Set TCP keepalive on Redshift (#782) #826
Conversation
|
@ryanmcdonnell If you're interested in testing out a branch that may fix your connectivity issues, you can give the |
| @@ -9,6 +9,9 @@ | |||
|
|
|||
|
|
|||
| class RedshiftAdapter(PostgresAdapter): | |||
|
|
|||
| DEFAULT_TCP_KEEPALIVE = 240 | |||
There was a problem hiding this comment.
just curious... where did this number come from?
There was a problem hiding this comment.
Drew linked me to this which has a docstring bit that says:
Note: For Redshift, use keepalives_idle in the extra connection parameters and set it to less than 300 seconds.
I wasn't sure how precise the 300 value was or where they sourced it from, so I figured one minute less would give us a bit more wiggle room on slightly flaky connections and stuff. Now that I think about it, they probably did the same thing with whatever their source was and 300 is also conservative, but hey, a few extra keepalives shouldn't hurt.
Set TCP keepalives on Redshift (and provide it as an option on Postgres). Users can disable setting the value if they want by explicitly setting the
keepalives_idleparameter in the connection profile to0, which will make it use the system default.This should fix #782