We´re running a view AIX 6.1 (VIO) machines with Networker 7.6SP1 and NPIV enabled. Every time we reboot a machine the Networker command cfglus -2 -l lus was invoked with the problem that the boot time was lengthen up to 1 hour or longer. That happens on machines with tape drive as well as on machines without tape drives connected. Any experience, hints or workarounds for that consideration?
Thanks in Advance.
Wonder if it could possibly be the lus driver taking its time going through the rounds on the reboot?
Here is article esg61779 explaining the driver configuration.
Hope its useful
|Goal: How to troubleshoot lus driver problems on AIXFact: AIXFact: LUS driverFact: Jukebox robotics driverFix: Run the following set of commands and check displayed outputs to see if Lusnneds to be reconfigured.lslpp -h lusFileset Level Action Status Date Time----------------------------------------------------------------------------Path: /usr/lib/objreposlus188.8.131.52 COMMIT COMPLETE 07/25/03 18:20:18lsdev -C -l luslus Available Legato SCSI User Interface Release nw_6_1_3_PM.Build.26Redefine lus if need be as:mkdev -l luslus DefinedIf the user is updating or adding a replacement of the existing lus driver dothe following (after copying lusdd to the drivers dir.):rmdev -d lusmkdev -c port -s scsi -t lusrather than:mkdev -l lusHere is the sequence to follow with the sample output:1. Make sure the following directories/files exist:In /usr/lib/methods, the following files should exist:cfglusucfglusdefluslus.catstartlusThe file /usr/lib/drivers/lusdd* exists.2. Make sure the Legato driver category has been added to theAIX device configuration database.This will be apparent in the error message returned bymkdev when you try to define and make the Legato driveravailable:mkdev -c port -s scsi -t lusWill return the error:mkdev : 0514-520 cannot find information in the predefineddevice configuration database for the following : type = "lus"subclass = "scsi" class ="port"Or you can verify with the following command, which looksup the device configuration database for our driver category:odmget -q"type=lus" PdDvYou should see output like this:PdDv:type = "lus"class = "port"subclass = "scsi"prefix = "lus"devid = ""base = 2has_vpd = 0detectable = 1chgstatus = 0bus_ext = 0fru = 1led = 1638setno = 1msgno = 1catalog = "lus.cat"DvDr = "lusdd"Define = "/usr/lib/methods/deflus"Configure = "/usr/lib/methods/cfglus"Change = ""Unconfigure = "/usr/lib/methods/ucfglus"Undefine = "/usr/lib/methods/undefine"Start = ""Stop = ""inventory_only = 0uniquetype = "port/scsi/lus"If the Legato driver category is not in the device configurationdatabase, you can add it manually with the following command:odmadd /usr/lpp/lus/lus.addThen verify with the odmget command.3. Make sure the driver has been defined and made available.Run lsdev -C -l lus. You should see output like this:# lsdev -C -l luslus Available Legato SCSI User Interface Release nw_6_1_3_PM.Build.26If that's not the case, you can manually run the script:/usr/lib/methods/startlusYou should see "lus Available" when the script completes.|
Well the esg61779 didn't help because the problem also exists on servers where no jukeboxes are connected.
We opened a Service Request at EMC². The problem is still under investigation. I'll post an update if we get information from EMC²
I don't have much experience with AIX, but if LUS is just used for robotic control then you could remove it from those storage nodes hosting just drives.... but perhaps EMC should be the one to say that (many years ago I was working with lus on Solaris and if I remember correctly I had to use it on storage nodes where I had only drives due to otherwise unknown HBA or something). Not sure what lus does while you it is supposed to unload during reboot, but I believe there was lusdebug command to get it into debug mode... might be worth of checking it out and perhaps run it to see what does it do.