Here are some more details. I used GBAK to backup and then restored the
database. Here are the gstat details:
Database header page information:
Flags 0
Checksum 12345
Generation 30
Page size 4096
ODS version 11.2
Oldest transaction 1
Oldest active 2
Oldest snapshot 2
Next transaction 22
Bumped transaction 1
Sequence number 0
Next attachment ID 2
Implementation ID 26
Shadow count 0
Page buffers 0
Next header page 0
Database dialect 3
Creation date Dec 19, 2015 14:27:42
Attributes force write
Variable header data:
Sweep interval: 0
*END*
I ran the import process again for several minutes and here are the revised
stats. It had stalled again and I let it continue doing whatever it does.
I am not sure if I took these stats during or after the stall completed:
Database header page information:
Flags 0
Checksum 12345
Generation 16527
Page size 4096
ODS version 11.2
Oldest transaction 16387
Oldest active 16388
Oldest snapshot 16388
Next transaction 16515
Bumped transaction 1
Sequence number 0
Next attachment ID 37
Implementation ID 26
Shadow count 0
Page buffers 0
Next header page 0
Database dialect 3
Creation date Dec 19, 2015 14:27:42
Attributes force write
Variable header data:
Sweep interval: 0
*END*
Is there something obviously wrong with the transactions? Thank you.
On Sat, Dec 19, 2015 at 4:43 PM, Caroline Beltran <
[email protected]> wrote:
> I have happily used Firebird 2.5 for a web application that run 24x7 for
> at least 2 years. The sweep interval is set to the default 20,000 and it
> works very well. There is at least 10 inserts per minute but not much more
> than that. I would guess that there are at least 20 reads per second and
> I've been completely happy with this trouble free application.
>
> Now for the other problematic application...
>
> This is also an online web application which runs wonderfully except
> during our data import process. This process will INSERT on average about
> 200 records per minute. But approximately 10% of those INSERTS require a
> large number of SELECTs and several UPDATEs to other records. This is not
> a simple case of INSERTing new records.
>
> Everything will run very well for a short period of time but then
> everything just stalls for several minutes. CPU consumption is always low,
> even when the stall happens. But the I/O Read and Write is quite high,
> even when stalled. This tells me that FB is busy doing something.
>
> For this particular database, I decided to set the Sweep Interval to 0 but
> this stalling still happens.
>
> Just as soon as I notice that FB stalls, I stop the data import process
> and let FB do whatever it needs to do until it finishes. I can then resume
> data importation as normal until it stalls again in a few hours.
>
> Is it possible that sweeping is still being triggered even though I have
> set the interval to 0?
>
> Thank you
>
