Skip to content

Releases: ifwe/twemproxy

0.5.2-tmg19

30 Jun 13:54
7d3cafc
Compare
Choose a tag to compare
0.5.2-tmg19 Pre-release
Pre-release

Add support for a failover pool with arbitrary depth (tyson, john irwin)
Improve tolerance of DNS failures in commands sent to multiple
redis/memcache servers such as memcache get(multiget), redis mget, etc.
(i.e. fragmented commands) (tyson)

Raise the severity of establishing connections, connection errors and pool update logs to
LOG_NOTICE (tyson)

Fix memory corruption seen when all hosts are ejected from a pool with
auto_eject_hosts: true (tyson)

Update tests to work with python 3 (tyson)

Ensure file permissions aren't chosen based on uninitialized memory. (tyson)

Add a heartbeat patch to detect dead servers (charsyam)

Add improvements to the heartbeat patch (tyson)

Decrease the verbosity of logging clients establishing/closing connections to
nutcracker to LOG_INFO for use cases that can't support persistent connections (tyson)

Micro-optimization: Don't fragment memcache multigets with a single key
into an array of commands to coalesce. (tyson)

Support new redis commands and changes to argument counts of existing
commands. (tyson)

Support nested multi-bulk redis commands (arrays of arrays...)
with an arbitrary depth. (qingping209, tyson)

Log the event processing backend used and whether debug assertions are
enabled in the --help/--version output. (tyson)

Micro-optimizations for allocating msg structures, etc (tyson)

Don't bother computing hash or looking up in distribution when a pool
has only one server - the result will always be the same (tyson)

Fix parsing bug when a redis error body contains no spaces (tyson)

Ignore server_failure_limit and treat it as if it's always 1 - the heartbeat
patch is responsible for reconnecting after a server failure. (tyson)

Don't send more than one heartbeat request over a single connection to a
server when reconnecting to a server. Reduce concurrent reconnection attempts. (tyson)