It happened again this week, here what we have done:

1. close the gatevalve
2. stop the turbo
3. stop the roughing
4. put the cryostat and the roughing back to atmosphere
5. restart the roughing

Then we wanted to restart the pump down so we needed to open the gatevalve.
But we couldn't do it, so I've power cycle the interlock and then it worked.

Is that a normal behaviour ? If the pressure is at atmosphere in the cryostat, we shouldn't have an interlock right ?

Are you describing a logic or a physical problem? If the cryostat is at atmosphere and the roughing line is pumped down, the gate valve will not and should not open.

I belatedly looked at the logs. I had hoped for a user error (sorry!), but we do need to try to duplicate this at JHU: the behavior is wrong. Specifically, the gatevalve was closed before the turbo was stopped. Having the gatevalve open when the turbo spins down through 75% is the only thing (besides pulling the jumper) which should trigger the interlock. shope [This was on r0, BTW: the thermal cryostat]

2017-05-16 13:31:37.487Z cmds             20 Actor.py:425 new cmd: gatevalve open
2017-05-16 13:31:37.520Z root             20 gatevalve.py:85 port 0 read: 0xf2
2017-05-16 13:31:37.520Z cmds             20 CommandLink.py:118 > 2 99 i gatevalve=0xf2,closed,closed
2017-05-16 13:31:37.521Z root             20 gatevalve.py:64 port 0 read: 0xf2
2017-05-16 13:31:37.521Z root             20 gatevalve.py:64 setting port 0 to 0x08
2017-05-16 13:31:37.522Z root             20 gatevalve.py:64 port 0 read: 0xfa
2017-05-16 13:31:37.522Z root             20 gatevalve.py:85 port 0 read: 0xfa
2017-05-16 13:31:37.522Z cmds             20 CommandLink.py:118 > 2 99 i gatevalve=0xfa,closed,blocked

That said, any new xcuActor (1.7.11+) would have rejected your request: with the roughing pump on and the cryostat at atmosphere the pressure differential across the valve would be too high.

New interlock board will address latching logic.