random :erpc, :timeout with partitioned get

[benonymus]

Hey,

I have the following cache:

  use Nebulex.Cache,
    otp_app: :app,
    adapter: Nebulex.Adapters.Partitioned

I randomly saw this error:

Nebulex.RPCError The RPC operation failed on node :"app@fdaa:0:0000:a9b:80:fdac:28d3:2" with reason:

{:erpc, :timeout}

See :erpc.call/5 for more information about the error reasons.

from here (lib/nebulex/rpc.ex in Nebulex.RPC.rpc_call/6 at line 138):

nebulex/lib/nebulex/rpc.ex

Line 138 in b12bd5a

:erpc.call(node, mod, fun, args, timeout)

Does this mean I should raise the timeout? https://hexdocs.pm/nebulex/getting-started.html#timeout-option

I also saw one of these:

Uncaught exit - {:timeout, {Task.Supervised, :stream, [5000]}}

the same time from another function that uses get from nebulex.

It was a one off not around deployment or any scaling.

[cabol]

Hey!

Does this mean I should raise the timeout? https://hexdocs.pm/nebulex/getting-started.html#timeout-option

By default the timeout is 5000 (5 sec), so maybe you should use the :timeout option, but the error may happen anyway because the cache is trying to reach a node to perform the operation and timed out, that may still happen. So I think if this is something you see from time to time, better off handling/rescuing the exception, then, if the reason is {:erpc, :timeout} perhaps you may want to retry, or just ignore it.

I also saw one of these:
Uncaught exit - {:timeout, {Task.Supervised, :stream, [5000]}}

This one looks a bit weird, because :erpc is used for OTP >= 23, otherwise, it uses supervised tasks. May you add the whole stacktrace please?

[benonymus]

Have not seen this since, so closing it.