Storage Resource Manager Enterprise Edition Installation Guide

Storage Resource Manager Enterprise Edition C–23
SRM Troubleshooting Guide
Then it is supposed to report the scan failure to SRM, which in turn should write it to
the NT Event log.
NOTE: When a program terminates abnormally, it is not always possible to cleanly catch the
exception and clean everything up. In any case, SRM will eventually be given a “timeout” status
to declare that the scan failed. In the event log, this will be seen as the following:
Type: ErrorSource: SRMOEMScannerCategory: NoneEvent: 12Description: An SRM
timeout has occurred on pipe \\.\pipe\FfromMips15_990527020017_Srm_raid5 for the file
scan on Mips15.
Type: ErrorSource: SRManagerServer-RMAGG_EVTEvent:10Description: SRM Server:
Partition /Srm_raid5 failed file scan with code 8007000d (The data is invalid.).
What Happens if the SRM Server Fails?
If agent scans were in progress at the time of SRM termination, UNIX agents may
eventually try to post their results, depending on the timing of the failure and when the
SRM Server comes back up. NT Agents will finish their scan but will not send the data
back to the server, because the original RPC connection has been torn down.
When the SRM Server comes back up, it checks its schedule to see if any jobs were to
be run while it was down. If so, it will run them immediately. If a job was in the
process of running when the server went down, the schedule shows that it was not run,
and the job will be started immediately.
To cause a scan to NOT be started when the SRM Server is started, set the scan
schedule for that particular group to “Scan On Demand.”
“Scan To Pipe - No Connection To Pipe …, Giving Up?”
This occurs when the UNIX agent attempts to post the results of a scan, and SRM
never picked them up.
For 2 minutes the OEMSCANIO CGI executable tries to make a direct pipe
connection to SRM. Then it will divert agent data into a file (allowing the agent
process to terminate on UNIX), create another process to handle it, and then continue
attempting to post the data for up to 30 more minutes.