zookeeper-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Andor Horvath <ahorv...@cloudera.com.INVALID>
Subject how to check consistency in ZK 3.4
Date Thu, 31 Jan 2019 13:57:28 GMT
Hello Gurus,

I came across an issue in which I would need to check if the state of my ZK
servers are consistent and healthy in matters of their snapshots. I would
like to ask your guidance related to this.

The root cause of the issue was that the ZK dataDir (which is the same as
the logDIr despite the recommendations) filled up, which caused the ZK
service go down, leaving behind a total of 2 invalid snapshot files.

When I sidelined them (with the rest of the snapshot files and log files
too) to a temporary dir, starting up the service I see that the newly
created snapshot files' transaction id suffix matches in some occasion, and
do not in some other.
Is this a normal behavior? I mean I think it is, because the service is
healthy, but I would like to be sure that if next time the service halts,
whichever ZKS is elected as leader, I won't have dataloss.

Looking forward for your assistance,
Andor Horvath

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