> Are you using a memory database or a file-backed database? 

memory.

> If a memory database, the error appears to be it creates a new 
connection...

Ok, you're saying that a new connection is created for every thread, and 
that means, an empty memory database? Even when I ran the migrations in the 
main thread's connection? That seems a bit odd.

What'd be the best way to workaround that? I've tried changing to a 
file-backed database, but now I'm seeing lock contention issues on the file 
(see: https://gitlab.com/honeyryderchuck/rodauth-oauth/-/jobs/710170917). 

> Not sure if this is an issue with sequel-active_record_adapte...

Do you mean sequel-activecord_connection, or jdbc-sqlite3, or 
activerecord-jdbc-adapter? I could try to narrow this down to a small 
purpose script to see if I can pinpoint the dependency. I was hoping you 
had seen this before and could suggest a workaround, as that might take me 
some time. 
A sábado, 29 de agosto de 2020 à(s) 17:18:39 UTC+1, Jeremy Evans escreveu:

> On Saturday, August 29, 2020 at 7:27:28 AM UTC-7, Tiago Cardoso wrote:
>>
>>
>> Hi,
>>
>> I'm currently trying to enable parallel tests in "rodauth-oauth" across 
>> all tested databases, and I'm having issues particularly with running the 
>> tests on jruby against a sqlite database.
>>
>> An example of the problem I'm finding can be seen here: 
>> https://gitlab.com/honeyryderchuck/rodauth-oauth/-/jobs/709664165 (debug 
>> log was enabled to better have a look at what is going on).
>>
>> Of note, I can share the interesting parts of the configuration:
>>
>> ---------------
>>
>>
>> https://gitlab.com/honeyryderchuck/rodauth-oauth/-/blob/master/test/support/roda_integration.rb#L38-41
>>
>> I initialize the db in the beginning of the config, and run the 
>> migrations immediately.
>>
>> ---------------
>>
>>
>> https://gitlab.com/honeyryderchuck/rodauth-oauth/-/blob/issue-26/test/support/roda_integration.rb#L141-147
>>
>> This is where a wrap the test around an auto-rollbackable transaction, 
>> the same way "database-cleaner" or "rodauth" itself sets it up in its 
>> tests. This is also where the bug happens, when I insert the account. Of 
>> importance, the error only happens when running the tests in parallel-mode 
>> (when PARALLEL=1).
>>
>>
>> I know this is not a small purpose script reproducing the issue, but I'm 
>> banking on someone having seen and reported that before, and maybe giving 
>> me hints on how to work around this. This can also be an issue with the 
>> "jdbc-sqlite-adapter" gem, or with sqlite itself.
>>
>> If someone has some hint on this, or can give me tips on how to best 
>> debug this, that'd be very helpful.
>>
>
> Are you using a memory database or a file-backed database?  If a memory 
> database, the error appears to be it creates a new connection, which uses a 
> new (empty) memory database.  Not sure if this is an issue with 
> sequel-active_record_adapter or something else.  Sequel defaults to only 
> allowing a single connection if it thinks you are using a memory database.
>
> Thanks,
> Jeremy
>

-- 
You received this message because you are subscribed to the Google Groups 
"sequel-talk" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/sequel-talk/c2d89f51-518c-456c-b443-86049cf4d71an%40googlegroups.com.

Reply via email to