cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jason Brown (JIRA)" <>
Subject [jira] [Commented] (CASSANDRA-3648) Repair should validate checksums before streaming
Date Thu, 31 Jan 2013 14:37:14 GMT


Jason Brown commented on CASSANDRA-3648:

Overall, I think the approach and code look fine, with the only major nit being the misspelling
in the DataIntegratyMetadata class name :). (should be DataIntegrityMetadata).
I like that this patch will also catch problems with bootstrapping, as well. 

bq. But we might not need checksum over the wire (let TCP handle errors)?

This seems reasonable. We've checked that it's legit before shipping it (from the jvm).

I guess my main question is: are compressed sstables validated, as well?  I think they are,
via CRAR.decompressChunk(), but just want to confirm.

Another question I have, and it might be for a separate ticket as the problem existed before,
is what should happen when one file fails to stream when there are several (or many) to stream.
Currently the entire stream session is cancelled, starting from the failed file onward. Thus,
if we wanted to stream ten files, and the second one fails, we're not going to stream the
remaining eight.

Otherwise, LGTM. +1
> Repair should validate checksums before streaming
> -------------------------------------------------
>                 Key: CASSANDRA-3648
>                 URL:
>             Project: Cassandra
>          Issue Type: New Feature
>            Reporter: Jonathan Ellis
>            Assignee: Vijay
>            Priority: Minor
>             Fix For: 2.0
>         Attachments: 0001-CASSANDRA-3648.patch
> We have block checksums and sha1 full-file hashes, but currently we do not validate these
before streaming in a repair.

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see:

View raw message