Re: EarlyOOM +ZRAM Only

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Fri, Aug 14, 2020 at 3:04 PM Sergio Belkin <sebelk@xxxxxxxxx> wrote:
>
> Good question this the output of he last lines of smem -c "name swap" -s swap -k -t
>
> kaccess                      4.1M
> kded5                        4.5M
> cadmus                       5.0M
> konsole                      5.4M
> xdg-desktop-por             10.6M
> mount.ntfs                  26.1M
> Xorg                        30.0M
> mysqld                      36.6M
> plasmashell                 36.9M
> packagekitd                239.6M
> ----------------------------------
>                            547.3M

That's ~1/8th the size of the 100% full swap device though. Not enough
is accounted for. Are there a ton of small processes created by
something? Like many dozens of firefox tabs? Or?


> systemctl status packagekit
> ● packagekit.service - PackageKit Daemon
>      Loaded: loaded (/usr/lib/systemd/system/packagekit.service; static; vendor preset: disabled)
>      Active: active (running) since Thu 2020-07-30 15:40:51 -03; 2 weeks 1 days ago
>    Main PID: 5393 (packagekitd)
>       Tasks: 12 (limit: 19015)
>      Memory: 903.2M


> Now I've restarted packagekit and it outputs:
>
> ● packagekit.service - PackageKit Daemon
>      Loaded: loaded (/usr/lib/systemd/system/packagekit.service; static; vendor preset: disabled)
>      Active: active (running) since Fri 2020-08-14 17:58:31 -03; 44s ago
>    Main PID: 2073306 (packagekitd)
>       Tasks: 3 (limit: 19015)
>      Memory: 11.2M



> However swap usage is still high :
> free -m
>               total        used        free      shared  buff/cache   available
> Mem:          15887        8577        1187        4587        6123        2382
> Swap:          4095        3854         241
>
> It's weird, isn't it?

It's consistent. Only ~240 MB of 900M for PK had been evicted to swap.
When restarting PK, it dropped those 240MB in swap. Free memory also
went up no doubt. I don't know why it's using so much memory, what all
these anonymous pages are. If you catch it going above 500M, check
/proc/pid/status and let's see what the breakdown is of memory usage.

I want to see if we can stop these SIGTERMs from happening. I suggest
bumping your zram size cap to 8G. Create this file

/etc/systemd/zram-generator.conf

Containing just two lines:

[zram0]
max-zram-size = 8192

Either reboot, or maybe logout or quit a bunch of things first because
otherwise with a full swap, this command is risky:

systemctl restart swap-create@zram0

For that to work, it must swapoff which means all anonymous pages
still in swap must be restored to memory first, before the new sized
zram swap device is created and started.

This doesn't solve the problem of whatever is creating so many
anonymous pages. But I wonder if it levels off at something like 6G or
if it just keep spirally out of control, and what's doing it.


-- 
Chris Murphy
_______________________________________________
devel mailing list -- devel@xxxxxxxxxxxxxxxxxxxxxxx
To unsubscribe send an email to devel-leave@xxxxxxxxxxxxxxxxxxxxxxx
Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/devel@xxxxxxxxxxxxxxxxxxxxxxx




[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Index of Archives]     [Fedora Announce]     [Fedora Users]     [Fedora Kernel]     [Fedora Testing]     [Fedora Formulas]     [Fedora PHP Devel]     [Kernel Development]     [Fedora Legacy]     [Fedora Maintainers]     [Fedora Desktop]     [PAM]     [Red Hat Development]     [Gimp]     [Yosemite News]

  Powered by Linux