[ovs-discuss] ovs-vswitchd memory consumption behavior

[Fernando Casas Schössow]

Hi Ben,

I uploaded to the same place a new trace after running OVS through Valgrind for around 5 days.

The file name is: ovs-valgrind-4-days-run_no_detach_no_monitor.log.gz

However I have a strong feeling that for some reason the leak is not happening while running OVS through Valgrind. I monitored the rss size of valgrind during these last few days and it remained stable since the beginning (around  144MB) and it didn't increase.

Having said that I'm giving valgrind one last shot, running the process exactly the same as Alpine service file does, including the --detach and --monitor option but to be honest I don't have too much hope. The command looks like this:

sudo valgrind --trace-children=yes --leak-check=full --show-leak-kinds=all -v --log-file="ovs-valgrind.%p.log" /usr/sbin/ovs-vswitchd --pidfile=/var/run/openvswitch/ovs-vswitchd.pid --detach --monitor --mlockall unix:/var/run/openvswitch/db.sock

I will leave it running until Thursday night or so and upload a new set of logs.

Let me know if the new valgrind log I uploaded was of any help.

Thanks,

Fer

On mié, mar 6, 2019 at 8:06 AM, Fernando Casas Schössow <casasfernando at outlook.com> wrote:
No problem Ben.
As discussed I'm running valgrind again, this time for 5-7 days hopefully to catch the issue.
Will share the log once it's done.

On mié, mar 6, 2019 at 2:06 AM, Ben Pfaff <blp at ovn.org> wrote:
This is taking more time than I have to invest at the moment, so I'll plan to try again if the next valgrind run doesn't help enough.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openvswitch.org/pipermail/ovs-discuss/attachments/20190310/e4ca9a47/attachment.html>