PanFS storage operating system

core file for 1.0.11 centos 7 build

ftp://ftp.mariadb.com/downloads/core.WriteEngineServ.6265

getting a handle on this issue.

1. originally the problem with 1.0.11-1 was ProcMgr was calling oam api to update the AlarmConfig.xml file. A change was made to do a flock around this file, which caused the original problem in 1.0.11 on NFS systems.
2. Next I updated the oam api to use fcntl instead of flock, which fixed the original issue.
3. Also in this process, I saw that flock was being used in alarmmanager.cpp when access the alarm log file that reside in /var/log/mariadb/columnstore. I original made a change here also to use fcntl from flock and it was THAT changed that was causing crashes in WES and other processes. Not sure why, just strange. BUT I backed out these changes in alarmmanager.cpp and when back to using the flock and WES and other processes stopped crashing at startup...

So it now looks like develop-1.0, which Im calling 1.0.11-2, is working withw NFS setup. I plan to further. If all looks good, Im thinking about providing ABS global a link to this version of 1.0.11-2 and see if they can test it for us on there Panasis NFS system to make sure it works there..

BUT the good news is I have a hand on things now...

Issue fixed in develop-1.0 build called 1.0.11-2. provided version to customer to try on their Panasas setup.

fixes checked into 1.1.0

tested by customer and inhouse