UNSOLVED
Savesets fail "failed immediate send 130625 bytes"
Hello all,
We are running networker 7.3.2 on solaris 10 with 100+ clients. Four of them are windows (2003 enterprise server) clusters, configured as dedicated storage nodes. We are seeing frequent failures part way through the backups which affects any save streams still running. The error message in the email notification and in the daemon.log file is "failed immediate send 130625 bytes" (the number of bytes varies).
The backup hardware is LTO3 over FC through Qlogic SANBox 5200 FC switches. Our monitoring has not detected an obvious correlation with interruptions to either our FC or ethernet infrastructure. Data volumes are of the order of 1-2 Tb per cluster.
Has anyone else experienced similar symptoms, and maybe found a resolution?
regards
Andy Garman
We are running networker 7.3.2 on solaris 10 with 100+ clients. Four of them are windows (2003 enterprise server) clusters, configured as dedicated storage nodes. We are seeing frequent failures part way through the backups which affects any save streams still running. The error message in the email notification and in the daemon.log file is "failed immediate send 130625 bytes" (the number of bytes varies).
The backup hardware is LTO3 over FC through Qlogic SANBox 5200 FC switches. Our monitoring has not detected an obvious correlation with interruptions to either our FC or ethernet infrastructure. Data volumes are of the order of 1-2 Tb per cluster.
Has anyone else experienced similar symptoms, and maybe found a resolution?
regards
Andy Garman
- I think we've got to the bottom of this one. We noticed the error occurred within a few minutes of 0545 every Sunday morning, coincidentally when we usually run Sun's Explorer diagnostics gathering package on our unix systems, some of which share drives with our windows cluster clients. It looks the probing of SCSI and FC devices carried out Explorer doesn't bother other unix nodes, but windows ones throw a wobbly. We have now managed to isolate the drives from most of our unix dedicated storage nodes via our switch zoning, and the few hosts we can't unmap have had their explorer runs retimed to less disruptive times of the week.
We'll know for sure on Sunday morning!
atremols
29 Posts
91
0
Posted August 27th, 2008 11:00
Please create an empty file with the name of: noimmediate
the file should be created in nsr\debug
if debug directory is not created then please created, this should be done on
all storage nodes showing that problem
regards,
Abdon