From Renee, elog:
http://midtwist.triumf.ca:8081/EL/021209.5785?exp=twist

ODB got corrupted again. Actually, I tried recuperating the system without
rebuilding the ODB. It did work. 

What triggered it?  Glen wanted to change some variables in the history
panel "ramp". He removed the 1st entry and replaced it by the last
entry (having not deleted the last entry). The whole system froze. 

How I fixed it? 
- From browser, go to SlowLauncher (e614slow:8082) and stop all tasks. 
-    2  7:18    killall mlxspeaker
     3  7:18    killall mlogger
     4  7:18    killall mevb
     5  7:18    killall -KILL fedaq
     6  7:18    killall -9 mhttpd
    10  7:18    killall mserver
-  ps -efw | grep tk614
and kill -9 for all process using tk614
- ipcs 
   look for Shared Memory segments that have a non zero key value;
   these are the MIDAS shared sections. Check nattach for these. 
   They are not zero yet. Look for more MIDAS processes.  
-  ps -efw | grep twistonl 
   found also  mevb, odbedit and lazylogger 
   kill -9 these tasks. 
- ipcs - now "nattach" is zero for the MIDAS sections.
  remove the sections:
    36  7:24    ipcrm shm 19824642
    37  7:24    ipcrm shm 25755651
    38  7:24    ipcrm shm 25788420

- try odb command
  Now it works..... 
- go to SlowLauncher.
    Start all tasks.
- start_daq
and again
- start_daq