shanthoosh commented on issue #1079: SAMZA-2250: Support large job models in
standalone.
URL: https://github.com/apache/samza/pull/1079#issuecomment-505232922
Travis build for this patch is green, just that it is not getting updated
here.
shanthoosh commented on issue #1079: SAMZA-2250: Support large job models in
standalone.
URL: https://github.com/apache/samza/pull/1079#issuecomment-503360594
@sborya
I addressed both of your comments. Can you please take a look.
shanthoosh commented on issue #1079: SAMZA-2250: Support large job models in
standalone.
URL: https://github.com/apache/samza/pull/1079#issuecomment-503359306
@prateekm
Added checksum to the end of metadata-value in `ZkMetadataStore`. It is
used for verification during read and
shanthoosh commented on issue #1079: SAMZA-2250: Support large job models in
standalone.
URL: https://github.com/apache/samza/pull/1079#issuecomment-503309436
@sborya
1. I've moved the chunking logic into the zookeeper metadata store
implementation.
2. I've already replied to
shanthoosh commented on issue #1079: SAMZA-2250: Support large job models in
standalone.
URL: https://github.com/apache/samza/pull/1079#issuecomment-502860180
@dnishimura
> Do we know if batch puts would increase performance by an order of
magnitude
It was not a significant
shanthoosh commented on issue #1079: SAMZA-2250: Support large job models in
standalone.
URL: https://github.com/apache/samza/pull/1079#issuecomment-502859889
@prateekm
> After leader election the job model is always the latest version? Are
there special requirements around
shanthoosh commented on issue #1079: SAMZA-2250: Support large job models in
standalone.
URL: https://github.com/apache/samza/pull/1079#issuecomment-501934727
@prateekm
Thanks for the review.
1. Segmenting the `JobModel` and storing it as independent byte array chunks
will