/*
* Copyright 2000, International Business Machines Corporation and others.
* All Rights Reserved.
- *
+ *
* This software has been released under the terms of the IBM Public
* License. For details, see the LICENSE file in the top-level source
* directory or online at http://www.openafs.org/dl/license10.html
#include <afsconfig.h>
#include <afs/param.h>
-RCSID
- ("$Header$");
#include <sys/types.h>
#include <string.h>
*/
/*!
- * if this flag is set, then ubik will use only the primary address
- * (the address specified in the CellServDB) to contact other
- * ubik servers. Ubik recovery will not try opening connections
- * to the alternate interface addresses.
+ * if this flag is set, then ubik will use only the primary address
+ * (the address specified in the CellServDB) to contact other
+ * ubik servers. Ubik recovery will not try opening connections
+ * to the alternate interface addresses.
*/
int ubikPrimaryAddrOnly;
* routine called when a non-sync site server goes down; restarts recovery
* process to send missing server the new db when it comes back up.
*
- * \note This routine should not do anything with variables used by non-sync site servers.
+ * \note This routine should not do anything with variables used by non-sync site servers.
*/
int
urecovery_LostServer(void)
#if !defined(AFS_PTHREAD_ENV)
/* No corresponding LWP_WaitProcess found anywhere for this -- klm */
LWP_NoYieldSignal(&urecovery_state);
- return 0;
#endif
+ return 0;
}
/*!
* and we must have a currently-good sync site.
*/
int
-urecovery_AllBetter(register struct ubik_dbase *adbase, int areadAny)
+urecovery_AllBetter(struct ubik_dbase *adbase, int areadAny)
{
- register afs_int32 rcode;
+ afs_int32 rcode;
- ubik_dprint("allbetter checking\n");
+ ubik_dprint_25("allbetter checking\n");
rcode = 0;
rcode = 1;
}
- ubik_dprint("allbetter: returning %d\n", rcode);
+ ubik_dprint_25("allbetter: returning %d\n", rcode);
return rcode;
}
int
urecovery_AbortAll(struct ubik_dbase *adbase)
{
- register struct ubik_trans *tt;
+ struct ubik_trans *tt;
for (tt = adbase->activeTrans; tt; tt = tt->next) {
udisk_abort(tt);
}
* \brief this routine aborts the current remote transaction, if any, if the tid is wrong
*/
int
-urecovery_CheckTid(register struct ubik_tid *atid)
+urecovery_CheckTid(struct ubik_tid *atid, int abortalways)
{
if (ubik_currentTrans) {
/* there is remote write trans, see if we match, see if this
* is a new transaction */
if (atid->epoch != ubik_currentTrans->tid.epoch
- || atid->counter > ubik_currentTrans->tid.counter) {
+ || atid->counter > ubik_currentTrans->tid.counter || abortalways) {
/* don't match, abort it */
/* If the thread is not waiting for lock - ok to end it */
#if !defined(UBIK_PAUSE)
* the log.
*/
static int
-ReplayLog(register struct ubik_dbase *adbase)
+ReplayLog(struct ubik_dbase *adbase)
{
afs_int32 opcode;
- register afs_int32 code, tpos;
+ afs_int32 code, tpos;
int logIsGood;
afs_int32 len, thisSize, tfile, filePos;
afs_int32 buffer[4];
sizeof(afs_int32));
if (code != sizeof(afs_int32))
break;
+ opcode = ntohl(opcode);
if (opcode == LOGNEW) {
/* handle begin trans */
tpos += sizeof(afs_int32);
if (code != 3 * sizeof(afs_int32))
break;
/* otherwise, skip over the data bytes, too */
- tpos += buffer[2] + 3 * sizeof(afs_int32);
+ tpos += ntohl(buffer[2]) + 3 * sizeof(afs_int32);
} else {
- ubik_dprint("corrupt log opcode (%d) at position %d\n", opcode,
- tpos);
+ ubik_print("corrupt log opcode (%d) at position %d\n", opcode,
+ tpos);
break; /* corrupt log! */
}
}
sizeof(afs_int32));
if (code != sizeof(afs_int32))
break;
+ opcode = ntohl(opcode);
if (opcode == LOGNEW) {
/* handle begin trans */
tpos += sizeof(afs_int32);
len -= thisSize;
}
} else {
- ubik_dprint("corrupt log opcode (%d) at position %d\n",
- opcode, tpos);
+ ubik_print("corrupt log opcode (%d) at position %d\n",
+ opcode, tpos);
break; /* corrupt log! */
}
}
if (code)
return code;
} else {
- ubik_dprint("Log read error on pass 2\n");
+ ubik_print("Log read error on pass 2\n");
return UBADLOG;
}
}
* This routine is called after replaying the log; it reads the restored labels.
*/
static int
-InitializeDB(register struct ubik_dbase *adbase)
+InitializeDB(struct ubik_dbase *adbase)
{
- register afs_int32 code;
+ afs_int32 code;
code = (*adbase->getlabel) (adbase, 0, &adbase->version);
if (code) {
* We replay the logs and then read the resulting file to figure out what version we've really got.
*/
int
-urecovery_Initialize(register struct ubik_dbase *adbase)
+urecovery_Initialize(struct ubik_dbase *adbase)
{
- register afs_int32 code;
+ afs_int32 code;
code = ReplayLog(adbase);
if (code)
char tbuffer[1024];
struct ubik_stat ubikstat;
struct in_addr inAddr;
+ char hoststr[16];
#ifndef OLD_URECOVERY
char pbuffer[1028];
int flen, fd = -1;
/* Every 30 seconds, check all the down servers and mark them
* as up if they respond. When a server comes up or found to
- * not be current, then re-find the the best database and
+ * not be current, then re-find the the best database and
* propogate it.
*/
if ((now = FT_ApproxTime()) > 30 + lastProbeTime) {
+
+#ifdef AFS_PTHREAD_ENV
+ DBHOLD(ubik_dbase);
+#endif
+
for (ts = ubik_servers, doingRPC = 0; ts; ts = ts->next) {
if (!ts->up) {
doingRPC = 1;
urecovery_state &= ~UBIK_RECFOUNDDB;
}
}
+
+#ifdef AFS_PTHREAD_ENV
+ DBRELE(ubik_dbase);
+#endif
+
if (doingRPC)
now = FT_ApproxTime();
lastProbeTime = now;
}
urecovery_state |= UBIK_RECSYNCSITE;
- /* If a server has just come up or if we have not found the
+ /* If a server has just come up or if we have not found the
* most current database, then go find the most current db.
*/
if (!(urecovery_state & UBIK_RECFOUNDDB)) {
rxcall = rx_NewCall(bestServer->disk_rxcid);
ubik_print("Ubik: Synchronize database with server %s\n",
- afs_inet_ntoa(bestServer->addr[0]));
+ afs_inet_ntoa_r(bestServer->addr[0], hoststr));
code = StartDISK_GetFile(rxcall, file);
if (code) {
}
#ifndef OLD_URECOVERY
flen = length;
- afs_snprintf(pbuffer, sizeof(pbuffer), "%s.DB0.TMP", ubik_dbase->pathName);
+ afs_snprintf(pbuffer, sizeof(pbuffer), "%s.DB%s%d.TMP", ubik_dbase->pathName, (file<0)?"SYS":"", (file<0)?-file:file);
fd = open(pbuffer, O_CREAT | O_RDWR | O_TRUNC, 0600);
if (fd < 0) {
code = errno;
code = close(fd);
if (code)
goto FetchEndCall;
-#endif
+#endif
code = EndDISK_GetFile(rxcall, &tversion);
FetchEndCall:
tcode = rx_EndCall(rxcall, code);
#ifdef OLD_URECOVERY
(*ubik_dbase->sync) (ubik_dbase, 0); /* get data out first */
#else
- afs_snprintf(tbuffer, sizeof(tbuffer), "%s.DB0", ubik_dbase->pathName);
+ afs_snprintf(tbuffer, sizeof(tbuffer), "%s.DB%s%d", ubik_dbase->pathName, (file<0)?"SYS":"", (file<0)?-file:file);
#ifdef AFS_NT40_ENV
- afs_snprintf(pbuffer, sizeof(pbuffer), "%s.DB0.OLD", ubik_dbase->pathName);
+ afs_snprintf(pbuffer, sizeof(pbuffer), "%s.DB%s%d.OLD", ubik_dbase->pathName, (file<0)?"SYS":"", (file<0)?-file:file);
code = unlink(pbuffer);
if (!code)
code = rename(tbuffer, pbuffer);
- afs_snprintf(pbuffer, sizeof(pbuffer), "%s.DB0.TMP", ubik_dbase->pathName);
+ afs_snprintf(pbuffer, sizeof(pbuffer), "%s.DB%s%d.TMP", ubik_dbase->pathName, (file<0)?"SYS":"", (file<0)?-file:file);
#endif
- if (!code)
+ if (!code)
code = rename(pbuffer, tbuffer);
if (!code) {
(*ubik_dbase->open) (ubik_dbase, 0);
#ifndef OLD_URECOVERY
}
#ifdef AFS_NT40_ENV
- afs_snprintf(pbuffer, sizeof(pbuffer), "%s.DB0.OLD", ubik_dbase->pathName);
+ afs_snprintf(pbuffer, sizeof(pbuffer), "%s.DB%s%d.OLD", ubik_dbase->pathName, (file<0)?"SYS":"", (file<0)?-file:file);
unlink(pbuffer);
#endif
#endif
if (code) {
#ifndef OLD_URECOVERY
unlink(pbuffer);
- /*
+ /*
* We will effectively invalidate the old data forever now.
* Unclear if we *should* but we do.
*/
inAddr.s_addr = ts->addr[0];
if (!ts->up) {
ubik_dprint("recovery cannot send version to %s\n",
- afs_inet_ntoa(inAddr.s_addr));
+ afs_inet_ntoa_r(inAddr.s_addr, hoststr));
dbok = 0;
continue;
}
ubik_dprint("recovery sending version to %s\n",
- afs_inet_ntoa(inAddr.s_addr));
+ afs_inet_ntoa_r(inAddr.s_addr, hoststr));
if (vcmp(ts->version, ubik_dbase->version) != 0) {
ubik_dprint("recovery stating local database\n");
}
/*!
- * \brief send a Probe to all the network address of this server
- *
+ * \brief send a Probe to all the network address of this server
+ *
* \return 0 if success, else return 1
*/
int
{
struct rx_connection *conns[UBIK_MAX_INTERFACE_ADDR];
struct rx_connection *connSuccess = 0;
- int i, j;
+ int i, j, success_i = -1;
afs_uint32 addr;
char buffer[32];
+ char hoststr[16];
extern afs_int32 ubikSecIndex;
extern struct rx_securityClass *ubikSecClass;
}
assert(i); /* at least one interface address for this server */
+#ifdef AFS_PTHREAD_ENV
+ DBRELE(ubik_dbase);
+#endif
+
multi_Rx(conns, i) {
multi_DISK_Probe();
if (!multi_error) { /* first success */
- addr = server->addr[multi_i]; /* successful interface addr */
-
- if (server->disk_rxcid) /* destroy existing conn */
- rx_DestroyConnection(server->disk_rxcid);
- if (server->vote_rxcid)
- rx_DestroyConnection(server->vote_rxcid);
-
- /* make new connections */
- server->disk_rxcid = conns[multi_i];
- server->vote_rxcid = rx_NewConnection(addr, ubik_callPortal, VOTE_SERVICE_ID, ubikSecClass, ubikSecIndex); /* for vote reqs */
-
- connSuccess = conns[multi_i];
- strcpy(buffer, (char *)afs_inet_ntoa(server->addr[0]));
- ubik_print
- ("ubik:server %s is back up: will be contacted through %s\n",
- buffer, afs_inet_ntoa(addr));
+ success_i = multi_i;
multi_Abort;
}
} multi_End_Ignore;
+#ifdef AFS_PTHREAD_ENV
+ DBHOLD(ubik_dbase);
+#endif
+
+ if (success_i >= 0) {
+ addr = server->addr[success_i]; /* successful interface addr */
+
+ if (server->disk_rxcid) /* destroy existing conn */
+ rx_DestroyConnection(server->disk_rxcid);
+ if (server->vote_rxcid)
+ rx_DestroyConnection(server->vote_rxcid);
+
+ /* make new connections */
+ server->disk_rxcid = conns[success_i];
+ server->vote_rxcid = rx_NewConnection(addr, ubik_callPortal,
+ VOTE_SERVICE_ID, ubikSecClass,
+ ubikSecIndex);
+
+ connSuccess = conns[success_i];
+ strcpy(buffer, afs_inet_ntoa_r(server->addr[0], hoststr));
+
+ ubik_print("ubik:server %s is back up: will be contacted through %s\n",
+ buffer, afs_inet_ntoa_r(addr, hoststr));
+ }
+
/* Destroy all connections except the one on which we succeeded */
for (j = 0; j < i; j++)
if (conns[j] != connSuccess)
if (!connSuccess)
ubik_dprint("ubik:server %s still down\n",
- afs_inet_ntoa(server->addr[0]));
+ afs_inet_ntoa_r(server->addr[0], hoststr));
if (connSuccess)
return 0; /* success */
git://git.openafs.org / openafs.git / blobdiff