> I double-checked and it is possible that the checkpoint thread is
> sharing an sqlite3* with a writer thread.
That was the problem. I gave the SQLITE_CHECKPOINT_PASSIVE it's own
sqlite3*, and it is now working fine. Thanks much.
___
sqlite-users
> Are each of your threads using their own handles or do they all share the
> same handle?
I double-checked and it is possible that the checkpoint thread is
sharing an sqlite3* with a writer thread. If so, that would account
for the problem, I think. I'll look into this further and post a
> > sqlite3_busy_timeout(db, 0x7000)
> If you want to turn timeouts off, specify zero or a negative number.
My understanding is that timeouts off means the functions immediately
return BUSY. I don't want timeouts off, I want them set to infinity,
so the functions never in my lifetime return
Sorry I can't solve your problem (too technical for me) but I have notes on the
testing code.
On 5 Sep 2016, at 8:54pm, Allen wrote:
> sqlite3_busy_timeout(db, 0x7000)
Please don't do this. It makes things hard to read and dependent on
architecture details, i.e. the
On 09/06/2016 02:54 AM, Allen wrote:
I'm doing some stress testing of an app that uses
sqlite-amalgamation-3140100 compiled under gcc 4.8.5
(x86_64-posix-seh-rev0, Built by MinGW-W64 project) on Windows 7 x64 with
the compile options:
#define SQLITE_MAX_MMAP_SIZE0
#define
I'm doing some stress testing of an app that uses
sqlite-amalgamation-3140100 compiled under gcc 4.8.5
(x86_64-posix-seh-rev0, Built by MinGW-W64 project) on Windows 7 x64 with
the compile options:
#define SQLITE_MAX_MMAP_SIZE0
#define SQLITE_OMIT_AUTORESET1
#define
6 matches
Mail list logo