Sysctl and out of memory killer

[fvdw]

fan1_pulses seems to be different you could try to set the value as found on my 5big2 although I never changed or have set that value.

Does the problem go away if you now perform a reboot ?

[samrise]

No, the problem persist after reboot. The good thing is that fan is indeed working. It just cant be control.

Code: Select all

Aug 24 10:00:38 (5big1) local2.notice sudo:     root : TTY=unknown ; PWD=/ ; USER=root ; COMMAND=/bin/echo 1
Aug 24 10:00:40 (5big1) local2.notice sudo:     root : TTY=unknown ; PWD=/ ; USER=root ; COMMAND=/bin/echo 1
Aug 24 10:00:41 (5big1) local2.notice sudo:     root : TTY=unknown ; PWD=/ ; USER=root ; COMMAND=/bin/echo 150
Aug 24 10:10:41 (5big1) user.info fan-d: fan5big: disk temperatures sda 50 sdb 51 sdc 52 sdd 52 sde 51
Aug 24 10:10:41 (5big1) user.info fan-d: fan5big: temperature above limit(48) fan on, temp = 52
Aug 24 10:10:41 (5big1) local2.notice sudo:     root : TTY=unknown ; PWD=/ ; USER=root ; COMMAND=/bin/echo 1
Aug 24 10:10:42 (5big1) local2.notice sudo:     root : TTY=unknown ; PWD=/ ; USER=root ; COMMAND=/bin/echo 1
Aug 24 10:10:43 (5big1) local2.notice sudo:     root : TTY=unknown ; PWD=/ ; USER=root ; COMMAND=/bin/echo 150
Aug 24 10:20:44 (5big1) user.info fan-d: fan5big: disk temperatures sda 50 sdb 52 sdc 52 sdd 52 sde 51
Aug 24 10:20:44 (5big1) user.info fan-d: fan5big: temperature above limit(48) fan on, temp = 52
Aug 24 10:20:44 (5big1) local2.notice sudo:     root : TTY=unknown ; PWD=/ ; USER=root ; COMMAND=/bin/echo 1
Aug 24 10:20:45 (5big1) local2.notice sudo:     root : TTY=unknown ; PWD=/ ; USER=root ; COMMAND=/bin/echo 1
Aug 24 10:20:46 (5big1) local2.notice sudo:     root : TTY=unknown ; PWD=/ ; USER=root ; COMMAND=/bin/echo 150


Seems like its missing something after echo?

Anyway, i stop looking into this because, like you said, this is not the cause for my out of memory problem. I've been searching and i think i am on the good path. The problem is related with nfs server. The problem is cause by nfs trying to cache one "file" bigger than the available memory causing out-of-memory killer to start killiing process trying to get enough memory. I'm still trying some suggestions, wish me luck.

This link might explain the problem better: https://major.io/2008/12/03/reducing-in ... er-at-bay/

[fvdw]

about the syntax of output in syslog, it is normal that it only shows the command and not the redirect.
Ps I would have expected that the temp drops when the fan is running at full speed. At least within 30 minutes it should be below, unless you have this nas in a very hot room

mmm..interesting that nfs can cause this OOM effect.

[samrise]

After intence research i've been using the following settings with very good results avoiding out of memory errors but with slightly worse transfer performance. How can i make those settings permanent?

Code: Select all

root@nas:/ # cat /proc/sys/vm/vfs_cache_pressure
10000
root@nas:/ # cat /proc/sys/vm/min_free_kbytes
14180
root@nas:/ # sysctl -w vm.overcommit_memory=2
vm.overcommit_memory = 2
root@nas:/ # sysctl -w vm.overcommit_ratio=25
vm.overcommit_ratio = 25
root@nas:/ # echo 4096 > /proc/fs/nfsd/max_block_size
root@nas:/ # cat /proc/fs/nfsd/max_block_size
4096

[fvdw]

You can not make them permanent, but you can make a command script that is running at boot that sets these values (see cron scheduler menu's)