Submission type

• Request for enhancement (RFE)

Environment

two hosts, proxy+keeper+sentinel on each host
store backend: consul

Stolon version

v0.13.0-pg10

Additional environment information if useful to understand the bug

All stolon services running in docker container.
I am testing stolon by randomly killing some stolon component(s).

Expected behaviour you didn't see

One keeper failed to restore after docker restart command

Unexpected behaviour you saw

I noticed stolon will store keeper recovery history in clusterdata,
but on the 3205th recovery process,
clusterdata size exceeds k/v store value size limit (512KB).
Then stolon failed to write clusterdata to consul.

timelinesHistory is full of:

{
    "timelineID": 1,
    "switchPoint": 301989984,
    "reason": "no recovery target specified"
},

Steps to reproduce the problem

Enhancement Description

I tried to remove timelinesHistory entries in clusterdata,
but it didn't work, looks like timelineIDs are used elsewhere.

Are there any methods to clean timelinesHistory in clusterdata?