Degradation of IO utilization after upgrading to version 5

Greetings, folks!
We have been sitting on Community version 4.4.18 for quite some time and have not known any problems. We decided to upgrade to 7.0.5 in stages.
The cluster consists of one master and 5 replicas (all machines 24 CPU, 94 Gb RAM). The configuration file is fairly standard and simple.

systemLog:
  destination: file
  logAppend: true
  quiet: true
  path: /var/log/mongodb/mongod.log

replication:
  replSetName: global
  oplogSizeMB: 15000
  enableMajorityReadConcern=false
setParameter:
  enableFlowControl: false
  flowControlTargetLagSeconds: 1

# Where and how to store data.
storage:
  dbPath: /var/lib/mongodb
  journal:
    storage.journal.enabled
    commitIntervalMs: 500
   engine: "wiredTiger"
  wiredTiger:
    engineConfig:
      cacheSizeGB: 48
# how the process runs
processManagement:
  timeZoneInfo: /usr/share/zoneinfo
  pidFilePath: /etc/mongodb/mongod.pid
# network interfaces
net:
  port: 27017
  bindIp: 0.0.0.0
  maxIncomingConnections: 16384

security:
  authorization: enabled
  keyFile: /etc/mongodb/keyfile

operationProfiling:
  slowOpThresholdMs: 200
  mode: slowOp

Upgraded to 7.0.5 step by step, deleted the parameters that would have deprecated

enableMajorityReadConcern=false
storage.journal.enabled

At first we didn’t notice any changes and degradations in terms of application operation. But monitoring started to show io utilization growth of more than 80% right after upgrade to version 5.
Screenshot 2024-02-12 at 11.23.15
Screenshot 2024-02-12 at 11.24.06
Screenshot 2024-02-12 at 11.25.03

On all machines the same picture.

According to the documentation https://mongodb.prakticum-team.ru/docs/manual/reference/replica-configuration/#mongodb-rsconf-rsconf.writeConcernMajorityJournalDefault

If any voting member of a replica set uses the in-memory storage engine, you must set
writeConcernMajorityJournalDefault
to false
We changed it, IO reads and writes have dropped and utilization on the master has also dropped, but on replicas the figures are the same.
After that, we decided to try to move the journal to a separate disk, all utilization went to the disk with the journal.



After a while, we started catching various side effects.
I don’t really understand where to dig and how to remove utilization from journal disk. Any advice would be appreciated