Re: during a checkpoint
Thanh wrote: > > I seem to read somewhere that queries are not blocked > during a checkpoint. Is this correct ? If not, what operations > are not blocked ? Yes, only operations entering a critical stage are blocked. Typically, anything that wants to change data is going to be blocked. Although you might want to read up on Fuzzy Checkpoints in TFM. Cheers, -- Mark. +----------------------------------------------------------+-----------+ | Mark D. Stock mailto:mdstock@mydas.freeserve.co.uk |//////// /| | http://www.informix.com http://www.informixhandbook.com |///// / //| | http://www.iiug.org +-----------------------------------+//// / ///| | |This email will self-destruct in |/// / ////| | |10 sec. If you received this email |// / /////| | |in error, sorry about the mess. |/ ////////| +----------------------+-----------------------------------+-----------+
"Mark D. Stock" wrote: > > Thanh wrote: > > > > I seem to read somewhere that queries are not blocked > > during a checkpoint. Is this correct ? If not, what operations > > are not blocked ? > > Yes, only operations entering a critical stage are blocked. Typically, > anything that wants to change data is going to be blocked. Although you > might want to read up on Fuzzy Checkpoints in TFM. Well almost. If you have a version earlier than 7.31 (or 9.30) then any query with threads running in CPU VP #1 will be blocked during the first half of the checkpoint while the engine uses that VP exclusively to organize the allocation of CLEANERS to BUFFERS for flushing. Other non-critical threads running in other CPU VPs (assuming you have any) will not be affected. Code changes in 7.31 and fuzzy checkpointing combine to alleviate this problem. Art S. Kagel
In article <39F89DA3.90910385@bloomberg.net>, kagel@bloomberg.net wrote: > "Mark D. Stock" wrote: > > > > Thanh wrote: > > > > > > I seem to read somewhere that queries are not blocked > > > during a checkpoint. Is this correct ? If not, what operations > > > are not blocked ? > > > > Yes, only operations entering a critical stage are blocked. Typically, > > anything that wants to change data is going to be blocked. Although you > > might want to read up on Fuzzy Checkpoints in TFM. > > Well almost. If you have a version earlier than 7.31 (or 9.30) then any > query with threads running in CPU VP #1 will be blocked during the first > half of the checkpoint while the engine uses that VP exclusively to organize > the allocation of CLEANERS to BUFFERS for flushing. Other non-critical > threads running in other CPU VPs (assuming you have any) will not be > affected. Code changes in 7.31 and fuzzy checkpointing combine to > alleviate this problem. > > Art S. Kagel > Thanks Mark and Art. There is great tech note 'Checkpoint Tuning: A Case Study' but this is specific to IDS7.x which does not have the fuzzy checkpoint feature. I would be great if this document can be updated to cover 9.x as well. There are some scripts that were mentioned in this paper. Anyone knows where I can get these scripts ? (The author, Geoffrey Bergren, did not leave a point of contact) Thanks, Thanh Sent via Deja.com http://www.deja.com/ Before you buy.