I don’t think the problem is specific to ‘recon-all -long’. When there is no synchronization mechanism to guard sequential access to shared resources, the race condition can cause collisions between processes
and lead to program failures. In general, ‘-parallel’ should not be used for recon-all submissions.
Yujing
From: Peter Najdzionek <najdziop@mcmaster.ca>
Sent: Tuesday, February 24, 2026 11:52 AM
To: Freesurfer support list <freesurfer@nmr.mgh.harvard.edu>
Subject: [Freesurfer] Re: FreeSurfer Longitudinal Pipeline Exits with Errors - Prior to Sphere registration?
External Email - Use Caution
Thank you for the suggestion.
You were correct, removing the -parallel portion of the script resulted in successful longitudinal processing.
For documentation purposes on my end, any ideas why '-parallel' cannot be used for 'recon-all -long'? My lab has previously used '-parallel' to parallel-process cross-sectional data ('recon-all
-all') successfully.
Thanks again for the insight!
~Peter
Peter Najdzionek,
MSc
(He/Him/His)
Ph.D. Candidate - Neuroscience Graduate Program
phone:
(289) 440-3892
email:
najdziop@mcmaster.ca

McMaster University recognizes and acknowledges that it is located on the traditional territories of the Mississauga and Haudenosaunee nations, and within the lands protected by the
Dish With One Spoon wampum agreement.
From: Huang, Yujing <YHUANG43@mgh.harvard.edu>
Sent: Friday, February 20, 2026 1:50 PM
To: Freesurfer support list <freesurfer@nmr.mgh.harvard.edu>
Subject: [Freesurfer] Re: FreeSurfer Longitudinal Pipeline Exits with Errors - Prior to Sphere registration?
|
You don't often get email from
yhuang43@mgh.harvard.edu. Learn
why this is important |
Caution: External email.
Your recon-all was run with ‘-parallel’ option. It is the multi-processing done at the recon-all level. There is no synchronization mechanism to ensure shared resources access/update.
Can you try to submit the recon-all without ‘-parallel’?
Yujing
From: Peter Najdzionek <najdziop@mcmaster.ca>
Sent: Friday, February 20, 2026 12:47 PM
To: freesurfer@nmr.mgh.harvard.edu
Subject: [Freesurfer] FreeSurfer Longitudinal Pipeline Exits with Errors - Prior to Sphere registration?
External Email - Use Caution
Hi FreeSurfer folks,
I'm attempting to use cloud computing (via Compute Canada) to run the FreeSurfer longitudinal pipeline on a batch of participants' images.
It seems like every participant exits at the same point in the pipeline: after
mris_curvature.
Unfortunately, the error is extremely non-descript (recon-all -s subject exited with ERRORS"), making troubleshooting difficult.
Any advice or thoughts are greatly appreciated; recon-all.log and recon-all.error attached.
Hardware details:
Freesurfer 8 (freesurfer-linux-centos8_x86_64-8.0.0-20250203-0f2bd3a)
Linux (PlatformVersion: 5.14.0-611.13.1.el9_7.x86_64)
uname -a: Linux nc30435 5.14.0-611.13.1.el9_7.x86_64 #1 SMP PREEMPT_DYNAMIC Thu Dec 11 04:57:59 EST 2025 x86_64 AMD EPYC 7532 32-Core Processor AuthenticAMD GNU/Linux
~Peter N
Peter Najdzionek,
MSc
(He/Him/His)
Ph.D. Candidate - Neuroscience Graduate Program
phone:
(289) 440-3892
email:
najdziop@mcmaster.ca
McMaster University recognizes and acknowledges that it is located on the traditional territories of the Mississauga and Haudenosaunee nations, and within the lands protected by the
Dish With One Spoon wampum agreement.