java - Is it a bad practise to pass jedis instance between methods?

Question

I was referring to this SO question, and I did couple of additions in this benchmark test. The main problem is my apis are getting slow as load increases on server. I am using jedis pool configuration.

// get a new instance
    public synchronized Jedis getJedi() {
    try {
        return jedisPool.getResource();
    } catch (Exception e) {
        log.fatal("REDIS CONN ERR:", e);
        return null;
    }
    }

// intialize at start
    public void initialize() {
    if (jedisPool == null) {
        IniUtils cp = PropertyReader.getConnPoolIni();

        String host = cp.get(REDIS, REDIS_HOST);
        int port = Integer.parseInt(cp.get(REDIS, REDIS_PORT));
        String password = cp.get(REDIS, REDIS_PASSWORD);
        int timeout = Integer.parseInt(cp.get(REDIS, REDIS_TIMEOUT));

        JedisPoolConfig poolConfig = new JedisPoolConfig();
        poolConfig.setMaxTotal(Integer.parseInt(cp.get(REDIS, REDIS_MAX_TOTAL_CONNECTIONS)));
        poolConfig.setMaxIdle(Integer.parseInt(cp.get(REDIS, REDIS_MAX_IDLE)));
        poolConfig.setMinIdle(Integer.parseInt(cp.get(REDIS, REDIS_MIN_IDLE)));
        poolConfig.setMaxWaitMillis(Long.parseLong(cp.get(REDIS, REDIS_MAX_WAIT_TIME_MILLIS)));

        poolConfig.setTestOnBorrow(true);
        poolConfig.setTestOnReturn(true);
        poolConfig.setTestWhileIdle(true);

        if (password != null && !password.trim().isEmpty()) {
        jedisPool = new JedisPool(poolConfig, host, port, timeout, password);
        } else {
        jedisPool = new JedisPool(poolConfig, host, port, timeout);
        }

        test();

      }
    }

    @Override
    public void destroy() {
      if (jedisPool.isClosed() == false)
          jedisPool.destroy();
    }

    private void test() {
      try (Jedis test = getJedi()) {
        log.info("Testing Redis:" + test.ping());
      }
    }

And while using, I get Jedis instance in try-with-resources and works on it. I use very less pipelining and there are various calls to Redis, so each time a method call, a new jedis instance gets created.

As per SO question shared, my implementation will lead to very slow results. So, can I pass around Jedis instance to methods and work with pipeline as per business logic. Something like this -

  public void push5(int n) {
    try (Jedis jedi = redisFactory.getJedi()) {
        pushWithResource(n, jedi, 0);
    }
  }

  public void pushWithResourceAndPipe(int n, Jedis jedi, int k) {
    if (k >= n)
        return;
    Pipeline pipeline = jedi.pipelined();
    map.put("id", "" + i);
    map.put("name", "lyj" + i);
    pipeline.hmset("m" + i, map);
    ++i;
    pushWithResourceAndPipe(n, jedi, ++k);
    pipeline.sync();
  }

    public void pushWithResource(int n, Jedis jedi, int k) {
    if (k >= n)
        return;
    map.put("id", "" + i);
    map.put("name", "lyj" + i);
    jedi.hmset("m" + i, map);
    ++i;
    pushWithResource(n, jedi, ++k);
  }

Is there any way to improve on api calls? Could you recommend some projects which uses jedis on server side, so I will have a better understanding on how to use jedis effectively.

Redis / Jedis Configuration

Jedis version:2.8.1 Redis version:2.8.4 Java version:1.8