The new compute cluster is beginning to feel like a production system. I’m currently run off my feet installing software for the stream of new users. Mostly this is fine, but occasionally I run into software that makes me want to band my head repeatedly on my desk until the pain goes away; or more accurately makes me want to bang the programmer’s head on the desk.<\/p>\n
Just today we received a linux port of a code that has been running on the Windows Condor pool for a while now. Everything seemed fine except for it’s stubborn refusal to run if it couldn’t find a windowing system. Bear in mind that it doesn’t actually produce any graphical output it just dies if it can’t connect to X. After a bit of futzing around we discover that the people that normally run this code do something like:<\/p>\n
\r\nXvfb :1 -server 1 1024x1024x8 &\r\nexport DISPLAY=:1\r\n.\/stupid_code_that_wants_X\r\n<\/pre>\nXvfb is the X virtual framebuffer. It creates a running X client without actually needing any graphics to be running.<\/p>\n
Which works just great locally but if you want to launch that as a script in the job scheduling system (we use PBSpro) then you need to be a bit more careful. What happens if two of these jobs try to launch on the same machine? Obviously one of them will fail because display 1 is already allocated. What I really needed was a script that will try to launch Xvfb and increment DISPLAY on failure until it finds a display that is free. For your edification here it is:<\/p>\n
\r\nget_xvfb_pid () {\r\n\tXVFB_PID=`ps -efww | grep -v grep | grep Xvfb |\\\r\n grep $USERNAME | tail -n 1 | awk '{print $2}'`\r\n\t}\r\n\r\ncreate_xvfb () {\r\n\tUSERNAME=`whoami`\r\n\tDISPLAYNO=1\r\n\twhile [ -z $xvfb_success ]\r\n\t\tdo\r\n\t\tget_xvfb_pid\r\n\t\told_XVFB_PID=$XVFB_PID\r\n\t\tXVFB_PID=""\r\n\t\tXvfb :${DISPLAYNO} -screen 0 1024x1024x8 >& \/dev\/null &\r\n\t\tsleep 1\r\n\t\tget_xvfb_pid\r\n\t\tif ! [ -z $old_XVFB_PID ]\r\n\t\t\tthen\r\n\t\t\tif [ -z $XFVB_PID ] && ! [ $XVFB_PID == $old_XVFB_PID ]\r\n\t\t\t\tthen\r\n\t\t\t\techo "Started XVFB on display $DISPLAYNO process $XVFB_PID"\r\n\t\t\t\txvfb_success=1\r\n\t\t\telse\r\n\t\t\t\tDISPLAYNO=$(($DISPLAYNO + 1))\r\n\t\t\t\tXVFB_PID=""\r\n\t\t\tfi\r\n\t\telse\r\n\t\t\tif [ -z $XFVB_PID ]\r\n then\r\n echo "Started XVFB on display $DISPLAYNO process $XVFB_PID"\r\n xvfb_success=1\r\n else\r\n DISPLAYNO=$(($DISPLAYNO + 1))\r\n echo "FAIL!" $XVFB_PID\r\n XVFB_PID=""\r\n fi\r\n\t\tfi\r\n \t\tdone\r\n\texport XVFB_PID\r\n\texport DISPLAY=:${DISPLAYNO}\r\n\t}\r\n\r\nkill_xvfb () {\r\n\tkill $XVFB_PID\r\n\t}\r\n<\/pre>\nWhich you can call from a script like thus:<\/p>\n
\r\n[arccacluster8]$. .\/xvfb_helper\r\n[arccacluster8]$ create_xvfb\r\nStarted XVFB on display 1 process 9563\r\n[arccacluster8 ~]$ echo $DISPLAY\r\n:1\r\n[arccacluster8 ~]$ echo $XVFB_PID\r\n9563\r\n[arccacluster8 ~]$ ps -efw | grep Xvfb\r\nusername 9563 9498 0 19:31 pts\/8 00:00:00 Xvfb :1 -screen 0 1024x1024x8\r\n[arccacluster8 ~]$ kill_xvfb\r\n[arccacluster8 ~]$ ps -efw | grep Xvfb\r\n[arccacluster8 ~]$\r\n<\/pre>\nI submit that this is a disgraceful hack, but it might come in handy to someone else.<\/p>\n","protected":false},"excerpt":{"rendered":"
The new compute cluster is beginning to feel like a production system. I’m currently run off my feet installing software for the stream of new users. Mostly this is fine, but occasionally I run into software that makes me want to band my head repeatedly on my desk until the pain goes away; or more … Continue reading Dealing With Stupid Programs That Think They Need X<\/span>