A few questions about replication and queues

[LinearGradient]

Hi,

I have heard some contradicting statements about CQ 5's (5.6.1 to be exact) replication system. Could someone please give a definitive answer for the following questions:

• Are replication queues FIFO? Or are they processed concurrently, hence leading to non-FIFO behavior?
• If passed a ReplicationOptions object, would the replicate method respect:
  • isSynchronous()? What would happen when there are already some items in the queue before a synchronous replication (activation) of a Node starts? Would it wait for all the prior items to be replicated before the method returns?
  • Would it respect a custom aggregate handler? Are there any sample codes for aggregate handlers?
    • Should cq:ClientLibraryFolder nodes be treated as aggregate nodes? I have heard that if the css.txt or js.txt files get replicated before the css or js files references in them, they could lead to errors on the publisher instances.
    • What should exactly be returned by the "AggregateHandler.prepareForReplication" method?
    • When a node is identified to be aggregate, which nodes underneath that node get actually published?
• What is the purpose of the ReplicationOptions's updateAlias property?

Thanks in advance.

[Jörg_Hoh · Accepted solution]

Hi,

just to give you a few answers:

• Each replication agent holds its own queue. These queues have a FIFO semantic. But the queues are not linked, so queue A might have processed the items 1,2 and 3, while queue B is still processing item 1.
• I don't have experience with "isSynchronous", because in one case it would take a long time (if all items already in the queue must be processed first) or it breaks the FIFO semantics.
• For the remaining parts please file a Daycare ticket, so the API documentation is enhanced.

Regards,

Jörg