zookeeper-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Norbert Kalmar <nkal...@cloudera.com.INVALID>
Subject Re: About the error zknode of replication
Date Mon, 28 Jan 2019 19:03:24 GMT
Hi Shen,

This looks like you should ask in the HBase user list. It is HBase related,
as it failed to cleanup the old znodes. Nothing in ZooKeeper you can do
really. You can either rmr recursively (which will break your replication)
or one-by-one by date maybe? Anyway, I wouldn't recommend that either.

Perhaps there is some way to do a cleanup from HBase command line. Maybe
even stop and re-enable replication so that the znodes can be removed, thus
the oldWALs cleaned.

I would write to HBase user list.

Regards,
Norbert

On Mon, Jan 28, 2019 at 6:05 PM 沈忆珠 <xzr1391@gmail.com> wrote:

> Hi,
>
> I am using hbase replication in my clusters.
> Perhaps because of the region server crash previously, I find that
> there are hundreds of thousands of zknode in /hbase/replication/rs.
> And this makes files in /hbase/oldWALs cannot be cleaned.
> I want to remove theses invalid zknode in zookeeper. But it's
> unrealistic to delete them manually because the number is too large.
> Now I am confused about the method to do the deletion without
> affecting those valid zknode for current replication.
> Can you please help me.
>
> Regards,
> shen
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message