Hi all,

We are facing a issue in reading the data from zookeeper. Actually whenever
there is a failure in processing
on bolt, storm writes it on zookeeper.(by raising reportedfailedexception)
we are using trident with some code changes in trident code base.

We have 7 tuple(filenames) batch. we are processing the data of a file in
bolts.

We have written the read api for zookeeper which parsing the znodes data
and return the list of failures.

In some cases we are getting 0 number failures even there were failures in
the processing.

We tested our read api extensively. No issues in that.

There are no issues in local cluster but facing problem on multi node
cluster.

Any inputs ?

Regards
Tarkeshwar

Reply via email to