Folks, Anyone with experience running the 5.1 HSM client on Solaris, read on. Ideas are welcome.
Thanks, Kelly J. Lipp Storage Solutions Specialists, Inc. PO Box 51313 Colorado Springs, CO 80949 [EMAIL PROTECTED] or [EMAIL PROTECTED] www.storsol.com or www.storserver.com (719)531-5926 Fax: (240)539-7175 -----Original Message----- From: Thee, Gwen [mailto:[EMAIL PROTECTED]] Sent: Monday, April 29, 2002 12:34 PM To: '[EMAIL PROTECTED]' Subject: HSM on Solaris 8 > Hi Kelly, > > I installed hsm on one of our Solaris 8 machines, (the new 5.1 client with > 64 bit support), and I am having some problems. Hopefully you can help me. > The good news is that the essential task of migrating and retrieving files > works fine if we do a manual migration. The bad news is twofold. It will > not pre-migrate any files, or start the migration automatically even > though it is above the threshold to start, and the following bad events > have been going on since I activated the first hsm file system. > > The dsmrecalld daemon will not stay up. It disappears about every 30 to 60 > minutes. There are no errors issued at the time that it dies. We have a > cron that restarts it, but obviously this is a bad thing since if someone > tries to get a file in the 60 seconds that it's not running, they get the > stub. > > The other thing is I have constant errors in the dsmerror.log. These also > started at the time we activated the first file system with hsm, and they > are issued about every 10th of a second. The Tivoli site is very unhelpful > with this error, here's their description of it: > > ANS9511E > program-name: cannot read DM attributes on session session for > file handle = handle token = token. Reason : error > Explanation: TSM space management cannot read the DM attributes of a DM > object, usually a file. > System Action: Processing of the file is interrupted. > User Response: Continue with normal operation. > > I did not find this very useful in determining my problems. This is what > the dsmerror.log looks like: > > 04/29/02 14:54:00 ANS9511E dsmmonitord: cannot read DM attributes on > session > 11847 for > file handle = 0000000000ff01ff 0000000700000000 00000000001a766b > 000001001011ff0c token = DM_NO_TOKEN. Reason : No such process > 04/29/02 14:54:10 ANS9511E dsmmonitord: cannot read DM attributes on > session > 11847 for > file handle = 0000000000ff01ff 0000000700000000 00000000001a766b > 000001001011ff0c token = DM_NO_TOKEN. Reason : No such process > > It has been issuing these errors non-stop since we activated an hsm file > system. To give you an overall feel for how things are set up, here's the > parameters set in the dsm.sys file: > CANDIDATESInterval 24 > CHECKFororphans yes > CHECKThresholds 5 > MAXCANDProcs 5 > MAXMIGRators 1 > MAXRecalldaemons 20 > MAXRECOncileproc 3 > MAXThresholdproc 3 > MIGFILEEXPiration 7 > MINMIGFILESize 10000000 > MINRECAlldaemons 3 > RECOncileinterval 24 > > SErvername puppy_tsm > COMMmethod TCPip > TCPPort 1500 > TCPServeraddress 129.228.65.204 > passwordaccess generate > schedlogretention 7 > errorlogretention 7 > INCLexcl /opt/tivoli/tsm/client/ba/bin/include_exclude > schedlogname /opt/tivoli/tsm/client/ba/bin/dsmsched.log > errorlogname /opt/tivoli/tsm/client/ba/bin/dsmerror.log > > and the dsm.opt: > COMPRESSIon No > OPTIONFormat SHort > RESToremigstate No > SErvername puppy_tsm > > The management class parameters are set to: > spacemgtechinique Auto > automignonuse Three > migrequiresbkup Yes > migdestination hsmdisk > > The only thing he did successfully on his own is create the candidate > list. He did not pre-migrate and he did not migrate automatically. > > Please help, > Thanks, > Gwen >
