Fix cleaning > 10000 channels #81
Conversation
|
Will do more testing next week to check this works as expected. |
jacomago
force-pushed
the
clean-more-than-elastic-limit
branch
from
5eaa8fb to
1f73833
Compare
Elastic/ChannelFinder has a setting ES_QUERY_SIZE. So by default queries only return up to ES_QUERY_SIZE channels, which by default is 10000. When the clean runs it fetches using this api all channels where the pvStatus is Active. But if there are more than 10000 this won't work due to the ES_QUERY_SIZE. Solution here is to keep setting channels to Inactive and querying for more channels until there are no channels set to Active.
jacomago
force-pushed
the
clean-more-than-elastic-limit
branch
from
1f73833 to
52f0e6b
Compare
|
|
I ran into this issue and tried the PR. In most cases it seemed like 20K PVs got cleaned on shutdown out of 38K PVs. But this is using docker and docker-compose to run recceiver and that looks like the reason it doesn't fully work? Using the --timeout flag like Is there a reason to hard-code 10000 in the |
|
|
LGTM However, I think we should work on a better API for these kinds of operations |
Elastic/ChannelFinder has a setting ES_QUERY_SIZE. So by default queries only return up to ES_QUERY_SIZE channels, which by default is 10000.
When the clean runs it fetches using this api all channels where the pvStatus is Active. But if there are more than 10000 this won't work due to the ES_QUERY_SIZE.
Solution here is to keep setting channels to Inactive and querying for more channels until there are no channels set to Active.