sphenix-production-l AT lists.bnl.gov

Subject: Sphenix-production-l mailing list

List archive

Re: [Sphenix-production-l] [EXTERNAL] runs which had memory overflow

From: Alex Lebedev <lebedev AT iastate.edu>
To: "Pereira Da Costa, Hugo Denis Antonio" <hugo.pereira-da-costa AT lanl.gov>
Cc: "sphenix-production-l AT lists.bnl.gov" <sphenix-production-l AT lists.bnl.gov>, Hugo Pereira Da Costa <hugo.pereira.da.costa AT gmail.com>
Subject: Re: [Sphenix-production-l] [EXTERNAL] runs which had memory overflow
Date: Fri, 21 Jun 2024 08:08:03 -0500

Hi Hugo,

The runs in this list run out of memory within the first day of running.

I did not include runs which reached the 3 day running limit on condor in this list.

We suspect that at least right now the main culprit for memory leaks is mvtx,

so, as a test, last night we excluded mvtx from full streaming production to

see if it will improve success rate.

Sasha.

On Wed, Jun 19, 2024 at 4:29 PM Pereira Da Costa, Hugo Denis Antonio <hugo.pereira-da-costa AT lanl.gov> wrote:

Hi Sasha,

Many of the runs in your list are non-Zero suppressed low trigger rate TPC runs. My understanding is that, exepectedly these are very slow to reconstruct fully ? And maybe that is the reason for the See list below.

The runs which I marked 'ok' are high trigger rates ZS runs. For those at least, TPC cannot be blamed. I'm trying to run TPOT alone on them to double-check stability.

Hugo

45828

45827

45826

45825

45824 <- all with non ZS TPC.

45749 <- TPC non ZS

45747 <- TPC non ZS

45723 <- ok

45721 <- ok

45699 <- ok

45675

45674

45670

45669 <- cosmics with TPC non ZS

45625 <- ok

On 6/19/24 11:49 AM, Pereira Da Costa, Hugo Denis Antonio wrote:

From: Sphenix-production-l <sphenix-production-l-bounces AT lists.bnl.gov> On Behalf Of Alex Lebedev Sent: Tuesday, June 18, 2024 1:19 PM To: sphenix-production-l AT lists.bnl.gov Subject: [EXTERNAL] [Sphenix-production-l] runs which had memory overflow

Hi all,

Here is a list of runs which were evicted from condor due to

memory overflow (in less than 3 days) in the past several days,

in no particular order.

There are also a bunch of runs evicted due to reaching the 3 day limit,

but I guess they are ok, just too long.

Sasha.

45828

45825

45826

45827

45824

45699

45675

45674

45625

45723

45721

45670

45669

45749

45747

--
Sphenix-production-l mailing list
Sphenix-production-l AT lists.bnl.gov
https://lists.bnl.gov/mailman/listinfo/sphenix-production-l

[Sphenix-production-l] runs which had memory overflow, Alex Lebedev, 06/18/2024
- Message not available
  - Message not available
    - [Sphenix-production-l] [EXTERNAL] runs which had memory overflow, Pereira Da Costa, Hugo Denis Antonio, 06/19/2024
      - Re: [Sphenix-production-l] [EXTERNAL] runs which had memory overflow, Alex Lebedev, 06/21/2024