Minutes from the 03/30/05 LCLS SLC IOC meeting:

(0) New Test IOCs:

    KenU agreed to add XL04 and XL05 to the SLC database for testing.  That 
    will give us two IOCs for RTEMS testing and for now, we'll assign XL04 to 
    cdvw4 and XL05 to cdvw5 and boot these as RTEMS.  (XL01,2 are solaris, 
    XL03 is linux).  Doug agreed to let Debbie have the borrowed SPEAR 
    Synergy CPU which they will set up as cdvw4 next week.  Perhaps soon we'll
    have a new mv5500 that we can set up as cdvw5 for SLC IOC testing.

(1) Database Service and Utilities:

    KenU has tested that the mccdev dbgen automatically puts the primary.map 
    file in the correct NFS dev directory.  When the dbinstall on mcc is done 
    in two weeks, he'll test that the file is put in the correct NFS prod
    directory.

    Debbie will change dbupdate to send both ST2 and ST3 data to DBEX.

    Debbie is done with dbSend including DBEX acknowledgement waits.
    Debbie tested every scenario she could think of where the acknowledgement
    isn't done or is corrupted and made sure the caller is notified of the
    bad status.

    Debbie added slcdbeditST1 iocsh routine for testing.

    We discussed that applications need to be smart about their dblput calls.
    If a dblput is doing multiple secondaries and only one secondary has 
    changed, ALL secondary values will be sent to DBEX (not just the one that
    changed).  Applications need to sort their secondaries into appropriate
    lists so that secondaries that don't change often are in separate
    lists from those that do and the application does multiple dblputs as 
    needed.  We could change dblput so that it checks for changes before
    updating the hi/lo table but the slc micros don't do this so it's
    possible we break something by adding this new feature.

    Tasks left for Debbie in this order:
    * finish dbupdate
    * add message logging
    * test on RTEMS
    * add any missing diagnostics (collision count, for one)
    * update documentation

    Estimate finish in mid-April unless RTEMS testing finds bugs.

(2) Message Service and General Utilities:

    Diane finished the slcExec changes proposed in last week's meeting 
    and task startup is reordered.  Did she add the SLC up/down messages?
    Debbie reported that msgRecv is still up after an slcStop.  Diane to 
    investiage after she's back from vacation.

    We discussed the assumption we made early in the design that slcExec
    cannot directly kill a task and deallocate that task's resources using
    a routine that the task registers with slcExec.  As a result, tasks
    without messages queues and tasks with message queues that may be busy
    processing a message for more than a few seconds need to periodically
    poll the stop flag that slcExec sets at about 1 Hz.  Kristi wants to
    investigate this assumption.

    Kristi is starting a list of additional helpful general utilities that
    she will propose to the group.
    
    Tasks left for Diane:
    * support others as they start using her utilities, plus consultation
    * diagnostics
    * correct thread priorities (waiting on requirements)

(3) Cluster Status Service and Async Utilities:

    RonM is currently coding cstrAsync and the slcAsync utilities called by it.
    He will have the 2 updated design documents out for review on Monday 
    morning. He has no issues.

    KenU agreed to add CSTR defaults for our test SLC IOCs to defaults.dbs
    and have XL01 to XL05 use it.  Stephanie sent him a first draft of the 
    default values.

    Tasks left for Ron:
    * update specs and send out for review
    * code cstrAsync and async utilities used only by cstrAsync
    * finish cstrHdlr and async utilities used only by cstrHdlr
    * code async utilities used by all Hdlr's

(4) Magnet Service:

    Kristi is updating her functional requirements (see minutes from last 
    week's meeting).  She is creating task lists for various areas associated 
    with the magnet job which we'll review as a group:
    * Minor VMS changes
    * Features that must be supported on the EPICS side
    * SLC database and displays that need changing/addition for LCLS

    Kristi wonders about the kicker magnet interface.  Will the interface
    be all EPICS or do we need to support it in the SLC magnet job?  Kristi
    mentions that the SCP side will need changes (maybe significant).  Who
    can answer this?

    Kristi is working on a prototype mgntHdlr.  She will have something
    running in 2 weeks.  Note that Kristi is not working on this full-time
    since she is responsible for all magnet issues for LCLS.

    Stephanie will provide a list of which side (EPICS or SLC) is the
    master and thus has control of a setpoint under various operating
    conditions.  The group will review and then the functional requirements
    updated.  This will help Kristi in her design.

(5) Digitizer Acquistion:

    Kristi mentioned that the SLC micro crate job was modified a long time
    ago for digitizer acquisition (not triggered on beam pulse).  We do
    not plan to support this in the SLC-aware IOC - Stephanie to update
    the functional requirements for this.

(6) CMLOG Message Logging:

    The CMLOG client still needs porting to RTEMS.  Can RonM do this after
    he finishes cstr and async?

(7) BPM Service:

    Can Debbie and Diane start this after they are finished with their cleanup?