[asterisk-bugs] [JIRA] (ASTERISK-25811) Unable to delete object from sorcery cache

[Ross Beer (JIRA)]

    [ https://issues.asterisk.org/jira/browse/ASTERISK-25811?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=229727#comment-229727 ] 

Ross Beer commented on ASTERISK-25811:
--------------------------------------

Ok, the original issues appears to be resolved. However I have now got another interesting issue with 'stale' entries.

When an entry becomes stale after 3600 seconds, the object is removed from cache which causes a phone de-registration.

Sourcery configuration is as follows:

{quote}
[res_pjsip] ; 
auth/cache=memory_cache,object_lifetime_stale=3600,object_lifetime_maximum=28800,expire_on_reload=yes,full_backend_cache=yes
auth=config,pjsip.conf,criteria=type=auth
auth=realtime,ps_auths
 
aor/cache=memory_cache,object_lifetime_stale=3600,object_lifetime_maximum=28800,expire_on_reload=yes,full_backend_cache=yes
aor=config,pjsip.conf,criteria=type=aor
aor=realtime,ps_aors
 
domain_alias/cache=memory_cache,object_lifetime_stale=3600,object_lifetime_maximum=28800,expire_on_reload=yes,full_backend_cache=yes
domain_alias=config,pjsip.conff,criteria=type=domain_alias
domain_alias=realtime,ps_domain_aliases
 
endpoint/cache=memory_cache,object_lifetime_stale=3600,object_lifetime_maximum=28800,expire_on_reload=yes,full_backend_cache=yes
endpoint=config,pjsip.conf,criteria=type=endpoint
endpoint=realtime,ps_endpoints
 
contact=realtime,ps_contacts
{quote}

Exactly at the point of the 'stale' marking each device states 'Removed contact' and 'Endpoint is now Unreachable' messages are outputted to the CLI.

Removing the 'object_lifetime_stale=3600,object_lifetime_maximum=28800' options resolves the issue.

> Unable to delete object from sorcery cache
> ------------------------------------------
>
>                 Key: ASTERISK-25811
>                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-25811
>             Project: Asterisk
>          Issue Type: Bug
>      Security Level: None
>          Components: Core/Sorcery
>    Affects Versions: 13.7.2
>         Environment: Centos 7
>            Reporter: Ross Beer
>            Assignee: Unassigned
>
> Sorcery cache throws an error on unregister:
> {quote}
> [2016-02-25 13:47:02] ERROR[17353]: res_sorcery_memory_cache.c:1559 sorcery_memory_cache_delete: Unable to delete object '<ENDPOINT>;@115bb1375dae1799c68048e7abef7e05' from sorcery cache
>  Contact <ENDPOINT>/sip:<ENDPOINT>@<IP Address>:39212;transport=TLS has been deleted
>     -- Added contact 'sip:<ENDPOINT>@<IP Address>:39212;transport=TLS' to AOR '<ENDPOINT>' with expiration of 60 seconds
>  Contact <ENDPOINT>/sip:<ENDPOINT>@<IP Address>:39212;transport=TLS has been created
>  Contact <ENDPOINT>/sip:<ENDPOINT>@<IP Address>:39212;transport=TLS is now Unknown.  RTT: 0.000 msec
> {quote}
> The device had previously registered and therefore the device should have been in the cache. When this issue occurs it blocks the registration of an endpoint causing it to go offline. I can replicate this issue by pressing 'Re-register' within the Snom interface.
> Sourcery is configured in the following way:
> {quote}
> contact/cache=memory_cache,object_lifetime_stale=3600,object_lifetime_maximum=28800,expire_on_reload=yes,full_backend_cache=yes
> contact=config,pjsip.conf,criteria=type=contact
> contact=realtime,ps_contacts
> {quote}
> Looking at the real-time database the ID is present, all be it encoded with <ENDPOINT>^3B which is the encoding for a semicolon. Therefore it looks like the cache isn't matching the object correctly or not being inserted in the first place. 



--
This message was sent by Atlassian JIRA
(v6.2#6252)