sPHENIX discussion of software

[pinkenburg <pinkenburg AT bnl.gov>]

Hi folks,

it looks like the problem was identified and fixed. gpfs has a limit of 
inodes (basically number of directories and files) and we were at the 
edge. This explains the seeming randomness we saw spread over all our 
gpfs volumes. sdcc increased this limit and writes which previously 
failed reproducibly go through now.

If you still encounter write issues, please let me know

Chris


On 7/20/2024 10:37 AM, pinkenburg wrote:
> Hi folks,
>
> it looks like the problem we had on /sphenix/user (device full event 
> though it had 40TB free) also affects other filesystems in gpfs. So 
> far I see /sphenix/WWW, cvmfs (build fails with out of space) and 
> where the production writes (/sphenix/data/data02) and our simulation 
> logs (/sphenix/sim/sim01) as well as a PHENIX filesystem. It seems to 
> be intermittent with no rhyme or reason. I filed a ticket and raised 
> the alarm bells in the MM channel for this. If you encounter gpfs 
> write issues please pop a message to our software help channel:
>
> https://chat.sdcc.bnl.gov/sphenix/channels/sphenix-software-help
>
> Chris

--
*************************************************************

Christopher H. Pinkenburg       ;    pinkenburg AT bnl.gov
                                ;    http://www.phenix.bnl.gov/~pinkenbu

Brookhaven National Laboratory  ;    phone: (631) 344-5692
Physics Department Bldg 510 C   ;    fax:   (631) 344-3253
Upton, NY 11973-5000

*************************************************************