incubator-cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jonas Borgström <jonas.borgst...@trioptima.com>
Subject Re: Possible EOFException regression in 0.7.1
Date Wed, 16 Feb 2011 16:54:34 GMT
On 02/16/2011 03:54 PM, Jonathan Ellis wrote:
> It does look a lot like 1932.  Make sure everything is really running
> 0.7.2, 0.7.0 can't read data files created by 0.7.1+.

All nodes are running 0.7.2

> If the versions are ok, take a snapshot, then compact, and see if the
> problem still occurs with all files on the [compacted] new format.

Yes I can confirm that rewriting all sstables by issuing "compact" did
the trick. Before I started I noticed that some nodes had both e and f
sstables and some only f files. So I guess the problem is that the
version f files generated by 0.7.1 sometimes contained corrupted column
name bloom filters somehow.
Anyway, "compact" did make the problem go away.

Thanks!

/ Jonas

Mime
View raw message