[Spce-user] URGENT: mysql database ist growing until disk is full (mr6.5.4.1)

Daniel Grotti dgrotti at sipwise.com
Fri May 10 07:22:16 EDT 2019

The reason of the error message is that some of those ACC entry does not 
have the required branch_id field that mediator requires to created the CDR.

Did you upgrade the system to 6.5.4 recently?
Do you have the following lines in your

[Optima]root at sp1:/usr/lib/ngcp-sems/audio/it# grep "branch_id" 
modparam("db_redis", "keys", 
modparam("acc", "multi_leg_info", 

If not, it seems you have al older version of kamailio templates, so you 
should update the to latest 6.5.4.x

Once you have that lines in place, what I suggest, to fix the ERROR 
messages, is to :

1. stop mediator
2. install the previous version of ngcp-mediator (who does not 
have the branch_id patch inside), and let it run in order to create CDRs 
for those ACCs.
3. re update nediator to latest package (apt-get update && apt-get 
install ngcp-mediator)

After that you should not see the error related to "branch_id" anymore.

Regarding the disk usage, I cannot say why acc_trash is so big.

Daniel Grotti

Head of Customer Support                     Sipwise GmbH
e: dgrotti at sipwise.com                     Europaring F15
t: +43(0)130120332                A-2345 Brunn Am Gebirge
w: www.sipwise.com   FN: 305595f   FG: LG Wiener Neustadt

On 5/10/19 1:15 PM, Alex Lutay wrote:
> Hi,
> There is a process mediator which converts accounting
> into CDRs. Accounting are stored into Redis, and for some
> reason you have a garbage? there. It should be investigated
> and fixed, as it is the origin of the problem:
>>> May 10 07:24:04 spce (warning) mediator[5130]: medredis.c:303
>>> [medredis_check_reply_string]: Received Redis reply type 4 instead of 1
>>> (string) for branch_id field of cid '3218c40fc8dc84f2' using key
>>> 'acc:entry::3218c40fc8dc84f2:1556875485.599718'
> Meanwhile all accounting which were not converted into CDRs are stored
> into acc_trash for further investigation. I would say you have too much
> of them. As a fast solution for weekend and I hint you to dump/backup
> acc_trash (just in case) and truncate the table to have a time with
> investigation of mediator<->redis issue.
> On 5/10/19 12:55 PM, Thomas Georg wrote:
>> -rw-rw----  1 mysql mysql  62G May 10 12:41 acc_trash.ibd
>> It seems that kamilio is eating all of our disk-space.
>> Why do we get as much  of „trash“? How can we stop that and delete the
>> „trash“?

More information about the Spce-user mailing list