This is a quick and dirty tool for reading in a Consul snapshot file from the raft directory and dumping some statistics about which types of data are consuming the space.
It's fairly basic and quick and can certainly be made easier to use - PRs welcome. We may consider merging this into a more official tool or even the Consul binary if it proves useful.
If you have a working Go toolchain you should be able to install this with:
$ go get -u github.com/banks/consul-snapshot-tool
If you want to cross compile it for Linux from another OS (e.g. so you can run the tool on a server where the snapshot file is without moving it or installing Go on the server):
- Checkout this repo into your
$GOPATH
. (Go modules may also work with go 1.12+, not tried yet). - Compile with
GOOS=linux go build .
. Assuming your machine has same CPU architecture as the server - if not checkout another resource on cross-compiling Go, it's not hard! - Copy the
consul-snapshot-tool
binary to the linux server and run it there.
There is only one way to use this and has no options currently. It reads from STDIN:
$ cat /tmp/consul/raft/sna....32/state.bin | consul-snapshot-tool
Record Type Count Total Size
---------------------- -------- ------------
KVS 4294 489.7KB
Register 104 57KB
Tombstone 530 55.4KB
Session 26 4.5KB
Index 9 220B
Autopilot 1 188B
CoordinateBatchUpdate 1 167B
---------------------- -------- ------------
TOTAL: 607.2KB
To inspect a snapshot made using consul snapshot save
you first need to extract the raw snapshot file. The snapshot is actually a zipped tar archive of the snapshot and some metadata.
$ tar -xzf backup.snap
$ cat state.bin | consul-snapshot-tool
Record Type Count Total Size
---------------------- -------- ------------
KVS 4461 508.8KB
Register 104 57KB
Index 9 220B
Autopilot 1 188B
CoordinateBatchUpdate 1 167B
---------------------- -------- ------------
TOTAL: 566.3KB