We are experiences a problem using rpc funcions inside a C program. Our application seems to have problems only on B2000,B2600 workstations. We received a suggestion to execute "ioscan -fnC disk" command on ws where our application runs and on a ws where our application fails. On last ws (ko) ioscan takes 45 sec. while on first ws (ok) it takes 5 sec. Can these two events be related ? Is it a problem of some hp patch ?
Are the two workstations the same verion of the O/S? Ioscan changed behaviour between 10.20 and 11.0 (I think). It now looks at all the physical devices. You can force it to just scan the kernel by using the -k option. If both ws are the same version, try the -k option and see what that results in.
Do the workstations differ in the amount of devices reported by ioscan? What I mean is, does ioscan have more legitimate work to do on one than it does on the other?
I fully agree with Steve. Terminator or other SCSI bus problems (like misbehaving device) is leading to ioscan working too long time and somethmes even hangs. Unfortunately there're no external SCSI connectors to terminate in B2000... maybe B2600 has them? Maybe you have addon SCSI card installed with bus terminated improperly? Try disconnecting all external devices (if any, exluding, of course, terminal, keyboard and mouse :o) ) Eugeny
I would change your application to add the -k option. It will always run instantly since it does not query every I/O channel and device on the system. Unless you are (dangerously) adding devices with power on, ioscan -kfnC disk will be just as valid as without -k, but will complete in a second or two. The -k tells ioscan to read all the info from kernel memory.
