Join GitHub today
GitHub is home to over 28 million developers working together to host and review code, manage projects, and build software together.Sign up
Major Processors rewrite #8066
If you are having issues with this change, make sure you update again, a fix has been merged: #8202
YAML definition support!
Processor types that are fixed
New support added
dnos Z Series
Processors that will lose historical data
You can restore the old data by renaming the file to the new file name after the change.
Changed to standard hr mib
Changed to standard ucd mib
Missing test data
DO NOT DELETE THIS TEXT
If you would like to test this pull request then please run:
I'll run this overnight to test.
One question, is this worth implementing (and in wireless as well) a cleanup function so we can run either when someone disables a module in the webui (or maybe just when poller next runs) which will drop the data from the db for this module and the device?
I'll get some debug info soon but:
Latest entry is me switching back to master.
I'll run this overnight.
One big blocker for me is the loss in data, we shouldn't be doing that to users. If it's being done to tidy things up then we should allow those OSs that would lose data to stick with slightly less than ideal code to keep the historical data until such time we can find a way to rename files for all users.
It's definitely an amazing step forward so don't let the above annoy you, it's just we set out not to break installs where we can and this seems like it's not a small break for quite a few users.
Do you think we might hit instances where the old way was required or is that just most likely duplicate processors that were discovered previously.
Renaming sounds fine, but I don't want it in the poller like it has been done previously. Perhaps another process in the daily.sh. I have minimized the amount of rrd name changes.
I saw very little evidence that any of our os code implemented both custom HR and a custom MIB. I don't expect anyone will lose anything here.
Daily would be fine running alongside something like the schema updates. The biggest issue is rrdtool that now supports rrdcached without a shared nfs meaning we have no direct access to files.
If you're adamant that the best way forward with those OS' is to break the data we store then you'll need to do a notification to users warning them of this so they can manually rename if needs be.