At 07:32 AM 20/12/2006, you wrote:

>Hi,
>
>Is a read-only read committed long running transaction harmfull in FB2.0?

If you are using the release version, no. If you are using a beta
version of Classic on a fast machine, possibly. An inherent bug
related to this issue (CORE-911) was unearthed and fixed quite late
in the release candidate cycle.

[ Users of Fb 1.5.x should be aware that this bug still exists in
releases < v.1.5.4. The fix has been backported to v.1.5.4, due out soon. ]

>Is it blocking OIT/OAT?

Not sure what you mean by "blocking" here, but it is both an active
transaction and an interesting transaction. As a read-only
transaction it has no potential to create its own garbage, so other
transactions are not interested in it and GC knows that *it* is not
interested in back versions, due to its Read Committed
isolation. However, unlike other transactions, it is not an
interesting transaction once it has been committed or rolled back.

>Is it blocking garbage collection?

No.

>Or can I use the read-only transaction running for a long time without
>performance issues (back versions aren't garbage collected?)

Yes. As a read-only transaction, it doesn't create garbage and, as
long as it uses Read Committed isolation, it doesn't care about what
is in the garbage. In other transaction isolation settings, the
normal garbage-flagging dependencies apply to read-only transactions.

./heLen