P*****x
发帖数: 72 | 1 we have a cluster on EC2. replication factor 3, simple strategy, quorum read
, quorum write. GC grace using default and node tool repair once a week
Cassandra V1.2.1 hector api 1.0.5
we have a column family used for indexing and all columns are dynamic
composite columns. Here is strange thing, when we do lots of writes and
deletes. Somehow lots of deletes are simply like never happened. Even later
we issued the same comments again, we still see some of columns not being
deleted. If we do range slice, some "deleted" columns always pop up.
Any suggestion?
Thanks a lot. Really appreciate your help. | w**z
发帖数: 8232 | 2 did you check after the repair is run? if you can repro it consistently,
make sure your delete method is implemented correctly. I suggest you to
post your question to Cassandra mailing list to get better response.
read
later
【在 P*****x 的大作中提到】
: we have a cluster on EC2. replication factor 3, simple strategy, quorum read
: , quorum write. GC grace using default and node tool repair once a week
: Cassandra V1.2.1 hector api 1.0.5
: we have a column family used for indexing and all columns are dynamic
: composite columns. Here is strange thing, when we do lots of writes and
: deletes. Somehow lots of deletes are simply like never happened. Even later
: we issued the same comments again, we still see some of columns not being
: deleted. If we do range slice, some "deleted" columns always pop up.
: Any suggestion?
: Thanks a lot. Really appreciate your help.
|
|
