ndb_redis: redis_cmd() doesn't return error with a faulty command

[linuxmaniac]

Description

Detected that a command like this was not returning <0

if(redis_cmd("test", "HSET dd field1", "r")) {
  [...]
}

Same command on redis-cli:

# redis-cli
127.0.0.1:6379> HSET dd field1
(error) ERR wrong number of arguments for 'hset' command

Troubleshooting

Reproduction

I've added some debug to the code and I find out that the command is returning:

Apr 23 10:55:02 spce proxy[12880]: DEBUG: ndb_redis [redis_client.c:956]: redisc_exec(): rpl->rplRedis->type:3

That's an INTEGER. From hiredis/read.h:

#define REDIS_REPLY_INTEGER 3
#define REDIS_REPLY_NIL 4
#define REDIS_REPLY_STATUS 5
#define REDIS_REPLY_ERROR 6

I've created a simple test code to check the same command:

#include <stdio.h>
#include <stdarg.h>
#include "hiredis.h"

void check(redisContext *c, const char *cmd, ...) {
        va_list ap;
        va_start(ap,cmd);
        redisReply *r = redisvCommand(c, cmd, ap);
        va_end(ap);

        if(r == NULL) {
                printf("reply NULL\n");
        }
        printf("reply->type:%d\n", r->type);
        freeReplyObject(r);
}

int main(void) {
        redisContext *c = redisConnect("127.0.0.1", 6379);

        if (c == NULL || c->err) {
            if (c) {
                printf("Error: %s\n", c->errstr);
            } else {
                printf("Can't allocate redis context\n");
            }
            return 1;
        }
        check(c, "HSET dd field");
        redisFree(c);
        return 0;
}

root@spce:/usr/local/devel# cc t.c -o t -lhiredis -I/usr/include/hiredis
root@spce:/usr/local/devel# ./t
reply->type:6

No problems there. The reply is ERROR.

Additional Information

• Kamailio Version - output of kamailio -v

version: kamailio 5.3.3 (x86_64/linux)
flags: USE_TCP, USE_TLS, USE_SCTP, TLS_HOOKS, USE_RAW_SOCKS, DISABLE_NAGLE, USE_MCAST, NO_SIG_DEBUG, DNS_IP_HACK, SHM_MMAP, PKG_MALLOC, Q_MALLOC, F_MALLOC, TLSF_MALLOC, DBG_SR_MEMORY, USE_FUTEX, FAST_LOCK-ADAPTIVE_WAIT, USE_DNS_CACHE, USE_DNS_FAILOVER, USE_NAPTR, USE_DST_BLACKLIST, HAVE_RESOLV_RES, TLS_PTHREAD_MUTEX_SHARED
ADAPTIVE_WAIT_LOOPS 1024, MAX_RECV_BUFFER_SIZE 262144, MAX_URI_SIZE 1024, BUF_SIZE 65535, DEFAULT PKG_SIZE 8MB
poll method support: poll, epoll_lt, epoll_et, sigio_rt, select.
id: unknown
compiled with gcc 8.3.0

[miconda]

I don't get exactly what you report here: is a problem with the kamailio code or a strange value for type field in the redis reply. Because that redis reply structure is what redisvCommand() returns. Any further details about the issue itself?

[linuxmaniac]

I would say our code is doing something wrong since the same command using redisvCommand() in my simple test code works as expected, returning an error reply type not an integer.

[linuxmaniac]

I will try to explain,

"HSET dd field" is missing the value parameter and it's a faulty command, it must fail.

redis_cmd() doesn't return <0 so there was no error 0_0 ( reply->type 3, INTEGER )
My simple code test fails as expected ( reply->type 6, ERROR )

[miconda]

First I was not sure that you report the redis_cmd() is not returning false (negative) in such case, or the type field has the wrong value.

Then we use the same hiredis function to get the response:

rpl->rplRedis = redisvCommand(rsrv->ctxRedis, cmd->s, ap );

Strange that gets different results on your system. Have you compiled the ndb_redis module where you did the test example, or is it from debian packages?

[linuxmaniac]

both are using the same library from buster: libhiredis-dev 0.14.0-3