think your data are fine. Try submit a support ticket. https://support.cognitect.com/hc/en-us/requests/new

jaret13:10:29

@U5GP9FMC0 I think its really important to note that you didn't "break it", what is happening here is the update CFT operation is failing, because a resource cannot be deleted by Datomic. It will roll back to the previous version. Any resource that has been modified will not be deleted by Datomic.

jaret13:10:46

and as @U0HLHE6JE says your data is totally safe and your system is still operational. I would be happy to walk you through upgrading or look at the specifics of your situation. If you do log a ticket please send me what steps you were following and what version you are upgrading from and to.

pedrorgirardi13:10:57

Thank you @U0HLHE6JE @U1QJACBUM. My wording was misleading @U1QJACBUM; the system is fine, and I didn’t loose anything. I will create a ticket and let you know. Thanks in advance.

Chris10:10:34

Can anyone recommend some open source codebases that use Datomic On-Prem? My team have been having discussions about different practices and what advice does and doesn't transfer from traditional DBs, and it would be nice to have some view of how other people approach it.

Twan12:10:42

We try to upgrade Datomic Client (on-prem) from 1.0.6202 to 1.0.6344. After upgrade, we are not able to perform queries on :db-after (as a result of a transaction) any more. We get errors either saying (`d/q`) Query db arg does not match connection or (`d/pull` )`db not found` . Do you have any clue?

Twan12:10:04

We downgraded to 1.0.6269 which resolves the issue, so something between 1.0.6269 and 1.0.6344 is likely the cause

jaret13:10:15

Hi @U9M6WJ9PV what version of client-pro are you using? Did you update your peer-server since you are using a peer-server to utilize client?

Twan13:10:10

Hi @U1QJACBUM! Our client-pro is on 1.0.72 The peer server was also on 1.0.6344 (and 1.0.6269 respectively)

jaret13:10:02

Can you share a full gist or snippet. I'd like to immediately try to re-create this 🙂

jaret13:10:36

A repl history would be enough for me to see if you are doing anything I am not doing in re-creating.

Twan13:10:56

(def res (d/transact conn {:tx-data [{:db/id 17592186190002 :nedap.source.people.person.wm/first-name "Some Name"}]})) (d/q '{:find [(pull ?e [*])] :where [[?e :nedap.source.bournedrasil.wm/key "something"]]} (:db-after res))

Twan13:10:14

I hope you get the gist of it

Joe Lane14:10:43

@U9M6WJ9PV What is the actual value of res? is it a map with the :db-after key or is it an anomaly?

Lennart Buit14:10:45

(I’m from the same company) Its just a successful transaction result, so with :db-after/`:db-before`/`:tempids`/`:t`. What we do notice btw is that the :database-id between (d/db conn) and (:db-after res) differs. The former being db-name + hex string (uuid?), the latter being a jdbc url.

Joe Lane14:10:36

Are y'all using on-prem with sql storage (which one?) and issuing these operations against a peer server via the client API?

Lennart Buit14:10:27

Yes, we use postgres, and we are issuing these operations against a peer with the client API

👍 2
Joe Lane14:10:10

Is it easy for your team to try with 1.0.6316?

Twan14:10:07

We already did. I copied the wrong version in my first thread message

Twan14:10:17

In that version everything was fine

Lennart Buit14:10:44

What I noted about :database-id being different between (d/db conn) and (:db-after res), that is not the case on 1.0.6316 and they both contain jdbc urls

Joe Lane14:10:03

Can you now try using the bin/repl command and the peer API to see if you get the same behavior?

Lennart Buit14:10:39

On 1.0.6344, or on 1.0.6316 ?

Lennart Buit15:10:39

Would you mind if we park this discussion until monday ^^. Its 5:02PM here, Friday afternoon, so not the best moment to start fiddling with the database 🙂. Have a good weekend!

jaret15:10:17

@UDF11HLKC @U9M6WJ9PV we have reproduced this issue. No further action needed on your end.

👀 3
jaret15:10:30

I am discussing with the team now and will update you.

2
👍 1
jaret17:10:48

As an update we isolated the problem and will work on a fix for an upcoming release. For now the work around is to downgrade ONLY the peer server process to 1.0.6316.

2
👍 1
jaret17:10:40

Thank you both for reporting this!

Twan18:10:43

Good to know, thank you for your follow-up

Lennart Buit19:10:43

Thanks for the openness & follow up :)!

Tatiana15:10:46

Hello, at https://docs.datomic.com/cloud/ions/ions-reference.html#web-ion, I noticed the request gets :uri. How do I get it right inside my http function? I am currently using metosin/reitit to create routing, can anyone have experience with using this inside datomics?

kenny22:10:16

Good afternoon. I'm executing an Ion deploy that is failing at the DownloadBundle stage. In the event log, the following message is present:

Cannot allocate memory - rm -rf /opt/codedeploy-agent/deployment-root/16ccad37-2dcd-49b4-81d5-65b944bab806/d-603RLTGJC 2>&1
I have not hit this error before, so I am curious what the best way to resolve it is. I can, of course, provide more info if relevant.

kenny22:10:39

fyi, terminating the instance and having the asg start a new one fixed it.

prnc14:10:32

I’ve seen this before, w/ the same workaround as yours, not sure what a “proper fix” is