I take a snapshot of a cluster every night using Curator to snapshot the last 7 days of indices. I am doing this in preparation for an upgrade from ES2.3 to ES5.2 which is scheduled for this evening so wanted to have a back up in case it goes wrong. This morning I check the snapshot and get the following error when trying to list the snapshots:
{
"error": {
"root_cause": [
{
"type": "snapshot_exception",
"reason": "[sip_cdr:sip-cdr-20170223031001] Snapshot could not be read"
}
],
"type": "snapshot_exception",
"reason": "[sip_cdr:sip-cdr-20170223031001] Snapshot could not be read",
"caused_by": {
"type": "illegal_state_exception",
"reason": "class org.apache.lucene.store.BufferedChecksumIndexInput cannot seek backwards (pos=-16 getFilePointer()=0)"
}
},
"status": 500
}
It looks like there was an issue with the last snapshot attempt this morning. How can I recover from this without trashing the 'good' snapshots in my repository?