-
-
Notifications
You must be signed in to change notification settings - Fork 2.1k
DNS failures connecting to IP address (!) for replication #9865
Comments
Ah, I think what probably happened was that the request was cancelled after 60 seconds. Which leads us to... why did the DNS lookup take 60 seconds? I suspect the answer to this is that we got a bunch of other DNS lookups stacked up, to the point that the |
I think we should just rip out |
I'm not actually sure that it would. We'd still be subject to the concurrency limits of |
So I suspect the real solution is to fix https://twistedmatrix.com/trac/ticket/10181, and/or to install a |
We should be sure to double check this doesn't skip the IP address blacklists if we do this...although that's why we have the |
Do we think #10190 fixed this? |
I don't really think so. The bottleneck is still the GAIResolver, which is limited to 10 concurrent lookups. |
We had a slew of this sort of error:
There are two things of concern here:
HostnameEndpoint
in twisted, which is used fairly widely, including the HTTP Agent. (Raised as https://twistedmatrix.com/trac/ticket/10181)The text was updated successfully, but these errors were encountered: