https://issues.apache.org/jira/browse/CASSANDRA-14672 is almost certainly due to pre-existing corruption . That the user is seeing 14672 is due to extra guards added in 3.11.3, but 14672 isn't likely going to hit you unless you're subject to https://issues.apache.org/jira/browse/CASSANDRA-14515 , which is a much more important bug (that is: 3.11.2 has a data loss bug, 3.11.3 just breaks the read)
On Tue, Sep 11, 2018 at 6:21 AM Ahmed Eljami <ahmed.elj...@gmail.com> wrote: > Any opinion please ? > > Le jeu. 6 sept. 2018 à 22:18, Ahmed Eljami <ahmed.elj...@gmail.com> a > écrit : > >> Hi, >> >> We are testing Cassandra 3.11.2 and we sawed that it contains a critcal >> bug wich was fixed in 3.11.3 ( >> https://issues.apache.org/jira/browse/CASSANDRA-13929). >> >> After about 1 months of testing, we haven't encountered this bug in our >> environnement, but to be sure before going in production, we would like to >> know if it's possible to backport this patch in version 3.11.2 ? >> >> Do you think that this patch could be backported wihtout >> impact/incompatibility on 3.11.2 ? Or it be safer to migrate to Cassandra >> 3.11.3 ? >> >> Our fears that 3.11.3 also not stable due to this bug: >> https://issues.apache.org/jira/browse/CASSANDRA-14672 >> >> Best regards. >> >> > > -- > Cordialement; > > Ahmed ELJAMI >
