Hello,
We are upgrading our servers to Kamailio 4.2 and we see a lot of error messages "Failed to delete subscription from memory".
Scenario: This happens every time there is a Notify Timeout (408) and Kamailio tries to delete the subscription.
Consequences: I don't know whether the delete fails because the entry did never exist, was already deleted or because the comparison (to_tag, callid or from_tag) is incorrect. It would be nice to at least have some details in the log message. I have not noticed any consequences to general functionality.
I suspect the commit (74d73953327ad7702922b27ba37c71e3283659c3) to be responsible for this error because it introduced some extra checks on deletion.
Does this happen to anybody else? What may be configured incorrectly on our system?
Kind regards
*Jan **Gaida* Ingeniero Desarrollo Software C/ Marconi 3 (PTM) 28760 Tres Cantos Spain jan.gaida@grupoamper.com | www.grupoamper.com
Has nobody ever seen something similar?
*Jan **Gaida* Ingeniero Desarrollo Software C/ Marconi 3 (PTM) 28760 Tres Cantos Spain jan.gaida@grupoamper.com | www.grupoamper.com
On 24 November 2014 at 11:52, Jan Gaida jan.gaida@grupoamper.com wrote:
Hello,
I will try to look asap to the code and at least add more debugging info to be easier to understand what happens in case all works fine at the end.
Cheers, Daniel
On 02/12/14 11:34, Jan Gaida wrote:
Hello,
the code seems to be similar to the one for searching a subscription. I am considering to make the matching criteria configurable to see the results.
Are you using presence for dialog notifications (e.g., blinking lamps, call pickup) or user presence states?
Cheers, Daniel
On 02/12/14 14:15, Daniel-Constantin Mierla wrote:
Can you try with the following commit:
- https://github.com/kamailio/kamailio/commit/85029d414384912879b45a34bdeb8f13...
If you installed from git, you can do a:
git pull origin git cherry-pick -x 85029d414384912879b45a34bdeb8f1372cc3780
Then re-compile, re-install and restart kamailio.
Let me know the results in order to decide whether to backport the patch or not.
Cheers, Daniel
On 03/01/15 10:48, Daniel-Constantin Mierla wrote: