Björn Svensson

As stated before this breaks clients, one example is `hiredis` when running with the async API. In async, hiredis waits for a successful reply to know the servers state and...

No, I don't think there is a way to protect against this currently..

As I understand it on a Linux, if cluster node timeout is 15000, `TCP_KEEPIDLE` is set to 30 seconds. So after after 30 sec idling/silence the kernel will attempt to...

Any more details of the arguments in frame 3, 4 and 5?

I got to test it a bit as well and it seems like it wont show single-result searches when continuously pressing Ctrl+S or Ctrl+R. You get a search result after...

Nice. Its a bit annoying that clang-format had some formatting changes somewhere between 18.1.3 and 18.1.6. Maybe that the reason for the failing CI check? The check seem to use...

Unfortunately, there are some wordings in the CLA for librdkafka that blocks me from contributing with a fix, otherwise that would have been simpler.

Hi! This will hopefully answer your questions: 1. `redisClustervCommandToNode` first appends the command to hiredis output buffer using `redisvAppendCommand`, then it calls `redisGetReply` in hiredis which will send the command...

Enabling the client to send CLUSTER SLOTS should help you since we support hostnames in CLUSTER SLOTS. We have this [testcase](https://github.com/Nordix/hiredis-cluster/blob/master/tests/scripts/redirect-with-hostname-test.sh) testing it. You can enable CLUSTER SLOTS with `redisClusterSetOptionRouteUseSlots(cc)`.

Hi @vgmahajanshetty, we have seen the same assert triggered in `nghttp2_pq_remove()` and have not yet found the reason. Do you remember if your investigations revealed anything more?