M3DB client asynchronous replication #1859
Conversation
|
Hey this is cool, adding some comments now. |
Codecov Report
@@ Coverage Diff @@
## master #1859 +/- ##
=======================================
Coverage 62.2% 62.2%
=======================================
Files 1104 1104
Lines 104725 104725
=======================================
Hits 65160 65160
Misses 35318 35318
Partials 4247 4247
Continue to review full report at Codecov.
|
simonrobb
changed the title
simonrobb
force-pushed
the
srobb/cluster-replication
branch
2 times, most recently
from
94c1ece to
eb7e7ec
Compare
simonrobb
force-pushed
the
srobb/cluster-replication
branch
2 times, most recently
from
476cee4 to
90eeb5d
Compare
richardartoul
force-pushed
the
srobb/cluster-replication
branch
from
444b2c5 to
ca16e6f
Compare
| func (s replicatedSession) Close() error { | ||
| err := s.session.Close() | ||
| for _, as := range s.asyncSessions { | ||
| as.Close() |
There was a problem hiding this comment.
Probably want to collect the errors here. (so you know if closing session fails, shouldn't matter too much on shutdown whether an error is returned in the case of an error with a real async session)
There was a problem hiding this comment.
Let me know if logging is enough, not sure if there's an idiomatic way to handle multiple errors without early abort. Capture and concat better?
|
Just wondering what the need is for ReplicationOptions, etc. Is there a way it could just be added to the existing Options/AdminOptions? It would help avoid breaking compat on the call to |
simonrobb
force-pushed
the
srobb/cluster-replication
branch
from
f8c4411 to
6e00f60
Compare
| environment: | ||
| - M3DB_HOST_ID=cluster_a_m3db_local_2 | ||
| volumes: | ||
| - "./m3dbnode-cluster-a.yml:/etc/m3dbnode/m3dbnode.yml" |
There was a problem hiding this comment.
🔥 multi-cluster test ftw
What this PR does / why we need it:
Adds support for asynchronously replicating client writes to multiple M3DB clusters.