The system has reached its maximum device addressable limit

[FTSE SysAdmin]

Hi,

I've been seeking a solution for this EMS error message:

"The system has reached its maximum device addressable limit"

"The total LUNs discovered by the host bus adapter through all the fibre channel devices is beyond the system addressable limit. This can happen if there is a large number of LUNs configured (either in a single fibre channel device or in multiple devices). "

Right what I have is a rp7410 with HP-UX 11.11; two dual FC cards connected to a EVA5000, plus securepath.

In terms of LUNs (VDisks) configured and displayed to the above server they are just 11.

Would someone come across this error?

Would be a kernel parameter?

Many thanks

[Alzhy]

You probably are connected on a SAN that has a misbheaving NSR (Storage Router) and it is presenting (or opened for scan) a large number of LUNs and targets.

If it is taking your ioscan forever to finish.. then that is a good indication of this possible scenario. You may also want to do a tdutil /dev/td0 get remote all.

[erics_1]

What verion of support tools is installed?
Use the 'monconfig' command to get STM and EMS versions.

Regards,
Eric

[Mark Greene_1]

See here for the actual hard-number limits and explainations of how to calculate:

http://docs.hp.com/en/1891/maxlun-whitepaper-v1.0.pdf

Now you said "two dual FC cards"; am I correct in reading this as 4 seperate physical connections? I assume they connect to a fibre swich, yes? If so, how many physical connections back to the EVA5000? And how are all these connections zoned?

If you have any circular references in your zoning you have a problem; like all 88 SCSI addresses being duplicated in each of the zones for the 4 physical connections * 2 for the connections to the EVA, you are now at 704 virtual LUNs and potentially over the limit.

mark

[FTSE SysAdmin]

Many thanks Mark Greene for the link to the pdf. I've found it useful and currently I'm running through it.

I've got dual FC cards however just two physical connections to EVA5000 and yes they are zoned.

regarding kernel parameters; maxfiles and maxfiles_lim, I have them with 2048.

The number of VGs is 13 and majority have 1 lvol except vg00 that has 12 lvols. Therefore all in the LVM limits

Nelson sugested that ioscan would take forever to finish. But that isn't my case, the ioscan runs quite fast and reports all devices including all vdisks made visible by EVA.

Regarding EMS version, I've got HPUX 11.11 Support Tools Bundle, Jun 2004 (B.11.11.14.15)

This seem a bit more than what EMS is reporting, spite the EMS and this errors, the system seem to work without major problems.

Dec 16 16:29:13 hp5 vmunix: 0/0/8/0/1: Fibre channel host port is OFFLINE, can not scan
Dec 16 16:29:13 hp5 vmunix:
Dec 16 16:29:13 hp5 vmunix: 1/0/2/0/1: Fibre channel host port is OFFLINE, can not scan
Dec 16 16:29:17 hp5 vmunix: 0/0/8/0/1: Fibre channel host port is OFFLINE, can not scan
Dec 16 16:29:17 hp5 vmunix: 1/0/2/0/1: Fibre channel host port is OFFLINE, can not scan

BTW I've logged this issue on HP

[Mark Greene_1]

Run this:

fcmsutil /dev/td0 stat

for both HBAs and see if it reports any errors.

mark