From 4da244caccd38a77de5428b6954f565219ef0719 Mon Sep 17 00:00:00 2001 From: Sanju Rakonde Date: Fri, 6 Apr 2018 01:53:45 +0530 Subject: glusterd: handling brick termination in brick-mux Problem: There's a race between the glusterfs_handle_terminate() response sent to glusterd from last brick of the process and the socket disconnect event that encounters after the brick process got killed. Solution: When it is a last brick for the brick process, instead of sending GLUSTERD_BRICK_TERMINATE to brick process, glusterd will kill the process (same as we do it in case of non brick multiplecing). The test case is added for https://bugzilla.redhat.com/show_bug.cgi?id=1549996 Change-Id: If94958cd7649ea48d09d6af7803a0f9437a85503 fixes: bz#1545048 Signed-off-by: Sanju Rakonde --- glusterfsd/src/glusterfsd-mgmt.c | 24 ++++++++++++------------ 1 file changed, 12 insertions(+), 12 deletions(-) (limited to 'glusterfsd') diff --git a/glusterfsd/src/glusterfsd-mgmt.c b/glusterfsd/src/glusterfsd-mgmt.c index db961c86304..e30d9287575 100644 --- a/glusterfsd/src/glusterfsd-mgmt.c +++ b/glusterfsd/src/glusterfsd-mgmt.c @@ -174,7 +174,6 @@ glusterfs_terminate_response_send (rpcsvc_request_t *req, int op_ret) ret = dict_allocate_and_serialize (dict, &rsp.output.output_val, &rsp.output.output_len); - if (ret == 0) ret = glusterfs_submit_reply (req, &rsp, NULL, 0, NULL, (xdrproc_t)xdr_gd1_mgmt_brick_op_rsp); @@ -188,15 +187,15 @@ glusterfs_terminate_response_send (rpcsvc_request_t *req, int op_ret) int glusterfs_handle_terminate (rpcsvc_request_t *req) { - gd1_mgmt_brick_op_req xlator_req = {0,}; + gd1_mgmt_brick_op_req xlator_req = {0,}; ssize_t ret; - glusterfs_ctx_t *ctx = NULL; - xlator_t *top = NULL; - xlator_t *victim = NULL; - xlator_t *tvictim = NULL; - xlator_list_t **trav_p = NULL; - gf_boolean_t lockflag = _gf_false; - gf_boolean_t last_brick = _gf_false; + glusterfs_ctx_t *ctx = NULL; + xlator_t *top = NULL; + xlator_t *victim = NULL; + xlator_t *tvictim = NULL; + xlator_list_t **trav_p = NULL; + gf_boolean_t lockflag = _gf_false; + gf_boolean_t still_bricks_attached = _gf_false; ret = xdr_to_generic (req->msg[0], &xlator_req, (xdrproc_t)xdr_gd1_mgmt_brick_op_req); @@ -240,15 +239,16 @@ glusterfs_handle_terminate (rpcsvc_request_t *req) glusterfs_terminate_response_send (req, 0); for (trav_p = &top->children; *trav_p; trav_p = &(*trav_p)->next) { tvictim = (*trav_p)->xlator; - if (!tvictim->cleanup_starting && !strcmp (tvictim->name, xlator_req.name)) { + if (!tvictim->cleanup_starting && + !strcmp (tvictim->name, xlator_req.name)) { continue; } if (!tvictim->cleanup_starting) { - last_brick = _gf_true; + still_bricks_attached = _gf_true; break; } } - if (!last_brick) { + if (!still_bricks_attached) { gf_log (THIS->name, GF_LOG_INFO, "terminating after loss of last child %s", xlator_req.name); -- cgit