Zabbix trapper items, trigger issue

Problem: item get triggered, and alert sent, when threshold is reached. But, it get recovered within seconds despite hysteresis.
Since data are collected every minutes using trappers, we got an alert+recovery every minute.

Zabbix version: 2.4.6 for both server & agent

Item is discovered through LLD. Here's the actual LLD paylod:

    {
    "{#RMQQUEUENAME}": "queue_name",
    "{#RMQRATIOTHRES}": 5,
    "{#RMQVHOSTNAME}": "queue_vhost",
    "{#RMQMSGTHRESH}": 3500
    }

Trigger prototype is as fallow:

    (
    {TRIGGER.VALUE}=0 and {BBC_TPL_RABBITMQ:rabbitmq.queue[{#RMQVHOSTNAME},{#RMQQUEUENAME},master].last()}=1 and {BBC_TPL_RABBITMQ:rabbitmq.queue[{#RMQVHOSTNAME},{#RMQQUEUENAME},count,message].min(5m)}>({#RMQMSGTHRESH}*10)
    ) or (
    {TRIGGER.VALUE}=1 and {BBC_TPL_RABBITMQ:rabbitmq.queue[{#RMQVHOSTNAME},{#RMQQUEUENAME},count,message].max(10m)}<{#RMQMSGTHRESH}
    )

Actual trigger on hostname is:

    (
    {TRIGGER.VALUE}=0 and {real_hostname:rabbitmq.queue[queue_vhost,queue_name,master].last()}=1 and {real_hostname:rabbitmq.queue[queue_vhost,queue_name,count,message].min(5m)}>(3500*10)
    ) or (
    {TRIGGER.VALUE}=1 and {real_hostname:rabbitmq.queue[queue_vhost,queue_name,count,message].max(10m)}<3500
    )


Raw items values are as follow. As you can see, we should never have the flapping effect :-/

|-----------|------------------|----------|----|
| itemid   | clock              | value   | ns |
|-----------|------------------|----------|----|
| 127424 | 1441992304 |  10138 |  0 |
| 127424 | 1441992365 |  12576 |  0 |
| 127424 | 1441992426 |  17052 |  0 |
| 127424 | 1441992487 |  21800 |  0 |
| 127424 | 1441992548 |  26441 |  0 |
| 127424 | 1441992609 |  30759 |  0 |
| 127424 | 1441992670 |  35482 |  0 |
| 127424 | 1441992731 |  39735 |  0 |
| 127424 | 1441992792 |  44115 |  0 |
| 127424 | 1441992853 |  48654 |  0 |
| 127424 | 1441992914 |  52902 |  0 |
| 127424 | 1441992975 |  57206 |  0 |
| 127424 | 1441993036 |  61052 |  0 |
| 127424 | 1441993097 |  65225 |  0 |
| 127424 | 1441993158 |  69401 |  0 |
| 127424 | 1441993219 |  73779 |  0 |
| 127424 | 1441993280 |  78420 |  0 |
| 127424 | 1441993341 |  82797 |  0 |
| 127424 | 1441993402 |  86617 |  0 |
| 127424 | 1441993463 |  91222 |  0 |
| 127424 | 1441993524 |  95373 |  0 |
| 127424 | 1441993585 |  99768 |  0 |
| 127424 | 1441993646 | 104122 |  0 |
| 127424 | 1441993707 | 108785 |  0 |
| 127424 | 1441993768 | 112611 |  0 |
| 127424 | 1441993829 | 116940 |  0 |
| 127424 | 1441993890 | 121295 |  0 |
| 127424 | 1441993951 | 125623 |  0 |
| 127424 | 1441994012 | 130242 |  0 |
| 127424 | 1441994073 | 134281 |  0 |
| 127424 | 1441994135 | 138360 |  0 |
| 127424 | 1441994196 | 142716 |  0 |
| 127424 | 1441994257 | 119110 |  0 |
| 127424 | 1441994319 |  88993 |  0 |
| 127424 | 1441994380 |  58073 |  0 |
| 127424 | 1441994441 |  25217 |  0 |

Here are some example of events:

2015-09-11 22:55:50 real_hostname   We have too many messages in queue_vhost:queue_name OK  Average 15h 44m 50s No  Ok
2015-09-11 22:55:50 real_hostname   We have too many messages in queue_vhost:queue_name PROBLEM Average 0   No  12  5
2015-09-11 22:54:48 real_hostname   We have too many messages in queue_vhost:queue_name OK  Average 1m 2s   No  Ok
2015-09-11 22:54:48 real_hostname   We have too many messages in queue_vhost:queue_name PROBLEM Average 0   No  12  5
2015-09-11 22:53:47 real_hostname   We have too many messages in queue_vhost:queue_name OK  Average 1m 1s   No  Ok
2015-09-11 22:53:47 real_hostname   We have too many messages in queue_vhost:queue_name PROBLEM Average 0   No  12  5
2015-09-11 22:52:46 real_hostname   We have too many messages in queue_vhost:queue_name OK  Average 1m 1s   No  Ok
2015-09-11 22:52:46 real_hostname   We have too many messages in queue_vhost:queue_name PROBLEM Average 0   No  12  5
2015-09-11 19:59:40 real_hostname   We have too many messages in queue_vhost:queue_name OK  Average 2h 53m 6s   No  Ok
2015-09-11 19:59:40 real_hostname   We have too many messages in queue_vhost:queue_name PROBLEM Average 0   No  12  5