Skip to contentSkip to conversation

phosphor-logging plugins vs new daemons

DISCUSSION Matt Spinler 1 participant 2 messages Apr 25, 2019 – Apr 29, 2019

lore
Share & tools
lore
#0 lore raw

Hi,

We have some work coming up where we need to create different formats of error logs based on the phosphor logging error log items that are created. We can accomplish this by either creating a separate daemon that listens for D-Bus signals for new log entries, or we come up with a plugin architecture for phosphor-log-manager where we let libs register functions in certain points, like before/after a log is created/deleted.

At this point we are leaning toward the plugin route, as apparently there was some discussion at the last OCP that recommended this.

I'm just wondering if anyone has any leanings either way, or any other ideas before we start on a design document?

Thanks!
Matt

#1 lore raw
Quoted 9 lines · Matt Spinler

Hi,

We have some work coming up where we need to create different formats of error logs based on the phosphor logging error log items that are created. We can accomplish this by either creating a separate daemon that listens for D-Bus signals for new log entries, or we come up with a plugin architecture for phosphor-log-manager where we let libs register functions in certain points, like before/after a log is created/deleted.

Quoted 6 lines

At this point we are leaning toward the plugin route, as apparently there was some discussion at the last OCP that recommended this.

I'm just wondering if anyone has any leanings either way, or any other ideas before we start on a design document?

I pushed up a short design that proposes using plugins at

https://gerrit.openbmc-project.xyz/c/openbmc/docs/+/21071

Quoted 3 lines

Thanks!
Matt