Anyone ever get an error trying to install dbt wit...
# troubleshooting
a
Anyone ever get an error trying to install dbt with ``meltano add transformer dbt` ?
t
Can you share the command line output? Also try running with
meltano --log-level=debug add …
a
Yep. I have the log handy here: https://gist.github.com/andrewcstewart/974bb94f70f9459c64666ee864bbdd10 (will re-run with debug)
t
Transformer 'dbt' already exists in your Meltano project
seems curious. Is it already installed?
a
no but it’s repeated attempts after failed attempts
so I imagine the first attempt left behind an incomplete installation
t
hmm. I’m not sure how to clean up a failed installation. I’ll have to call for backup on this one. @douwe_maan or @aaronsteers?
a
I can try to see if the original failure is still in my scroll buffer
Alas! It has left us.
I think the “dbt already exists” is probably a bit of an aside. This mostly looks like an issue with
PyICU
which itself seems to be an issue with
agate
dbt just recently addressed this in a release a few days ago: https://github.com/fishtown-analytics/dbt/releases/tag/v0.18.2
d
"dbt already exists" is not an issue, it just means that the plugin is already in
meltano.yml
, so it doesn't need to be added anymore, and
meltano add
is effectively saying "it's already been added so I' not going to add it again, but I'll reinstall it for you"
The ICU error output is the same thing you would've gotten when you ran
meltano add transfromer dbt
initially, or if you were to run
meltano install transformer dbt
again now
I'd continue debugging the ICU issue, which is not Meltano specific. @andrew_stewart Have any of the workarounds in those issues you found worked?
a
Agree wrt “dbt already exists”.
What im going to do is test some separate venvs w/ just dbt vs dbt-via-meltano.
What version dbt is meltano currently using?
We should bump that to the latest
t
Yeah, 0.19.1 is out. Is there anything we’d have to consider when upgrading dbt like that?
a
The fact that https://github.com/fishtown-analytics/dbt/releases/tag/v0.18.2 was just released a few days ago and specifically addresses a know issue with ‘agate’, which relates to ‘PyICU’, I suspect is a promising lead here.
d
Indeed. You can update the pip_url in your meltano.yml and rerun
meltano install
a
(Or at least the tag was updated or something?)
Ill give that a shot.
Copy code
transformers:
  - name: dbt
    # pip_url: dbt==0.16.1
    pip_url: dbt==0.18.2
bingo
d
That worked? 🙂 Any chance you can create an MR to
discovery.yml
make that the new default for everyone?
a
looks like I do the following? 1. create issue to link MR back to 2. fork repo? 3. use web IDE to edit
discovery.yml
4. commit + create MR Did I miss anything?
d
You've got it!
a
d
Thanks @andrew_stewart!
a
np! Hope I did it right!
d
@andrew_stewart Can you add a changelog item as well?
a
can I add that manually w/o poetry ?
d
Totally, just edit changelog.md
t
any reason not to just upgrade to 0.19.1?
d
Not that I can think of
a
it’s a friday, I didn’t want to get too crazy.
t
hahaha
a
alright, stand by. Ill test 0.19.1
t
I'm down for the 0.18.2 upgrade and then I can do 19.1 next week if that's easier
a
This seems like an item we’d want to make easy to override locally also. Do we have docs on how one might do that? (I’ll create an issue if not.)
d
@aaronsteers The local pip_url in meltano.yml always takes precedence, and there’s a doc on “pinning a version” already
a
hmm, it can’t find distros for 0.19.0
im guessing 0.18 is the latest stable ?
anyway, for changelog… do I just add a new
## Unreleased
? (there’s already an entry under that header)
orrrrr.. do I add a single line under the current
## Unreleased
?
a
I think in the unreleased section under “Changes”, something like line 16 in the main branch: https://gitlab.com/meltano/meltano/-/blob/master/CHANGELOG.md#L16
a
Got it. I figured but wanted to confirm.
a
Thanks for confirming. I’m still getting used to the release workflow myself.