Meeting 10-23-2009 Installation Plans

ATTENDEES

Diane, Kukhee, Sonya, Kristi, Terri, Ernest, Sheng 

TOPICS: Installation

Need to know-

1)      the scope of the installation effort for each subsystem.

2)      How much can be done without beam, before April turn-on

Current Plan for April Start-up

• April 1st - begin 2 weeks of PPS
• April 19th - beam on
• April 19 through May 5 - beam tune-up
• May 6 - Users start

Possible Alternate Plan

• April 19th through September no 120Hz user experiments, only up to 30Hz
• April 19th through September testing at 120Hz through TD11 only
  

NETWORK 

add FCOM uplinks to

*- we will install this new switch swh-li21-nw01, move these iocs from swh-in20-nw02 to this switch

Network upgrade and test plan - to be done DURING downtime

• install all new fiber and new switches, leave old private switches in place for RF, BPMs, BLENs
• configure new switches for all BPMs, BLENs, RF EIOCs (set the EIOC ports to receive only multicast from their master IOC)
• THEN, for every IOC and EIOC involved in feedback,
  • test the new network connections by moving network cables to new switch and ping (add a couple lines in start-up file)
  • this will likely take several days and will involve both Network group and each subsystem engineer (Kristi, Sonya, Sheng, Kukhee)
  • when done testing an IOC / EIOC move connections BACK TO ORIGINAL SETUP. We must leave the system in its original state for start-up in April
  • QUESTION - do we need to disconnect the magnet IOCs, and UND BPM IOCs 2nd NIC when done testing? Definitely not the Controller IOCs.
• Terri will determine the overall schedule for this work, with input from Kristi, Sonya, Sheng, and Kukhee
  

 NOTE: Once beam is back up in April, the subsystems that now use private switches (RF, BPM, BLEN) will need to move cables from private switches to new FCOM switches when installing new software in production. Backout will include moving cables back.

Magnets

This involves two Magnet IOCs.  The new magnet sofware must be a single image that works in IOCs that control the new fast correctors, and IOCs that do not.

Before April the network connections will have been tested as described above. After start-up we must schedule time for the following installation plan:

• test new code in one IOC that controls new fast correctors.
• If it works, boot into both IOCs and leave it running about 1 week
• once they check out, boot the code into an IOC that does not control a new corrector, leave it running for about a week
• Finally, reboot all magnet IOCs with new code.

BPMs and BLENs

involves 18 BPM IOCs, and 2 BLEN IOCs.  Sonya pointed out that the undulator BPM IOCs do not have EIOCs so their network setups will be more like the Magnet IOCs ( NIC1 for Channel Access, NIC2 for FCOM).  I don't know if these different BPM types use different IOC images, but the goal should be to make a single image for IOCs used by feedback and those not. This is not an issue with BLENs - both BLEN IOCs are used by feedback.

Before April the network connections/switch configurations will have been tested as described above. After start-up we must schedule time for the following installation plan:

• BLENs
  • One MD day to install and test both BLEN IOCs - let them run if it checks out ok
  • NIC 2 cabling must be switched from private switch to the new FCOM network switch at the same time the software is installed.
• BPMs
  • Sonya, Till - I did not write down a specific plan for the BPM IOCs. Do you expect to need a couple weeks of check-out time, similar to Kristi's plan above? Please let me know what's required after start-up.

Controller IOCs

3 controller IOCs.  We should be able to install the Controller IOCs during the downtime, or really anytime before April. They may be useful for production network testing during the downtime (we could run Till's tests in production maybe?) - Terri and Ernest can plan to install the Controller IOCs with the network installations. 

After start-up

• we'll need an MD day per feedback loop to install, configure, and test. Maybe a couple days for the first feedback.

BLD IOC

A version 2 of the BLD IOC needs to be installed once we have three feedbacks, 1) Bunch Charge Feedback, the 2) Longitudinal Feedback, and the 3) LTU Launch Feedback, running and providing their States data over FCOM. This will not likely happen until later in the summer. We discussed that this IOCs NICs will be dedicated to 1) FCOM and 2)Photon Multicast network - leaving no NIC for Channel Access. Sheng to think about whether having Channel Access is necessary, and alternative ways to monitor this VME if there is no channel access.

After the three feedbacks are installed:

• one MD day, plus coordinate with photon group,  to install version 2 BLD and test.
  

RF IOCs

We did not get to this topic during the 1+ hours meeting. Diane, Kukhee, Ernest and Terri to meet again (Monday?) to continue.

ACTION ITEMS

• Terri to make schedule for Network installation and test during downtime, with input from Sonya, Kristi, Kukhee, Sheng
• Terri to add Controller IOC installation, with input from Ernest
•  Sonya, Till - detailed plan for BPM check-out after April start-up.
• Sheng - develop a plan for monitoring status of this BLD version 2  'concentrator' processor
• Kristi - which two magnet IOCs control the 4 new correctors? ltu0-mg01, ltu1-mg01
• Diane, Kukhee, Ernest, Terri - meet to plan installation of RF IOCs