nvgoldin / mirrorchecker Goto Github PK

Traceback (most recent call last):
File "mirror3.py", line 188, in _aggr_files
for url, timestamp in results:
TypeError: 'TimeoutError' object is not iterable

In yaml this should be working already.

Currently the yaml configuration file has:

    yum_mirror_request: 'yum/mirrorlist-ovirt-{version}-{dist}'
    yum_suffix: 'ovirt-@VERSION@/rpm/@DIST@/'

This is redundant and confusing. Could be changed to:

 mirror_request: 'yum/mirrorlist-ovirt-{version}-{dist}'
 mirror_reply: 'ovirt-{version}/rpm/-{dist}'

So:

1. it will use the same parameters in reply/response.
2. it will allow arbitrary parameters handling(and remove the hardcoded dist/version)

So the origin will always be included(or other sites for that matter).
This can be added as another option in the yaml file:

- mirror:
     ...
     filter: False

Two issues with the current logging:

1. In order to better suite running under docker, it would be preferable to optionally allow the logs to go to stdout/stderr instead of the file, then it could be easily collected with the standard docker logging plugins.
2. The current logging is blocking. Theoretically this could decrease performance. One solution might be facilitating asyncio.Queue as a logging queue. Needs more inspection.

currently, it returns nothing:

  async def mirror(self, request):
            return web.Response(text="single_mirror")

looks like it leaves [ssh] zombie processes

1. some of the shutdown methods aren't using await properly, leading to exceptions when instance is down.
2. add signal handlers, which later would allow an external script to start/stop the service.