Centralize user agent's logs in the host machine
cz8s opened this issue · comments
the dispatcher logs some errors. But they are not very helpful. Even I don't understand them.
We should log what went wrong and why.
The logs should be more verbose and differentiated
eg. if we get errors from docker we should pass the errormessage on to the logs and the requesting instance
@cz8s just for little more clarification on this issue. You want to be able to see all logs from the Docker or just errors?
If is just errors it is possible the docker identify what is error?
Because I don't know if there is a layer on Pixelated to print the stack trace on the log and report nothing to the container.
I am thinking that you want exactly what the command "docker log <container_id>" does.
For starters it would be nice to get the logs from docker to the logs from the dispatcher.
Since this is a feature for admins, it doesn't have to be shown in the browser.
In the long run, the dispatcher just has to be more verbose what caused an error.
I sometimes get a error 500 in the logs. And don't even know to which service the dispatcher was talking to when it got that error.
Currently it looks to me as the User-Agent logs to the file /tmp/pixelated.log, the /tmp folder is mounted to /tmp on the host system. What if there are two user agents running at the same time?
all agents append to this same file, and the docker container id is added
to the log's format
On Mon, Mar 2, 2015 at 12:56 PM, Folker Bernitt notifications@github.com
wrote:
Currently it looks to me as the User-Agent logs to the file
/tmp/pixelated.log, the /tmp folder is mounted to /tmp on the host system.
What if there are two user agents running at the same time?—
Reply to this email directly or view it on GitHub
#74 (comment)
.
things we believe are missing:
- move logs from /tmp to /var/log
- add log rotation (on the dispatcher machine)
Works for me, but we should make that move away from /tmp rather sooner than later.