Whitelist SIGUSR1 to avoid auto-triggering errors.

This commit fixes issue #875 that was caused by the following events:

1) There is an active child doing BGSAVE.
2) flushall is called (or any other condition that makes Redis killing
the saving child process).
3) An error is sensed by Redis as the child exited with an error (killed
by a singal), that stops accepting write commands until a BGSAVE happens
to be executed with success.

Whitelisting SIGUSR1 and making sure Redis always uses this signal in
order to kill its own children fixes the issue.
This commit is contained in:
antirez 2013-01-14 10:29:14 +01:00
parent 1e20c939fe
commit 39f0a33f78
4 changed files with 8 additions and 5 deletions

View File

@ -177,7 +177,7 @@ void stopAppendOnly(void) {
redisLog(REDIS_NOTICE,"Killing running AOF rewrite child: %ld", redisLog(REDIS_NOTICE,"Killing running AOF rewrite child: %ld",
(long) server.aof_child_pid); (long) server.aof_child_pid);
if (kill(server.aof_child_pid,SIGKILL) != -1) if (kill(server.aof_child_pid,SIGUSR1) != -1)
wait3(&statloc,0,NULL); wait3(&statloc,0,NULL);
/* reset the buffer accumulating changes while the child saves */ /* reset the buffer accumulating changes while the child saves */
aofRewriteBufferReset(); aofRewriteBufferReset();

View File

@ -219,7 +219,7 @@ void flushallCommand(redisClient *c) {
server.dirty += emptyDb(); server.dirty += emptyDb();
addReply(c,shared.ok); addReply(c,shared.ok);
if (server.rdb_child_pid != -1) { if (server.rdb_child_pid != -1) {
kill(server.rdb_child_pid,SIGKILL); kill(server.rdb_child_pid,SIGUSR1);
rdbRemoveTempFile(server.rdb_child_pid); rdbRemoveTempFile(server.rdb_child_pid);
} }
if (server.saveparamslen > 0) { if (server.saveparamslen > 0) {

View File

@ -1194,6 +1194,9 @@ void backgroundSaveDoneHandler(int exitcode, int bysignal) {
redisLog(REDIS_WARNING, redisLog(REDIS_WARNING,
"Background saving terminated by signal %d", bysignal); "Background saving terminated by signal %d", bysignal);
rdbRemoveTempFile(server.rdb_child_pid); rdbRemoveTempFile(server.rdb_child_pid);
/* SIGUSR1 is whitelisted, so we have a way to kill a child without
* tirggering an error conditon. */
if (bysignal != SIGUSR1)
server.lastbgsave_status = REDIS_ERR; server.lastbgsave_status = REDIS_ERR;
} }
server.rdb_child_pid = -1; server.rdb_child_pid = -1;

View File

@ -1695,7 +1695,7 @@ int prepareForShutdown(int flags) {
overwrite the synchronous saving did by SHUTDOWN. */ overwrite the synchronous saving did by SHUTDOWN. */
if (server.rdb_child_pid != -1) { if (server.rdb_child_pid != -1) {
redisLog(REDIS_WARNING,"There is a child saving an .rdb. Killing it!"); redisLog(REDIS_WARNING,"There is a child saving an .rdb. Killing it!");
kill(server.rdb_child_pid,SIGKILL); kill(server.rdb_child_pid,SIGUSR1);
rdbRemoveTempFile(server.rdb_child_pid); rdbRemoveTempFile(server.rdb_child_pid);
} }
if (server.aof_state != REDIS_AOF_OFF) { if (server.aof_state != REDIS_AOF_OFF) {
@ -1704,7 +1704,7 @@ int prepareForShutdown(int flags) {
if (server.aof_child_pid != -1) { if (server.aof_child_pid != -1) {
redisLog(REDIS_WARNING, redisLog(REDIS_WARNING,
"There is a child rewriting the AOF. Killing it!"); "There is a child rewriting the AOF. Killing it!");
kill(server.aof_child_pid,SIGKILL); kill(server.aof_child_pid,SIGUSR1);
} }
/* Append only file: fsync() the AOF and exit */ /* Append only file: fsync() the AOF and exit */
redisLog(REDIS_NOTICE,"Calling fsync() on the AOF file."); redisLog(REDIS_NOTICE,"Calling fsync() on the AOF file.");