Abortet FOS Downgrade

[sthiel]

Hello

after an aborted Downgrade Procedure on an 8Gb Bladeswitch from FOS 7.4.1 to 7.3.1d (Because of Performance Issues), the Switch comes up with Informations that let me assume, there is an Problem with Flash Memory.

Console-Output:

write the pid: No space left on device
/bin/sed: couldn't write 78 items to {standard output}: No space left on device

Typing typical FOS Commands like switchshow, respond "fabos not yet initialized: -1".

My try to boot from second hda brings the same Error.

Command firmwareshow tell me:

Appl     Primary/Secondary Versions
------------------------------------------
FOS      v7.3.1d
         v7.4.1

WARNING: Firmwaredownload is in progress.

After some Time, the Switch Reboots automaticaly.

It seemes like, the firmwaredownload could not complete due to insufficient memory.

Does somebody has any idea, how to switch back to previous FOS or how to Erase 7.3.1d Files and Reboot FOS v7.4.1 from second Partition?

[Antonio Bongiorno T…]

@sthiel

--->>> (Because of Performance Issues),

is you assumption that FOS 7.4.x caused performance issue ?

or what is the reason you do downgrade to 7.3 ? BTW 7.3.x  is no longer supported

--->>>It seemes like, the firmwaredownload could not complete due to insufficient memory.

this happened in  rare circumstance when a core file is present

try follow: at no waranty

"supportsave -R" to remove any core/dump file is present

and then start "firmwarecommit"

[sthiel]

@Antonio Bongiorno TechHelp24

Actually FOS 7.4.1 brings our older Blade Switches to an Memory Load > 76%.

So although this FOS-Version is no longer supported, i thougt it´s better to take a step back.

First of all, thanks for your suggestion.

I tried this way and the Dump-Files were also deleted. But after the firmwarecommit the issue stays the same.

The Switch comes up and reports:

write the pid: No space left on device

/bin/sed: couldn't write 78 items to {standard output}: No space left on device

no listening sockets available, shutting down
AH00015: Unable to open logs

evm(main): raslogm_init failed: -3
*** superd: daemon evmd returned -3; tfs= 2391
write the pid: No space left on device
rpc(main): raslogm_init failed: -3
*** superd: daemon rpcd returned -3; tfs= 2663

Afterward then he restarts again and again...

Any more Ideas?

[Antonio Bongiorno T…]

@sthiel

show as the switch try to boot from a partition with corrupted FOS.

you can try from boot to change the boot partition.

--->>>Actually FOS 7.4.1 brings our older Blade Switches to an Memory Load > 76%.

--->>>So although this FOS-Version is no longer supported, i thougt it´s better to take a step back.

the high memory load is not related to 7.4.x or 7.3.x most probable as sayed in my preview post the switch have core/dump file 

another problem is during the downgrade is follow:

by firmwaredownload as"upgrade" from 7.3.x to 7.4.x the FOS perform a Migration, and the upgrade procedure remove some uneeded file where are not longer required.

now you have perform firmwaredownload as "downgrade" from 7.4 to 7.3.1, and this is not a recommended release for a downgrade.

in feew word, if the switch continued with the failure, you need to open a TAC by HP

[sthiel · Accepted solution]

Fond a Solution!

During the Downgrade, the Filespace of the Switch was simply Empty.

My way to Solution:

Login via Root, compare Filespace with identical Switchtype by using Linux "du" command:

Result: There were many older Lofiles in /var/log from which I deleted the oldest ones.

Then i did an firmwarecleaninstall wih FOS v7.4.2c.

Warning! Before that, you should have backed up all licenses (for example by an SAN-Health Report - Thank´s for that hint to @Antonio Bongiorno TechHelp24).  All Switch-Configurations must also be reset!