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

Make sure DCBs are OK before closing them

Browse Source 
Added a check for the validity of the backend DCBs before they are
closed. This should guarantee that only valid DCBs are closed by
readwritesplit.

However, this is not the correct solution for the problem. The DCB should
not be in an invalid state in the first place and this fix just removes
the bad side effects of the double closing.

With the added logging in the readwritesplit error handler, more detailed
information should become available.
This commit is contained in:
Markus Makela
2016-12-01 17:05:23 +02:00
parent 1d9d325a01
commit 943aa48fb5
1 changed files with 11 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
server/modules/routing/readwritesplit/readwritesplit.c
View File

@ -959,12 +959,19 @@ static void closeSession(ROUTER *instance, void *router_session)
} }
bref_clear_state(bref, BREF_IN_USE); bref_clear_state(bref, BREF_IN_USE);
bref_set_state(bref, BREF_CLOSED); bref_set_state(bref, BREF_CLOSED);
/**
* closes protocol and dcb
*/
RW_CHK_DCB(bref, dcb); RW_CHK_DCB(bref, dcb);
/** MXS-956: This will prevent closed DCBs from being closed twice.
* It should not happen but for currently unknown reasons, a DCB
* gets closed twice; first in handleError and a second time here. */
if (dcb && dcb->state == DCB_STATE_POLLING)
{
dcb_close(dcb); dcb_close(dcb);
}
RW_CLOSE_BREF(bref); RW_CLOSE_BREF(bref);
/** decrease server current connection counters */ /** decrease server current connection counters */
atomic_add(&bref->bref_backend->backend_conn_count, -1); atomic_add(&bref->bref_backend->backend_conn_count, -1);
} }