Hide Forgot
Section 6.4.4, example 6.2, defines the size of the XID size at 53 bytes, while the previous paragraph defines it as 25 bytes. Which one is accurate?
Please provide further information for documentation. LKB
There's nothing else I can provide. Engineering will need to clarify which numbers are correct. Thanks.
Gordon, Are you able to help out with source documentation? LKB
Kim, can you have a look at the apparent inconsistency in the journal sizing text and comment on corrections needed?
I can't find any reference to 25 bytes, but 24 bytes is the size used for local transactions according to the text, I assume this is what you mean in comment #1. Section 6.4.4, paragraph 1 starts with: "When transactions are used, a transaction ID (XID) is added to each record. The size of the XID is 24 bytes for local transactions. For distributed transactions, the user supplies the XID, which is usually obtained from the transaction monitor, and may be any size. ..." In the following sections describing each record type, the transaction is described as: "Transaction ID (XID) size (24 bytes for local transactions, arbitrary size for distributed transactions)" The confusion probably arises from Example 6.2 not explicitly stating that the transactions will be distributed, and are supplied by the user. In this example, the supplied XID is a human-readable UUID format of 53 bytes (as opposed to the local transaction format which is binary and 24 bytes). I suggest the following changes to the text of Example 6.2: Old text: "This example shows how to calculate journal size when transactions are used with a persistent queue. Here are the characteristics of the messaging queue for this example:" New text: "This example shows how to calculate journal size when distributed transactions are used with a persistent queue. The XID will be supplied by the user in this case. Here are the characteristics of the messaging queue for this example:"