[FIX] Apply batch size to prefetch

[chibenwa]

Default value 32 might be too much

[Arsnael]

I'm confused here... You applying the batchSize to prefetch saying 32 might be too much... that implies that configration has been changed for batchSizes, with the assumption the user running James put it lower than 32. However, default batch size value in the code is defined at 200 if conf is missing in batchsize.properties, and also default values in the sample conf files are defined like this:

fetch.metadata=200
fetch.headers=200
fetch.body=100
fetch.full=50

Wouldn't this change make things worse as is in the end, as batchsize default values are higher than 32?

[chibenwa]

I agressively decreased this on one of our customer infrastructure

fetch.metadata=1000

fetch.headers=30
fetch.body=1
fetch.full=1

# IMAP COPY command
copy=10

[Arsnael]

I agressively decreased this on one of our customer infrastructure

James is an ASF project, not a Linagora/customer project :)

I understand you likely did that on that platform, my question is what happens for other people using James with default batch values here in that case? Not sure it's beneficial in this state

[chibenwa]

James is an ASF project, not a Linagora/customer project :)

This is completly unrelated. Please step down.

---

I mean if I set batchSizes to 1 I expect 1 message in memory not 1 message in memory and 32 prefetched...

BTW CassandraMessageMapper already does this

james-project/mailbox/cassandra/src/main/java/org/apache/james/mailbox/cassandra/mail/CassandraMessageMapper.java

Line 262 in d4bf925

.flatMapSequential(metadata -> toMailboxMessage(metadata, ftype), concurrency, concurrency);

[Arsnael]

Misunderstood on my end, sorry. Alright then