I’d be curious to see the debug log sql processing times to see which is actually the slow one first of all?
OvermindDL1
l00ker
This suggestion might be going out on a limb… but have a look at this recent post:
Disable prepared statements for an individual query
I didn’t bother to research when prepared statements where introduced in ecto etc., but since this was such a large upgrade jump, maybe you’re being affected in a similar way and it’s adding up fast due to the shear volume of queries.
It’s a long shot, but the read and eventual outcome might prove educational either way 
1
agacode
I put Repo.checkout in the controller action in my localhost, the pool size is 10
It is processing 10 request only at the same time, but it is getting stuck/hung.
I have run some commands that I’ve found on a closed issue for db_connection and the output is the following:
iex(7)> pid = spawn(fn -> Beacon.Repo.query("SELECT 1") end)
#PID<0.2153.0>
iex(8)> Process.info(pid)
[
current_function: {DBConnection.Holder, :checkout_call, 5},
initial_call: {:erlang, :apply, 2},
status: :waiting,
message_queue_len: 0,
links: [],
dictionary: [],
trap_exit: false,
error_handler: :error_handler,
priority: :normal,
group_leader: #PID<0.65.0>,
total_heap_size: 233,
heap_size: 233,
stack_size: 38,
reductions: 188,
garbage_collection: [
max_heap_size: %{error_logger: true, kill: true, size: 0},
min_bin_vheap_size: 46422,
min_heap_size: 233,
fullsweep_after: 65535,
minor_gcs: 0
],
suspending: []
]
iex(9)> Process.info(pid, :current_stacktrace)
{:current_stacktrace,
[
{DBConnection.Holder, :checkout_call, 5,
[file: 'lib/db_connection/holder.ex', line: 251]},
{DBConnection.Holder, :checkout, 2,
[file: 'lib/db_connection/holder.ex', line: 60]},
{DBConnection, :checkout, 3, [file: 'lib/db_connection.ex', line: 1030]},
{DBConnection, :run, 6, [file: 'lib/db_connection.ex', line: 1340]},
{DBConnection, :parsed_prepare_execute, 5,
[file: 'lib/db_connection.ex', line: 540]},
{DBConnection, :prepare_execute, 4,
[file: 'lib/db_connection.ex', line: 533]},
{Postgrex, :query_prepare_execute, 4, [file: 'lib/postgrex.ex', line: 219]}
]}
Any idea why might be getting hung? Thank you
0
benwilson512
Author of Craft GraphQL APIs in Elixir with Absinthe
What is getting hung, the 10 requests or the iex command? Are you running that spawn command while the 10 requests are happening? If you have a pool size of 10 and there are 10 requests happening each with a connection, your pid in console MUST hang until a connection becomes available.
0
agacode
The 10 requests are getting hung. Actually they all executed till the end I had a 200 (OK) response from each of them in JMeter. But it won’t execute any other request after that, no more logs in the console, no activity, nothing.
I ran those commands because I saw them in this post from @josevalim
I tried then replacing Repo.checkout by Repo.transaction (The queries can execute in a transaction) because I saw it in this article from @josevalim, but that did not solve the problem either, not getting hung with Repo.transaction, but the same behavior as before very slow.
I also tried the master branch for db_connection but I got the same behavior.
Thanks again for looking into this.
0
This. Give us some more details, there are practically zero at this point.