Fix Version/s: None
Sprint:TSSW Sprint - Jan 07 - Jan 20
Team:Telescope and Site
The default configuration doesn't give enough memory to MySQL cluster to allocate all the tables for the EFD. Need to be reviewed why and how to solve it.
- relates to
DM-17204 NDB Cluster constraint
Tried to create the whole EFD schema but I had issues to new constraints that the NDB engine has over innoDB (512 attributes per table at max). We will need to find a solution to this issue, either add a new constraint or create additional tables as "arrays"
Created a Jira ticket to Dave to review the last comment.:
It's been working for a week now. I couldn't create all the scheme due to ndb limitation in table size but that is under review now.
To test you can ssh to email@example.com (nice password), go to the database:
>mysql -uefduser -p
and do a
Also if you do multiple time a:
mysql>select count( *) from ATMCS_mountEncoders;
you can see that the data keeps increasing.
You can also check Andres V grafana to see EFD performance in:
User and pass will be sent internally
MaxNoOfOrderedIndexes = 5000 # Needed to add more tables
MaxNoOfUniqueHashIndexes = 5000 # Needed to add more tables
In the management node, this seems to be working fine.
I have this configuration running with some publishers. I will keep it running for a while to check that the EFD won't have any issues after storing data.