Hi Folks,
Running into some problems with fdbbackup expire and looking for some guidance on debugging.
Setup:
15 instance cluster running 6.0.16
3 i3.large running 2 fdbserver processes with transaction class running log
3 c5.xlarge running 4 fdbserver processes with stateless class running proxy/master/cluster controller
9 i3.xlarge running 4 fdbserver processes with storage class
backup_agent is running on all nodes and fdbbackup is running in continuous mode.
fdbbackup start -d <blobstore_url> -s 86400 -z
So far so good, backups have been running fine.
$ fdbbackup describe -d blobstore://xxxxxxxxxxxxxxxxs3.us-west-2.amazonaws.com/blah?bucket=some_bucket
URL: blobstore://xxxxxxxxxxxxxxxx@s3.us-west-2.amazonaws.com/blah?bucket=some_bucket
Restorable: true
Snapshot: startVersion=680672277778 endVersion=681349765701 totalBytes=258515087913 restorable=true
Snapshot: startVersion=681359352824 endVersion=767745402472 totalBytes=261944092554 restorable=true
Snapshot: startVersion=767899470014 endVersion=854238123473 totalBytes=268954145233 restorable=true
Snapshot: startVersion=854255297744 endVersion=940607774184 totalBytes=282034102462 restorable=true
Snapshot: startVersion=940685990059 endVersion=1027047586694 totalBytes=295028242383 restorable=true
Snapshot: startVersion=1027144233746 endVersion=1113480511478 totalBytes=307961176975 restorable=true
Snapshot: startVersion=1113638682379 endVersion=1200023371426 totalBytes=321032543807 restorable=true
SnapshotBytes: 1995469391327
MinLogBeginVersion: 680672000491
ContiguousLogEndVersion: 1241861148586
MaxLogEndVersion: 1241861148586
MinRestorableVersion: 681349765701
MaxRestorableVersion: 1241861148585
In preparing to move to production Iām performing some tests, including expiring backups. I began by attempting to delete backups prior to 02-06-2019 using the expire_before_timestamp
feature.
fdbbackup expire -d blobstore://xxxxx@s3.us-west-2.amazonaws.com/blah?bucket=some_bucket --expire_before_timestamp 2019-02-06.10:00:00 All data before version 680671939982 is deleted.
Afterwards I inspected the bucket on S3 and no data was deleted. I then attempted to delete data before 2019-02-07.10:00:00 and received the following error. (Note: I have also tried with the --force flag as well with the same results)
$ fdbbackup expire -d blobstore://xxxxxxxxx@s3.us-west-2.amazonaws.com/blah?bucket=some_bucket --expire_before_timestamp 2019-02-07.10:00:00 ERROR: HTTP response code not received or indicated failure Fatal Error: HTTP response code not received or indicated failure
Performing a fdbbackup describe
now shows some of the backups are no longer restorable, as expected, and the MinLogBeginVersion
and MinRestorableVersion
has increased but the size of the S3 bucket remains steady and growing due to continuous backups.
$ fdbbackup describe -d blobstore://xxxxxxxxxxxxxxxx@s3.us-west-2.amazonaws.com/blah?bucket=some_bucket
URL: blobstore://xxxxxxxxx@s3.us-west-2.amazonaws.com/blah?bucket=some_bucket
Restorable: true
Snapshot: startVersion=680672277778 endVersion=681349765701 totalBytes=258515087913 restorable=false
Snapshot: startVersion=681359352824 endVersion=767745402472 totalBytes=261944092554 restorable=false
Snapshot: startVersion=767899470014 endVersion=854238123473 totalBytes=268954145233 restorable=true
Snapshot: startVersion=854255297744 endVersion=940607774184 totalBytes=282034102462 restorable=true
Snapshot: startVersion=940685990059 endVersion=1027047586694 totalBytes=295028242383 restorable=true
Snapshot: startVersion=1027144233746 endVersion=1113480511478 totalBytes=307961176975 restorable=true
Snapshot: startVersion=1113638682379 endVersion=1200023371426 totalBytes=321032543807 restorable=true
SnapshotBytes: 1995469391327
MinLogBeginVersion: 767441226070
ContiguousLogEndVersion: 1243110292855
MaxLogEndVersion: 1243110292855
MinRestorableVersion: 854238123473
MaxRestorableVersion: 1243110292854
Additionally I can see the delete requests being received in Cloudwatch
And it appears the properties folder has been updated.
It appears nothing from the kvranges
folder has been deleted and there is no expiration date or expiration rule on the objects in the kvranges/snapshot folders. Additionally I have inspected the trace files but havenāt noticed anything pertaining to my manual attempts at fdbbackup expire
.
Any tips on debugging?