# Large datasets crash gemma

Running GEMMA on the NSNIH dataset in Genenetwork sends the server in a tail spin and logs `BUG: soft lockup CPU stuck` messages. This obviously is not great and appears to be a side effect of running openblas aggressively in parallel (I remember seeing some evidence of that, but I can no longer find that message). Or it may be GEMMA simply runs out of RAM and the kernel is busy cleaning up using the OOM reaper. See

=> https://lkml.iu.edu/hypermail/linux/kernel/2003.2/01012.html

## Tags

* assigned: pjotrp, zachs

## Tasks

* [ ] tux02: test out-of-band-access
* [ ] tux02: test GEMMA
* [ ] tux02: set overcommit memory on tux02 to 2 (see below)
* [ ] tux02: reboot and reinstate services on tux02
* [ ] tux02: test GEMMA
* [ ] tux02: try and optimize versions of openblas using -O2
* [ ] tux02: deploy GEMMA latest

And do the same on tux01 (production)

## Notes

A 'soft lockup' is defined as a bug that causes the kernel to loop in kernel mode for more than 20 seconds without giving other tasks a chance to run. The watchdog daemon will send an non-maskable interrupt (NMI) to all CPUs in the system who, in turn, print the stack traces of their currently running tasks.

After a gemma lockup we see

```
[2512382.403215] watchdog: BUG: soft lockup - CPU#118 stuck for 22s! [migration/118:609]
[2512404.477219] Out of memory: Kill process 1723 (gemma) score 87 or sacrifice child
[2512404.569158] Killed process 1723 (gemma) total-vm:44620288kB, anon-rss:25261688kB, file-rss:0kB, shmem-rss:0kB
[2512405.788221] oom_reaper: reaped process 1723 (gemma), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB
```

It is clear parallel GEMMA is running out of RAM.
We can make softlocks messages relax by setting `/proc/sys/kernel/watchdog_thresh` higher. Consider the message harmless.

Overcommit is set to 0 on Tux01. We may want to change that to

```
vm.overcommit_memory=2
vm.overcommit_ratio=90
```

That will make out-of-RAM problems less impactful. We have been running penguin2 like this for over a year with no more OOM problems. I have not set that before on tux01 because it requires rebooting the production server.

From Zach I got the K and GWA commands:

```
/usr/local/guix-profiles/gn-latest-20220122/bin/gemma-wrapper --json --loco 1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,X -- -debug -g /home/zas1024/gn2-zach/genotype_files/genotype/bimbam/HSNIH-Palmer_true_geno.txt -p /home/zas1024/gn2-zach/tmp/gn2/gn2/PHENO_2+FcfQiTVSC7FmmbsatUPg.txt -a /home/zas1024/gn2-zach/genotype_files/genotype/bimbam/HSNIH-Palmer_true_snps.txt -gk > /home/zas1024/gn2-zach/tmp/gn2/gn2/HSNIH-Palmer_K_TPTFHJ.json

/usr/local/guix-profiles/gn-latest-20220122/bin/gemma-wrapper --json --loco --input /home/zas1024/gn2-zach/tmp/gn2/gn2/HSNIH-Palmer_K_TPTFHJ.json -- -debug -g /home/zas1024/gn2-zach/genotype_files/genotype/bimbam/HSNIH-Palmer_true_geno.txt -p /home/zas1024/gn2-zach/tmp/gn2/gn2/PHENO_2+FcfQiTVSC7FmmbsatUPg.txt -a /home/zas1024/gn2-zach/genotype_files/genotype/bimbam/HSNIH-Palmer_true_snps.txt -lmm 9 -maf 0.05 > /home/zas1024/gn2-zach/tmp/gn2/gn2/HSNIH-Palmer_GWA_MWKKYW.json
```

The geno file is massive:

```
3.7G Mar 12 11:56 HSNIH-Palmer_true_geno.txt
 24K Mar 12 11:56 PHENO_2+FcfQiTVSC7FmmbsatUPg.txt
3.4M Mar 12 11:56 HSNIH-Palmer_true_snps.txt
```

Probably best to test on a different machine! Let's move to tux02. Running luna (a half year old version of GN2) gives `**** FAILED: number of columns in the kinship file does not match the number of individuals for row = 79`. So, that does not help! I think this is a known issue that got fixed later. Next up, try and run gemma by hand for chromosome 1 after installing gemma tools with a recent GNU Guix:

```

```

Now the goal is to try and crash the server before setting overcommit.