hcq/MaxScale
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

MXS-1323: Fix crash on attempted retry of read

Browse Source 
When a backend is waiting for a response but no statement is stored for
the session, the buffer where the stored statement is copied is not
modified. This means that it needs to be initialized to a NULL value.

Added a test that checks that the behavior works as expected even with
persistent connections. A second test reproduces the crash by executing
parallel SET commands while slaves are blocked.

There is still a behavioral problem in readwritesplit. If a session
command is being executed and it fails on a slave, an error is sent to the
client. In this case it would not be necessary to close the session if the
master is still alive.
This commit is contained in:
Markus Mäkelä
2017-07-25 09:46:40 +03:00
parent 2148d78d3f
commit ed44c45be1
7 changed files with 206 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
server/modules/routing/readwritesplit/readwritesplit.c
View File

@ -1541,8 +1541,8 @@ static bool handle_error_new_connection(ROUTER_INSTANCE *inst,
*/
if (BREF_IS_WAITING_RESULT(bref))
{
GWBUF *stored;
const SERVER *target;
GWBUF *stored = NULL;
const SERVER *target = NULL;
if (!session_take_stmt(backend_dcb->session, &stored, &target) ||
target != bref->ref->server ||
@ -1554,8 +1554,14 @@ static bool handle_error_new_connection(ROUTER_INSTANCE *inst,
*/
gwbuf_free(stored);
DCB *client_dcb = ses->client_dcb;
client_dcb->func.write(client_dcb, gwbuf_clone(errmsg));
if (!sescmd_cursor_is_active(&bref->bref_sescmd_cur))
{
/** The client expects a response from this exact backend.
* We need to route an error to the client to let it know
* that the query failed. */
DCB *client_dcb = ses->client_dcb;
client_dcb->func.write(client_dcb, gwbuf_clone(errmsg));
}
}
}