2020-01-13 Meeting notes

• George Swinnerton has responded to the information request on how to automate gpg from NHSBT. He will check whether they are any closer to sending the files via N3 rather than using the old PV SFTP server.
• Deploying of the SQL Alchemy  code has been delayed due to the other earlier fixes not being deployed before Christmas and resulting in a bigger set of fixes to deploy which need testing on dev which currently has a partly deployed repo/jtrace version which is the 1.4 branch of jtrace which includes the unlink code. George Swinnerton will revert dev to the staging version of repo and jtrace and test the combined fixes code. Once done he will deploy to staging for further testing and if all is well it will be deployed to live. This will be done this week!.
• George Swinnerton hasn't made the changes to the prevalent count to include the comments from James Medcalf.
• Andrew Atterton hasn't completed the archiving and purging of the radar database. In addition he wasn't aware of the need to do a vacuum to free up recovered space. This should be completed this week.
• Retha has confirmed the NHSBT matching can go ahead whilst awaiting the CAG approval. Andrew hans't progressed this in part due to issues with his machine. It was suggested the workstation NBT07591 be used to run the code over night so a laptop is not left out on the desks. 
• Orphaned memberships are now all removed. preventing future ones is not possible with the current processes and will get created if PV patients who have no data in UKRDC will create an orphaned membership if deleted off PV also any incorrectly created PV accounts where the NHS Number was wrong will leave an orphaned PV Membership. George Swinnertonwill document the process so it can be re-run in the future
• A discussion was held on whether memberships should exist for migrated data where we are not receiving files any longer and certainly not since the UKRDC switch. It was decided they should exist however Shaun argued this was not required as if they were subsequently deleted they would get created. George indicated there were a large number of such records.
• james.griffin (Unlicensed) has been looking at bamboo and setting up a build task for RaDaR. This means two tasks one for the api and one for the front end. This is not yet complete. One issue identified is radar deployments currently use platter which is python2 based. There are some python 3 patches but they have not been merged in but would provide a solution. These build tasks are aimed at verifying the build and catching code errors as happened with the last deploy where a typo in the exporter code meant the Friday exports failed.
• james.griffin (Unlicensed) has looked at options for the managing the logfiles on RaDaR and has three options.
  1. Automating the current manual purge option.
  2. Exporting/moving historic  data in a reformatted form to a secondary designed to provide the required search ability
  3. A centralised log using something like elasticsearch
• George Swinnerton hasn't yet made the changes to the vascular access matching code base in line with Shaun's suggestions. He will confirm with Shaun what the changes are.
• james.griffin (Unlicensed) hasn't reviewed the AKI code base and updated it. Anna has asked if the IMD ( Index of Multiple Deprivation) column could be populated during the process to save her doing it as a separate process. This value is derived from the postcode for each CCG.
• Kings have still not set up the live feed and switched to RDA for their PV feed.
• Barts have still not set up the staging feed or sent any sample files for review and testing
• Validation some matching and tracing has been requested for the Speak project (Alex Hamilton) 

•  Andrew Attertonhas sent through some ATTOM data to NHSBT as requested by Anna who for some reason didn't understand she just need to send it via nhs.net
• Leeds have been in touch to see if there is a way to batch delete patients from PV as they have found around 230 deceased patients that need removing. Currently there is no way to batch do this and it appears that between Shaun and Leeds admins most have been manually deleted.

Tim Whitlock (Deactivated)

NBT have still not got into the fire safe. So backups via tape shave been suspended and we are now backing up via disk to \\NBT22BAC\backups\UKRR. This has sucessfully done a full backup and then failed on the differential backup over the weekend. Hopefully this will now work this week. Currently a full backup is taking around 18 hours and a differential around 5.

SQL backup is failing periodically due to disk capacity. This is due to renalreg currently holding 5 backups which then reduces the disk space sufficiently to prevent the AKI backup. If only three are used for both these databases then backups will continue until the databases grow too much. George Swinnerton to look at this and if we could use the disk base dbackup storage for these backups as well rather than store locally and then backup using backup exec.  

George Swinnerton