AWS Thinkbox Discussion Forums

Phantom slave appearing

Having a strange issue with beta 10, where I’ll get an incomplete phantom slave entry in my Slaves pane.

How I caused this:

  1. Submitted a dead-simple CommandScript job to a locally-hosted repository, which has one slave (my workstation). All tasks failed.
  2. Selected the job, then in the Tasks pane, selected one of the tasks.
  3. After some clicking around in the Tasks list, the magic blue slave appeared in my Slaves list.

I’m not sure if this is a coincidence, but the Tasks list shows the fully-qualified domain name of my machine in the Slave column (which matches what the phantom entry shows as its “Slave Name”). However, the “real” slave entry shows only the short name as the Slave Name, with the FQDN as the “Machine Name”.

I can reliably reproduce this by closing the monitor, reopening it, selecting the job again, and then clicking around in the Tasks list a few times. It seems that the entry will not show up unless I start interacting with the Tasks list.

Here’s the JSON dump from the SlaveInfo collection in the “deadline6db” db (I’m assuming that’s the current one, but I also have “deadlinedb”)

{ "Arch" : "x86_64", "BadJobs" : 0, "CPU" : 2, "Disk" : NumberLong("9701429248"), "DiskStr" : "9.035 GB ", "Grps" : "", "Host" : "ws-082.luma-pictures.com", "IP" : "100.100.104.82", "JobGrp" : "", "JobId" : "", "JobName" : "", "JobPlug" : "", "JobPool" : "", "JobPri" : -1, "JobUser" : "", "LastWriteTime" : ISODate("2013-01-30T00:42:07.836Z"), "Lic" : "", "LicEx" : -1, "LicFree" : true, "LicPerm" : false, "Limits" : [ ], "MAC" : "00:1D:4F:48:0F:4C", "Msg" : "2013/01/29 16:42:07 Slave shutting down", "Name" : "ws-082", "OS" : "Fedora release 15 (Lovelock)", "OnTskComp" : "Continue Running", "Pools" : "", "Port" : 52060, "ProcSpd" : NumberLong(2800), "Procs" : 8, "Pulse" : false, "RAM" : NumberLong("16923025408"), "RAMFree" : NumberLong("12886806528"), "RndTime" : 0, "Stat" : 3, "StatDate" : ISODate("2013-01-30T00:42:07.712Z"), "TskComp" : 0, "TskFail" : 4, "TskId" : "", "TskName" : "", "TskProg" : "", "TskStat" : "", "Up" : 252.46429443359375, "User" : "ruschn", "Ver" : "v6.0.0.49986 R", "Vid" : "NVIDIA Corporation GF100GL [Quadro 4000] (rev a3)", "_id" : "ws-082" }

One more thing is that if I go to the Job Properties and click on Machine Limit, the Slave List shows both the machine name and FQDN as separate entries.

Thanks! We’ve seen this happen too, but we couldn’t reliably reproduce, so the steps you provided will definitely be helpful!

It’s logged as a bug.

Privacy | Site terms | Cookie preferences