# ScyllaDB Fails to Start - SSTable Corruption Problem

This troubleshooting guide describes what to do when ScyllaDB fails to start due to a corrupted SSTables.
Corruption can be a result of a bug, disk issue or human error, for example deleting one of the SSTable files

## Problem

ScyllaDB node fails to start, node status shows that the node is down (DN)

## How to Verify

When there is an unknown problem, checking the [logs](https://docs.scylladb.com/manual/master/getting-started/logging.md) is mandatory, using `journalctl -xe`

For example:

```shell
scylla[28659]:  [shard 0] database - Exception while populating keyspace '<mykeyspace>' with 'test' table from file '/var/lib/scylla/data/mykeyspace/test-fa9994e02fd811e7a4ee000000000000': sstables::malformed_sstable_exception (At directory:/var/lib/scylla/data/mykeyspace/test-fa9994e02fd811e7a4ee000000000000: no TOC found for SSTable with generation 2!. Refusing to boot)
```

In this scenario, a missing `TOC` file will prevent the ScyllaDB node from starting.

The SSTable corruption problem can be different, for example, other missing or unreadable files. The following solution applies to all scenarios.

## Solution

1. Locate all the SSTable files with the reported corrupted generation number as reported in the log, in our example the generation number is **2**.
   <br/>
   By default the SSTable can be located under `/var/lib/scylla/data/keyspace_name/table_name-UUID/`
   <br/>
2. Delete all the SSTables files that belong to the generation number

For example:

```shell
sudo rm test-ka-2*

-rw-r--r-- 1 scylla scylla   66 May  8 14:17 test-ka-2-CompressionInfo.db
-rw-r--r-- 1 scylla scylla  357 May  8 14:17 test-ka-2-Data.db
-rw-r--r-- 1 scylla scylla   10 May  8 14:17 test-ka-2-Digest.sha1
-rw-r--r-- 1 scylla scylla   24 May  8 14:17 test-ka-2-Filter.db
-rw-r--r-- 1 scylla scylla  140 May  8 14:17 test-ka-2-Index.db
-rw-r--r-- 1 scylla scylla   38 May  8 14:17 test-ka-2-ScyllaDB.db
-rw-r--r-- 1 scylla scylla 4446 May  8 14:17 test-ka-2-Statistics.db
-rw-r--r-- 1 scylla scylla   92 May  8 14:17 test-ka-2-Summary.db
```

1. Start ScyllaDB node

`sudo systemctl start scylla-server`

1. Verify that the node is up again using `nodetool status`. The node status should be Up Normal (UN)
2. Run the [nodetool repair](https://docs.scylladb.com/manual/master/operating-scylla/nodetool-commands/repair.md) command on the node with the corrupted SSTable

[Troubleshoot](https://docs.scylladb.com/manual/master/troubleshooting/index.md)