jbr: Sending rollback from failed fop to fdl

In case of a failed fop, the failure is detected
by the leader in the jbr-server in two places. First
during a quorum check of +ve responses when it
receives responses from all the followers. At this
point if the fop hasn't been successfully journaled
at a quorum of followers (as in there is no merit in
trying the fop in the leader as the quorum will never
be met), then we fail the fop.

Also if this quorum is met, then the fop is tried on
the leader, and after the leader completes the fop
a quorum check similar to the previous one is done
again, this time including the leaders outcome. If
quorum is not met, then we fail the fop.

In both these cases, when the fop fails we send a -ve
ack to the client. With this patch, now we will also
send a rollback through a GF_FOP_IPC to all the followers(and
also to the leader in the second case of failure). This
rollback will contain the index and term number of the
fop which failed. This will be recorded in the respective
journals of the bricks and will be used to rollback the
fop on that brick later.

A subsequent write, and it's respective rollback would
look something like the following in the journal.

The trusted.jbr.term and trusted.jbr.index present in the
dict of both the logs, relate them, and the presence of
"rollback-fop" in the dict of IPC indicates that it is a
rollback fop, and the value 13(stands for GF_FOP_WRITE)
indicates what kind of rollback operation it is.

=== GF_FOP_WRITE
fd = <gfid 77f12ea2-ca56-40e3-a46e-ba2308baa035>
vector = <158 bytes>
offset = 0 (0x0)
flags = 32769 (0x8001)
xdata = dict {
 trusted.jbr.term = 0 <2 bytes>
 trusted.jbr.index = 4 <2 bytes>
}
=== GF_FOP_IPC
xdata = dict {
 trusted.jbr.term = 0 <2 bytes>
 trusted.jbr.index = 4 <2 bytes>
 rollback-fop = 13 <3 bytes>
}

Change-Id: I70b6a143d20697153d58e2f719e34ecd1ed160a5
BUG: 1349385
Signed-off-by: Avra Sengupta <asengupt@redhat.com>
Reviewed-on: http://review.gluster.org/14783
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
Reviewed-by: Jeff Darcy <jdarcy@redhat.com>
Smoke: Gluster Build System <jenkins@build.gluster.org>

7 files changed, 126 insertions, 25 deletions

diff --git a/xlators/experimental/fdl/src/Makefile.am b/xlators/experimental/fdl/src/Makefile.am
index 9ec9d4f06ff..3f1eccc0ed6 100644
--- a/xlators/experimental/fdl/src/Makefile.am
+++ b/xlators/experimental/fdl/src/Makefile.am
@@ -1,7 +1,7 @@
 xlatordir = $(libdir)/glusterfs/$(PACKAGE_VERSION)/xlator/experimental
 xlator_LTLIBRARIES = fdl.la
 
-noinst_HEADERS = jnl-types.h
+noinst_HEADERS = fdl.h
 
 nodist_fdl_la_SOURCES = fdl.c
 fdl_la_LDFLAGS = -module -avoid-version
diff --git a/xlators/experimental/fdl/src/dump-tmpl.c b/xlators/experimental/fdl/src/dump-tmpl.c
index cac1071a9c1..32b0fef6af3 100644
--- a/xlators/experimental/fdl/src/dump-tmpl.c
+++ b/xlators/experimental/fdl/src/dump-tmpl.c
@@ -2,17 +2,42 @@
 #ifndef _CONFIG_H
 #define _CONFIG_H
 #include "config.h"
+#include <ctype.h>
 #endif
 
 #include "glfs.h"
 #include "iatt.h"
 #include "xlator.h"
-#include "jnl-types.h"
+#include "fdl.h"
+
+/*
+ * Returns 0 if the string is ASCII printable *
+ * and -1 if it's not ASCII printable         *
+ */
+int str_isprint (char *s)
+{
+        int ret = -1;
+
+        if (!s)
+                goto out;
+
+        while (s[0] != '\0') {
+                if (!isprint(s[0]))
+                        goto out;
+                else
+                        s++;
+        }
+
+        ret = 0;
+out:
+        return ret;
+}
 
 #pragma fragment DICT
         {
                 int key_len, data_len;
                 char *key_ptr;
+                char *key_val;
                 printf ("@ARGNAME@ = dict {\n");
                 for (;;) {
                         key_len = *((int *)new_meta);
@@ -23,8 +48,14 @@
                         key_ptr = new_meta;
                         new_meta += key_len;
                         data_len = *((int *)new_meta);
+                        key_val = new_meta + sizeof(int);
                         new_meta += sizeof(int) + data_len;
-                        printf (" %s = <%d bytes>\n", key_ptr, data_len);
+                        if (str_isprint(key_val))
+                                printf (" %s = <%d bytes>\n",
+                                        key_ptr, data_len);
+                        else
+                                printf (" %s = %s <%d bytes>\n",
+                                        key_ptr, key_val, data_len);
                 }
                 printf ("}\n");
         }
diff --git a/xlators/experimental/fdl/src/fdl-tmpl.c b/xlators/experimental/fdl/src/fdl-tmpl.c
index fdcfafbac31..a92f6676ce1 100644
--- a/xlators/experimental/fdl/src/fdl-tmpl.c
+++ b/xlators/experimental/fdl/src/fdl-tmpl.c
@@ -21,7 +21,7 @@
 #include "defaults.h"
 #include "syscall.h"
 #include "xlator.h"
-#include "jnl-types.h"
+#include "fdl.h"
 
 /* TBD: make tunable */
 #define META_FILE_SIZE  (1 << 20)
@@ -55,6 +55,9 @@ typedef struct {
         int                     first_term;
 } fdl_private_t;
 
+int32_t
+fdl_ipc (call_frame_t *frame, xlator_t *this, int32_t op, dict_t *xdata);
+
 void
 fdl_enqueue (xlator_t *this, call_stub_t *stub)
 {
@@ -341,8 +344,21 @@ err_unlocked:
 }
 
 int32_t
+fdl_ipc_continue (call_frame_t *frame, xlator_t *this,
+                  int32_t op, dict_t *xdata)
+{
+        /*
+         * Nothing to be done here. Just Unwind. *
+         */
+        STACK_UNWIND_STRICT (ipc, frame, 0, 0, xdata);
+
+        return 0;
+}
+
+int32_t
 fdl_ipc (call_frame_t *frame, xlator_t *this, int32_t op, dict_t *xdata)
 {
+        call_stub_t     *stub;
         fdl_private_t   *priv   = this->private;
         dict_t          *tdict;
         int32_t         gt_err  = EIO;
@@ -381,6 +397,20 @@ fdl_ipc (call_frame_t *frame, xlator_t *this, int32_t op, dict_t *xdata)
                 }
                 break;
 
+        case FDL_IPC_JBR_SERVER_ROLLBACK:
+                /*
+                 * In case of a rollback from jbr-server, dump  *
+                 * the term and index number in the journal,    *
+                 * which will later be used to rollback the fop *
+                 */
+                stub = fop_ipc_stub (frame, fdl_ipc_continue,
+                                     op, xdata);
+                fdl_len_ipc (stub);
+                stub->serialize = fdl_serialize_ipc;
+                fdl_enqueue (this, stub);
+
+                break;
+
         default:
                 STACK_WIND_TAIL (frame,
                                  FIRST_CHILD(this),
@@ -423,7 +453,6 @@ fdl_init (xlator_t *this)
          * bit cleaner than messing with the generation to add a hand-written
          * exception.
          */
-        this->fops->ipc = fdl_ipc;
 
         if (pthread_create(&priv->worker,NULL,fdl_worker,this) != 0) {
                 gf_log (this->name, GF_LOG_ERROR,
diff --git a/xlators/experimental/fdl/src/fdl.h b/xlators/experimental/fdl/src/fdl.h
new file mode 100644
index 00000000000..32e38c93f2d
--- /dev/null
+++ b/xlators/experimental/fdl/src/fdl.h
@@ -0,0 +1,30 @@
+/*
+   Copyright (c) 2016 Red Hat, Inc. <http://www.redhat.com>
+   This file is part of GlusterFS.
+
+   This file is licensed to you under your choice of the GNU Lesser
+   General Public License, version 3 or any later version (LGPLv3 or
+   later), or the GNU General Public License, version 2 (GPLv2), in all
+   cases as published by the Free Software Foundation.
+*/
+
+#ifndef _FDL_H_
+#define _FDL_H_
+
+#define NEW_REQUEST     (uint8_t)'N'
+
+typedef struct {
+        uint8_t         event_type;     /* e.g. NEW_REQUEST */
+        uint8_t         fop_type;       /* e.g. GF_FOP_SETATTR */
+        uint16_t        request_id;
+        uint32_t        ext_length;
+} event_header_t;
+
+enum {
+        FDL_IPC_BASE = 0xfeedbee5,       /* ... and they make honey */
+        FDL_IPC_CHANGE_TERM,
+        FDL_IPC_GET_TERMS,
+        FDL_IPC_JBR_SERVER_ROLLBACK
+};
+
+#endif /* _FDL_H_ */
diff --git a/xlators/experimental/fdl/src/gen_fdl.py b/xlators/experimental/fdl/src/gen_fdl.py
index 7f6b1aaaeaa..d59f12a4841 100755
--- a/xlators/experimental/fdl/src/gen_fdl.py
+++ b/xlators/experimental/fdl/src/gen_fdl.py
@@ -300,19 +300,44 @@ def get_special_subs (args):
 			ser_code += ser_tmpl.replace("@SRC@",src)
 	return len_code, ser_code
 
+# Mention those fops in the selective_generate table, for which
+# only a few common functions will be generated, and mention those
+# functions. Rest of the functions can be customized
+selective_generate = {
+		"ipc":			"len,serialize",
+	}
+
 def gen_fdl ():
 	entrypoints = []
 	for name, value in ops.iteritems():
 		if "journal" not in [ x[0] for x in value ]:
 			continue
+
+		# generate all functions for all the fops
+		# except for the ones in selective_generate for which
+		# generate only the functions mentioned in the
+		# selective_generate table
+		gen_funcs = "len,serialize,callback,continue,fop"
+		if name in selective_generate:
+			gen_funcs = selective_generate[name].split(",")
+
 		len_code, ser_code = get_special_subs(value)
 		fop_subs[name]["@LEN_CODE@"] = len_code[:-1]
 		fop_subs[name]["@SER_CODE@"] = ser_code[:-1]
-		print generate(LEN_TEMPLATE,name,fop_subs)
-		print generate(SER_TEMPLATE,name,fop_subs)
-		print generate(CBK_TEMPLATE,name,cbk_subs)
-		print generate(CONTINUE_TEMPLATE,name,fop_subs)
-		print generate(FOP_TEMPLATE,name,fop_subs)
+		if 'len' in gen_funcs:
+			print generate(LEN_TEMPLATE,name,fop_subs)
+		if 'serialize' in gen_funcs:
+			print generate(SER_TEMPLATE,name,fop_subs)
+		if name == 'writev':
+			print "#define DESTAGE_ASYNC"
+		if 'callback' in gen_funcs:
+			print generate(CBK_TEMPLATE,name,cbk_subs)
+		if 'continue' in gen_funcs:
+			print generate(CONTINUE_TEMPLATE,name,fop_subs)
+		if 'fop' in gen_funcs:
+			print generate(FOP_TEMPLATE,name,fop_subs)
+		if name == 'writev':
+			print "#undef DESTAGE_ASYNC"
 		entrypoints.append(name)
 	print "struct xlator_fops fops = {"
 	for ep in entrypoints:
diff --git a/xlators/experimental/fdl/src/jnl-types.h b/xlators/experimental/fdl/src/jnl-types.h
deleted file mode 100644
index 8cb39d01a25..00000000000
--- a/xlators/experimental/fdl/src/jnl-types.h
+++ /dev/null
@@ -1,14 +0,0 @@
-#define NEW_REQUEST     (uint8_t)'N'
-
-typedef struct {
-        uint8_t         event_type;     /* e.g. NEW_REQUEST */
-        uint8_t         fop_type;       /* e.g. GF_FOP_SETATTR */
-        uint16_t        request_id;
-        uint32_t        ext_length;
-} event_header_t;
-
-enum {
-        FDL_IPC_BASE = 0xfeedbee5,       /* ... and they make honey */
-        FDL_IPC_CHANGE_TERM,
-        FDL_IPC_GET_TERMS,
-};
diff --git a/xlators/experimental/fdl/src/recon-tmpl.c b/xlators/experimental/fdl/src/recon-tmpl.c
index 523bda39418..ab5edb1a378 100644
--- a/xlators/experimental/fdl/src/recon-tmpl.c
+++ b/xlators/experimental/fdl/src/recon-tmpl.c
@@ -11,7 +11,7 @@
 #include "xlator.h"
 #include "glfs-internal.h"
 
-#include "jnl-types.h"
+#include "fdl.h"
 
 #define GFAPI_SUCCESS 0