From DCT Science Report 20150415UT-C:
RC1/RC2/CAT/JOE/DeVeny:
1) 03:48 ScienceNote Just before Audrey was done with sky flats, Peter sent an email saying that it looked like some of the arc lamps were still on. We checked into this by taking some images on DeVeny & indeed it seemed that one was still on somehow even after I had gotten some flat images after doing some testing with the PDU connections earlier. I used the Joe Client PDU connection & the get status button didn't do anything, so I wasn't sure which lamp(s) was/were on. Without really thinking about the other outlets & the chiller being connected to one of them(they are also not labelled), I figured the easiest way to turn them off was the All off button(all while wanting to start the pointing check to get further evidence for or against an tangent pin being broken). Hence, the chiller was turned off & it didn't occur to me until Peter reminded me of the DeVeny CCD temp while I was staring at the PDU tab on Joe Client what had happened with Stephen on the phone as well.
I turned all the outlets on at this point & the ccd cooled back down nicely. After a Joe restart we took more images & used the Deveny Loui PDU control to turn the lamps off & make sure they were off.
It seems that we are having lots of problems with the DeVeny PDU and there is some confusion about how things work so I am going to write this rather long email here explaining the way it currently works.
When joe starts, it looks at the config file and tries to make a socket connection to any PDUs that are defined in the config file. Currently we only have the DeVeny PDU in the config file. joe's attempt at the socket connection will fail if the web interface to the PDU is up and running. The PDU allows one socket connection at a time.
So if the web interface is up and one wants to instead use the LOUIs for PDU communication, the web interface needs to be closed. It usually takes 30 seconds for the socket to be released so one should wait at least 30 seconds before starting joe. joe makes a socket connection but there is no login to the PDU yet. One can either login from the DeVeny or JOE Client not both. One login is allowed at a time. The login has to be successful before the button labels are populated with the right labels. If the login is not successful, the buttons look to be enabled but they will not be populated with the right labels. In other words if you don't see the labels correctly, there is no real communication with the PDU and that specific widget is not operational.
There is a timestamp at the bottom of the widget and if it turns red, it means that joe has lost the socket connection to the PDU. But being green does not mean that the widget can control the PDU. I am in the process of redesigning the code so it will be more robust and also potentially allow multiple points of access or monitoring of the PDU status. This will hopefully be ready by the next engineering run.
I ran a test earlier this morning at DCT and every thing worked. I first logged into the web interface and got the following after logging in. Please not that the buttons on the web interface do not show the ON/OFF state. In order the see the socket state, you need to bring up the "Text Interface" first and then click the "Get Status" button. In this case you'll see that the 4 lamps are off and all the other sockets are on.
I then closed the web interface but quitting safari and making sure that the applet is not running either. I waited ~ 30 seconds and brought up joe and joe Client and this is what I saw after logging in:
You see that the state of the buttons are obvious by whether the button has a blue background or not and that all the labels are updated. If the buttons had shown OUTLET-1 through OUTLET-8 as the label, it would have meant that I do not have a good connection. At this point, I did not bother with the DeVeny PDU because I knew that JOE Client would have pre-empted any other connections.