Commented by: @dyemanov

Do you use local or remote connection string?

Commented by: Vitaly (v6y)

> Do you use local or remote connection string?

Does not matter - it happens in both cases.
For tests I used the following code:
#⁠include <ibase.h>
#⁠include <string.h>
#⁠include <stdio.h>
#⁠include <time.h>
#⁠include <signal.h>
#⁠include <unistd.h>

isc_db_handle db = NULL;
ISC_STATUS_ARRAY status;
int ret;

int TERM=0;

void sigterm(int signal) {
TERM=1;
printf("Got INT or TERM signal\n");
}

#⁠define ERREXIT(status, rc) {isc_print_status(status); return rc;}

int main (void)
{

long long i=0;
signal(SIGTERM,sigterm);
signal(SIGINT,sigterm);

while(!TERM) {
if (isc_attach_database(status, 0, "localhost:employee", &db, 0, NULL))
{
ERREXIT(status,-1);
}
if (isc_detach_database(status, &db))
{
ERREXIT(status,-1);
};
i++;
printf("%i attaches/detaches\n",i);
usleep(10000);
}
signal(SIGTERM,SIG_DFL);
signal(SIGINT,SIG_DFL);
return 0;
}

Because of no dpb here, I used environment variables ISC_USER and ISC_PASSWORD to setup firebird login/password.

So, just after starting of my test, the 'top' programm showed following:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
32636 root 15 0 4368 1948 1516 S 2.0 0.1 0:00.02 fbtst

After about 3500 ataches/detaches:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
32636 root 15 0 4496 2072 1516 S 1.3 0.1 0:01.44 fbtst

After about 10000 ataches/detaches:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
32636 root 15 0 4624 2292 1540 S 0.3 0.1 0:02.74 tst

fter about 12800 ataches/detaches:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
32636 root 16 0 4752 2376 1540 S 0.3 0.1 0:02.95 tst

And so on.

Commented by: @dyemanov

On Win32, running v2.0.0 I see no memory leak at all. Tried 6000 iterations, both IPC and TCP connections.

Commented by: @dyemanov

Alex, could you please try this test case? Maybe it's posix-specific.

Commented by: @AlexPeshkoff

Failed to reproduce mentioned memory leak for both 1.5.4 and 2.0.1 RC2.