Mailing-List: contact commits-help@cassandra.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@cassandra.apache.org
Date: Wed, 1 Mar 2017 05:06:46 +0000 (UTC)
From: "Jeff Jirsa (JIRA)" <jira@apache.org>
To: commits@cassandra.apache.org
Message-ID: <JIRA.13047190.1488344561000.26386.1488344806106@Atlassian.JIRA>
In-Reply-To: <JIRA.13047190.1488344561000@Atlassian.JIRA>
References: <JIRA.13047190.1488344561000@Atlassian.JIRA> <JIRA.13047190.1488344561247@jira-lw-us.apache.org>
Subject: [jira] [Updated] (CASSANDRA-13282) Commitlog replay may fail if
 last mutation is within 4 bytes of end of segment
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
archived-at: Wed, 01 Mar 2017 05:06:51 -0000


     [ https://issues.apache.org/jira/browse/CASSANDRA-13282?page=3Dcom.atl=
assian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Jeff Jirsa updated CASSANDRA-13282:
-----------------------------------
    Description: Following CASSANDRA-9749 , stricter correctness checks on =
commitlog replay can incorrectly detect "corrupt segments" and stop commitl=
og replay (and potentially stop cassandra, depending on the configured poli=
cy). In {{CommitlogReplayer#replaySyncSection}} we try to read a 4 byte int=
 {{serializedSize}}, and if it's 0 (which will happen due to zeroing when t=
he segment was created), we continue on to the next segment. However, it ap=
pears that if a mutation is sized such that it ends with 1, 2, or 3 bytes r=
emaining in the segment, we'll pass the {{isEOF}} on the while loop but fai=
l to read the {{serializedSize}} int, and fail.   (was: Following CASSANDRA=
-9749 , stricter correctness checks on commitlog replay can incorrectly det=
ect "corrupt segments" and stop commitlog replay (and potentially stop cass=
andra, depending on the configured policy). In {{CommitlogReplayer#replaySy=
ncSection}} we try to read a 4 byte int {{serializedSize}}, and if it's 0 (=
which will happen due to zeroing when the segment was created), we continue=
 on to the next segment. However, it appears that if a mutation is sized su=
ch that it ends with 1, 2, or 3 bytes remaining in the segment, we'll hit p=
ass the {{isEOF}} on the while loop but fail to read the {{serializedSize}}=
 int, and fail. )

> Commitlog replay may fail if last mutation is within 4 bytes of end of se=
gment
> -------------------------------------------------------------------------=
-----
>
>                 Key: CASSANDRA-13282
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-1328=
2
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>            Reporter: Jeff Jirsa
>            Assignee: Jeff Jirsa
>             Fix For: 3.0.x, 3.11.x, 4.x
>
>
> Following CASSANDRA-9749 , stricter correctness checks on commitlog repla=
y can incorrectly detect "corrupt segments" and stop commitlog replay (and =
potentially stop cassandra, depending on the configured policy). In {{Commi=
tlogReplayer#replaySyncSection}} we try to read a 4 byte int {{serializedSi=
ze}}, and if it's 0 (which will happen due to zeroing when the segment was =
created), we continue on to the next segment. However, it appears that if a=
 mutation is sized such that it ends with 1, 2, or 3 bytes remaining in the=
 segment, we'll pass the {{isEOF}} on the while loop but fail to read the {=
{serializedSize}} int, and fail.=20


--
This message was sent by Atlassian JIRA
(v6.3.15#6346)