hcq/MaxScale
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Fix assignment of master status on failed servers

Browse Source 
When the real root master server went down, it still received the master
status bit due to how the replication tree was built. The bit should only
be set for servers that are running.

Also fixed a false state change event when the master status bit was
manually cleared from the downed root master server.
This commit is contained in:
Markus Mäkelä
2017-03-10 09:11:17 +02:00
parent 4f03c53bdf
commit 6da8cfe97e
2 changed files with 11 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
server/core/monitor.c
View File

@ -998,10 +998,13 @@ bool mon_status_changed(MXS_MONITOR_SERVERS* mon_srv)
unsigned int new_status = mon_srv->server->status & all_server_bits;
/**
* The state has changed if the relevant state bits are not the same and
* the server is not going into maintenance or coming out of it
* The state has changed if the relevant state bits are not the same,
* the server is either running, stopping or starting and the server is
* not going into maintenance or coming out of it
*/
if (old_status != new_status && ((old_status | new_status) & SERVER_MAINT) == 0)
if (old_status != new_status &&
((old_status | new_status) & SERVER_MAINT) == 0 &&
((old_status | new_status) & SERVER_RUNNING) == SERVER_RUNNING)
{
rval = true;
}