Re: ASMDF dependencies with MPATH and UDEV

Our standard lun size is 2 TB for diskgroups used for data, with around 100 tb per RAC cluster, and up to 100 luns counting FRA, OCR , GG etc. We do not have any issues with timing. Putting in a sleep in the AFD startup script seems reasonable for a quick workaround, but adding a new systemd service would be better.
You did not mention what arrays you are using, but we have a mix of Netapp FC and IBM Flashsystem 9200 FC, all using flash drives.

On Thu, Oct 3, 2024 at 5:16 AM Laurentiu Oprea <laurentiu.oprea06_at_gmail.com> wrote:

> THank you. After the LUN size increase did you experience any performance
> impact ?
>
> În mar., 1 oct. 2024 la 12:23, Ricard Martinez <ricard.martinez_at_gmail.com>
> a scris:
>
>> Hi, I had a similar problem. The way we fixed it was to reduce the number
>> disks/lun, creating new bigger ones, adding them to ASM, rebalance, remove
>> the small ones. We had 250g luns, created new 1TB ones, so we reduced the
>> number down by 4. That helped us in our situation, so maybe is something
>> you can consider if possible.
>> Regards
>>
>> On Tue, 1 Oct 2024, 09:43 Laurentiu Oprea, <laurentiu.oprea06_at_gmail.com>
>> wrote:
>>
>>> Hello all,
>>>
>>> Did anyone runned into a situation where due to the very high number of
>>> disks and multipaths the ASM AFD driver will fail as some disks are not yet
>>> ready when starting in the boot sequence after the server restart?
>>>
>>> I was thinking that as a workaround, adding in rc.d/init.d/af a sleep
>>> sounds very random as I cannot determine what is the correct value. Is
>>> there a smart way to treat this situation?
>>>
>>> Thank you.
>>>
>>

--
http://www.freelists.org/webpage/oracle-l