SP metadata cache keeps growing

[Spencer Thomas]

We are switching to using the MDQ feeds for the UK and InCommon, which change significantly reduces startup time, because those metadata are downloaded as needed, rather than all at once.


--
Spencer Thomas
Technical Architect / JSTOR and Artstor
ITHAKA <https://www.ithaka.org/> / 301 E. Liberty St, Suite 250, Ann Arbor, MI 48104
Email: Spencer.Thomas at ithaka.org
Voicemail: 734-887-7004
 
 
 


On 2/27/20, 4:05 AM, "users on behalf of Alan Buxey" <users-bounces at shibboleth.net on behalf of alan.buxey at myunidays.com> wrote:

    hi,
    
    with the full firehose-feed of eduGAIN there's a lot of metadata to be
    checked.... probably looking at almost 10minutes
    - whilst the systemd stuff can be adjusted to deal with theres a
    quicker fix and that's to have a local instantiation
    of PyFF that filters out just the IdPs (the SP doesn't need to know
    all other SPs) - that'll slash the data size. and if
    PyFF is doing the heavy lifting and dropping resulting file to disk
    then Shibboleth can be configured to ONLY read
    that local disk file (and not verify/check it either as its already
    been checked with cert in the PyFF staging). pretty
    fast restarting of the shibd process then.
    verifyBackup="false"  - the on-disk file won't get checked - but the
    on disk one will only be used if the system cannot
    successfully download the remote metadata (which it can...and then checks)
    
    alan
    -- 
    For Consortium Member technical support, see https://wiki.shibboleth.net/confluence/x/coFAAg
    To unsubscribe from this list send an email to users-unsubscribe at shibboleth.net