Disk volume is down Flex 5260

[rafanto]

Hello,

Please help with this query, based on your experience.

We have a Flex Appliance 5260 with 3 Primay, Media, WORM instances.

Lately we have had two Disk Volume is Down events (error 2074), but it has recovered without any action, after a few hours.

We believe that some tuning needs to be done, since we have up to 6 tape mirrors in parallel and more than 60 backup threads. Do you think it is because of this load that the MSDP crashes?

An Appliances 5260 (MSDP 300TB) with 3 instances (512GB RAM), how many are the maximum backup threads it can handle?

At the moment we have 100 in the DiskPool and 100 in the Storage Unit.

Thank you!

Rafael

[kiyan]

I think the Netbackup planning and performance Tuning guide has some kind of best practices.
https://www.veritas.com/support/en_US/doc/21414900-146141073-0/v151900227-146141073

[Slartybardfast]

Less is more with MSDP. Generally STU concurrent connections should be 10% less than the IO sessions for disk pools. For example start with 10 disk pool sessions and set the concurrent connections for the STU to 9. then maybe 15 disk pool sessions and 11 concurrent STU concurrent connection, you get the idea. Increace slowly until you find a sweet spot. There is no magic figure here you have to tune to your environment what works for my environment might not work for your system. Check out Kiyan's link in the above post that is good advice.