Summary: Sun Fire E6900 boots EXTREMELY slow

From: Rongsheng Fang <>
Date: Wed May 28 2008 - 10:28:00 EDT
Thanks to everyone who replied:

Anatoliy Nelyubin
Noel Milton Vega
Francisco Mauro
Bill Voight
James C. McPherson

I asked:

> We have a Sun Fire E6900 with two dual port Sun branded Qlogic 4GB/ 
> s HBAs. The servers is connected to a EMC Symmetrix DMX3 via two  
> redundant Brocade fibre switches. The DMX3 has 4 FA connections  
> going to each fibre switch. So each LUN presented on the OS has 16  
> paths. Currently 221 EMC LUNs are presented to this box, which  
> means the OS will see 3536 EMC LUNs.
> We are running Solaris 10 U5 on this server. We did a reconfigure  
> reboot today, and it's been more than 2 hours and the system is  
> still trying to do detect all the paths to the EMC devices:
> ......
> /ssm@0,0/pci@18,600000/SUNW,qlc@1/fp@0,0/ssd@w5006048ad52cc2c7,1c8  
> (ssd3079) online
> ssd3080 at fp6: name w5006048ad52cc2c7,1c7, bus address 17000
> ssd3080 is /ssm@0,0/pci@18,600000/SUNW,qlc@1/fp@0,0/ 
> ssd@w5006048ad52cc2c7,1c7
> /ssm@0,0/pci@18,600000/SUNW,qlc@1/fp@0,0/ssd@w5006048ad52cc2c7,1c7  
> (ssd3080) online
> ssd3081 at fp6: name w5006048ad52cc2c7,1c6, bus address 17000
> ssd3081 is /ssm@0,0/pci@18,600000/SUNW,qlc@1/fp@0,0/ 
> ssd@w5006048ad52cc2c7,1c6
> /ssm@0,0/pci@18,600000/SUNW,qlc@1/fp@0,0/ssd@w5006048ad52cc2c7,1c6  
> (ssd3081) online
> ssd3082 at fp6: name w5006048ad52cc2c7,1c5, bus address 17000
> ssd3082 is /ssm@0,0/pci@18,600000/SUNW,qlc@1/fp@0,0/ 
> ssd@w5006048ad52cc2c7,1c5
> /ssm@0,0/pci@18,600000/SUNW,qlc@1/fp@0,0/ssd@w5006048ad52cc2c7,1c5  
> (ssd3082) online
> ssd3083 at fp6: name w5006048ad52cc2c7,1c4, bus address 17000
> ssd3083 is /ssm@0,0/pci@18,600000/SUNW,qlc@1/fp@0,0/ 
> ssd@w5006048ad52cc2c7,1c4
> /ssm@0,0/pci@18,600000/SUNW,qlc@1/fp@0,0/ssd@w5006048ad52cc2c7,1c4  
> (ssd3083) online
> ssd3084 at fp6: name w5006048ad52cc2c7,1c3, bus address 17000
> ssd3084 is /ssm@0,0/pci@18,600000/SUNW,qlc@1/fp@0,0/ 
> ssd@w5006048ad52cc2c7,1c3
> /ssm@0,0/pci@18,600000/SUNW,qlc@1/fp@0,0/ssd@w5006048ad52cc2c7,1c3  
> (ssd3084) online
> ssd3085 at fp6: name w5006048ad52cc2c7,1c2, bus address 17000
> ssd3085 is /ssm@0,0/pci@18,600000/SUNW,qlc@1/fp@0,0/ 
> ssd@w5006048ad52cc2c7,1c2
> /ssm@0,0/pci@18,600000/SUNW,qlc@1/fp@0,0/ssd@w5006048ad52cc2c7,1c2  
> (ssd3085) online
> ......
> I am not sure if the long reboot time is due to the large number of  
> paths that the OS has to detect during the reconfigure boot process  
> or something else is wrong. Has anybody experienced this before?

This reconfigure reboot actually took about 3 hours. We had two bad  
links (our of 8) between the DMX and the switch during the time of  
the reboot. I wasn't and am still not sure if that was causing the  
problem. But once the links were fixed, the reconfigure boot time  
comes down to 20 minutes which is quite normal for a Sun Fire E6900.  
Yes, I did try a couple of reconfigure reboots and made sure the  
results were consistent.

James C. McPherson from Sun mentioned:

"if you've got that many paths resulting in an extraordinary
number of luns, then yes I would expect to see a crazylong
boot time."

I didn't enable MPxIO and gave it a shot as the problem seemed to go  
away once we fixed the bad links.

Thanks again to all who responded!

sunmanagers mailing list
Received on Wed May 28 10:29:07 2008

This archive was generated by hypermail 2.1.8 : Thu Mar 03 2016 - 06:44:11 EST