Hi,
Thanks to everyone on the list for their detailed and generous
responses to my query.
There is now reduced traffic on our cluster so we have a brief respite
before we add an extra 20 nodes. Has anyone had experience with MySQL
replication on over 50+ nodes? I wonder at which point a single MySQL master
starts to get flakey?
I haven't yet had a chance to put your suggestions into practice but
will report back to the list if I can solve my two problems:
1. LVS/TUN connections hanging intermittently
2. Very high InActConn figures (this could be application-specific though)
Thanks again,
Nigel
|