[ejabberd] connection_dropped seen in ejabberd.log and sasl.log

Tom tom.valdes at gmail.com
Wed Jun 11 21:51:25 MSD 2008


Am I not giving enough details to the list to get help?  I can provide 
more logs if needed.
Also, I'm running version 2.0 and noticed 2.0.1 is out, but the 
changelog doesn't mention my error so I haven't upgraded my 
install..should I?


Tom wrote:
> Further testing over the weekend showed the same results as below.
> Any ideas?
> thanks,
> tom
>
> Tom wrote:
>> Jeffrey Rogiers wrote:
>>>
>>> This has to do with MySQL's keep alive timer, it kills connections 
>>> after 8 hours of inactivity, (you may want to look further to get 
>>> the exact time). But an easy way to fix it is to just tell ejabberd 
>>> to send keep alive queries more frequently.
>>>
>>> "You can configure an interval to make a dummy SQL request to keep 
>>> alive the connections to the database. The default value is 
>>> 'undefined', so no keepalive requests are made. Specify in seconds: 
>>> for example 28800 means 8 hours.
>>>
>>> {odbc_keepalive_interval, undefined}."
>>>  
>>> If your using ejabberd 2.X you can look at the manual.
>>> http://www.process-one.net/docs/ejabberd/guide_en.html#htoc28
>>>
>>> Thanks,
>>> Jeffrey Rogiers
>> I'm aware of the keepalive interval, but I think something else is 
>> going on since I'm seeing this behavior alot sooner than 8 hours.  
>> There is never a period of inactivity even close to 8 hours.. We 
>> pretty much see traffic around the clock.
>>
>> For example, in my sasl.log, every CRASH REPORT is associated with 
>> the connection_dropped..  Here's a snippet of a results of" grep 
>> CRASH  sasl.log"
>> =CRASH REPORT==== 15-Mar-2008::10:57:05 ===
>> =CRASH REPORT==== 15-Mar-2008::10:58:05 ===
>> =CRASH REPORT==== 15-Mar-2008::10:58:07 ===
>> =CRASH REPORT==== 15-Mar-2008::11:00:06 ===
>> =CRASH REPORT==== 15-Mar-2008::11:01:04 ===
>> =CRASH REPORT==== 15-Mar-2008::11:01:06 ===
>>
>> In the snippet above, the first two crashes are a minute apart and 
>> the third is after only two seconds.
>>
>> In the snippet below from ejabberd.log, the two crashes are about two 
>> minutes apart.
>>
>> =ERROR REPORT==== 2008-06-06 04:25:02 ===
>> ** Generic server <0.19802.552> terminating
>> ** Last message in was {'DOWN',#Ref<0.0.460.168032>,
>>                               process,
>>                               <0.19803.552>,
>>                               normal}
>> ** When Server state == {state,<0.19803.552>,mysql}
>> ** Reason for termination ==
>> ** connection_dropped
>>
>> =ERROR REPORT==== 2008-06-06 04:27:39 ===
>> ** Generic server <0.20275.552> terminating
>> ** Last message in was {'DOWN',#Ref<0.0.460.170162>,
>>                               process,
>>                               <0.20276.552>,
>>                               normal}
>> ** When Server state == {state,<0.20276.552>,mysql}
>> ** Reason for termination ==
>> ** connection_dropped
>>
>> thanks,
>> tom
>



More information about the ejabberd mailing list