Brighstor degradation conection

[Fernando Boza]

ENVIRONMENT DETAILS
SERVER:
- Server Name = CRDSEN001
- Windows 2003 Server Build 3790
- CA AB Server version r12.1.5582.0
- MSSQL Agent r12.1.5582.0
- OFA r12.1.5582.0
- Client Agent r12.1.5582.0


REMOTE SERVER:
- Server Name = CRDUXDES
CA BrightStor ARCserve Backup Agent for UNIX/Linux r11.5 (Build 2427)
HP-UX Agents

Apart of the BAB Manager 12.0 There are 4 clients HP-UX

CRDHP01
CRDHP02
CRDUXDES
CRDUXDES2
the machines called CRDHP01 and CRDHP02.. are in Cluster, and they are sharing LUNs
this is and Active/Pasive Cluster. on this moment the customer backup only
on the active node.
The HP-UX agents are installed as StandAlone mode.
----------------------------------------------------------------------------
PROBLEM DESCRIPTION :

when backing up 5GB , 10GB data all is ok however when backing up more than
this (40gb as example in just one file syetem) the network connections stops
to work . The network activity goes down between the Manager and
HP'UX agent .
we had previously working the same manner but with BAB Manager
11.5 SP3 installed over HP-UX and 11.5 SP3 on Windows.
The problem we have is degradation on unix servers throughput.

TROUBLESHOOTING DONE :
** Tested after applied RO01340 , RO02316
QO89984 , HP -BAB R11.5 SERVICE PACK 3 FOR HP
RO02732 , HP -DEVICE SUPPORT UPDATE 18 - HP

** Disabled the TCPChimney on win2003
** added -m 1024 on uag.cfg file for HP-UX agent.
Problem continues with backup is only when a file system (unix box)
containing more than 40gb in size is backed full. If the backup is done in
multiple file system (small pieces) there is no problem and the network
activity appears normal. They had backed up 126GB on this tests with
multiple file system without problems.
when the backup is done on sybase file system (dump files 40GB in total)
the network activity and throughput goes to 0 and the HP-UX agent
goes to Sleep status.

It seems the communication between the unix agent and Manager stops to
work. They have waited during 10 mins to see if restart the activity but
nothing occurs. cancel the job is the next step
logs does not show errors on jobh 665 (job showed in screenshots)

any idea