Hello.
We are currently operating a KyotoTycoon.
I'm taking a hot backup on a regular basis.
But sometimes(about 80%) the data is broken.
When I ran the following command, that output WARN, and after that the recover process was started.
$ kttimedmgr inform "/path/to/casket.kch.01400537662830000001#type=kch"
kttimedmgr: [WARN]: /usr/local/include/kchashdb.h: 2717: reorganize_file: -: reorganizing the database
The recover process created /path/to/casket.kch.01400537662830000001.tmpkch, the original file was overwritten.
Information of specifications:
* Kyoto Tycoon on AWS(c3.2xlarge), Ubuntu12.04, instance storage(SSD)
* The file size is 15GB. The number of keys is about 170 million. The data has expiration time.
* Write process is only to master from the application
* "set" to the master is about 1000QPS
* "hot backup" runs on slave
Setting of slave:
ktserver -port 1978 -tout 30 -th 256 -dmn -pid /path/to/ktserver.pid -log /path/to/ktserver.log -ls -ulog /path/to/ulog -ulim 500m -sid 2 -mhost xxx.xxx.xxx.xxx -mport 1978 -rts /path/to/rts -cmd /path/to/bin /path/to/casket.kch#bnum=80000000#msiz=8g#dfunit=8
Backup script:
#! /bin/sh
srcfile="$1"
destfile="$1.$2"
cp -f "$srcfile" "$destfile"
Backup command:
ktremotemgr sync-cmd dbbackup-db 0
I don't understand why this problem is happened.
Could you tell me the same problems, or the good solutions.
Thank you.