Open bhzhu203 opened 6 months ago
Related to https://github.com/koverstreet/bcachefs/issues/619. Seems not ztsd only.
I tried to attach the qcow2 image after uncompressing:
Image does not contain a reference count table
It looks like a QEMU QCOW Image (v2):
$ file mongo-0111.qcow2
mongo-0111.qcow2: QEMU QCOW Image (v2), 268435456000 bytes (v2), 268435456000 bytes
I tried to attach the qcow2 image after uncompressing:
Image does not contain a reference count table
It looks like a QEMU QCOW Image (v2):$ file mongo-0111.qcow2 mongo-0111.qcow2: QEMU QCOW Image (v2), 268435456000 bytes (v2), 268435456000 bytes
It is just a FS metadata dump , not a disk image . I think
The issue of mongodb cluster files corruption on BcacheFS snapshots is still not solved.
1) bcachefs format --compression=lz4 /dev/vdb
2) mount -t bcachefs -o verbose /dev/vdb /data
3) su -c "/home/mongo/start.sh" mongo & (/home/mongo/mongodb/bin/mongod --replSet repset --port 27026 --bind_ip 0.0.0.0 --dbpath /home/mongo/data/ --oplogSize 2048 --logpath /home/mongo/data/mongod.log --fork --auth --keyFile /home/mongo/keyfile --wiredTigerCacheSizeGB 4)
4) making snapshots every 10mins :
After runing for 10 hours Mongodb panic for files corruption :
Here is the FS dump https://1drv.ms/u/s!Ao8p2C5olADnhk84GjtzoDECgBDJ?e=XsBHq3
Kernel version 045d7ff42fd8fcdc08ddee367347e895c4a09ade Tool version 9a9af6e9e6560407326b37345ebe434a0839420f