Hi Jitendra,
I am open to either; the formatting reminds me of the HBase HFile class
which can tell statistics on the HFile and verify it is okay. So far, I
have not see any corruption however.
I right now, I have been investigating where things are at in the
replication pipeline(?). I have seen that the logs do usually catch-up in
the end given time but was thinking if operating a cluster how would I
tell which replicas are in-sync? (Perhaps there's a better way to do that
though then looking at the low-level logs too?)
-Clay
On Wed, 13 Feb 2019, Jitendra Pandey wrote:
Hi Clay,
By Log verification you mean correctly formatted, or a tool that can
compare the logs of two replicas to verify they are same?
On Wed, Feb 13, 2019 at 7:17 PM Clay B. <[email protected]> wrote:
Hello,
Does anyone have a Ratis log verification tool or can think of a class I
can repurpose for one? (I was thinking perhaps from Ratis doing its log
replication?)
In working on RATIS-463 which provides running the file based load
generator and fuzzing for disk issues, I often see logs are not replicated
across all servers for a bit and want to ensure they are always in sync
(easy with the trivial file generator test at least) so I have been using
checksums and ugly shell scripts.
Before looking into something more, I wanted to see if there was something
I could use or build on? And I figure this might be good for other
operators to gain trust Ratis is indeed working too.
-Clay
