wait status: pipe

[Kenan Erdey]

hi;

for a while i have problem. the process nsrnmostart( process of legato oracle module that triggers backup in client side) doesn't end so the backup which i start from legato server doesn't end either. when i do "ps -ef | grep nsr" i see running processes and when i look at the process wait state of nsrnmostart from glabce, i can see wait state of reason %100 pipe.
client which is backed up is hp-ux 11.11
legato server windows 2003

why doesn't this process end ? thanks.

[spex]

Kenan,

What do the Legato and Oracle logs tell you?

PCS

[Kenan Erdey]

i couldn't find anything from logs.

[spex]

How far do the backups get? Partially complete? Fully complete?

Did anything on the HP-UX or Windows systems change around the time the problem began? How about other systems on the same network link? Any increase in network traffic?

[Kenan Erdey]

ok, let me tell. i get backups daily, as full. there is no problem with other oracle machine backups. i first get the backup to disk with rman at night. in the morning, i am getting it to legato server with rman again. after getting backup pieces from disk to legato server, index backup must start but it doesn't because nsrnmostart process in client doesn't end.
yes it can be because of something changed in hp-ux. i looked for any permission changes to the directory or files that processes can write. i couldn't find that either. no network traffic increase by the way.

[Sandman!]

Run tusc on the client-side process that is waiting on the pipe resource/event. Post the output of tusc here as the gurus will most likely be able to point out the cause of the problem.

~cheers

[Kenan Erdey]

i am attaching the tusc output. when the process hanged i terminated logging.