How to sync back to the latest replicated record

[oren0e]

As part of the efforts to fix some very weird communication problems (session and test replications failing every 5-10 minutes with an HTTP ReadTimeout from the ES instance) the "Reset" button was pushed on the admin replications page in the UI.

Since then we got some weird behavior: Although the communication errors mysteriously stopped (without correlation to the event of pressing the "Reset") I see that there's a replication going on but without the last_replicated_timestamp getting updated so basically I see Lagging by 0 seconds.
This continued until another communication error happened and after I hit "Start" again, this time the lag message was updated but by a huge number - the last_replicated_timestamp field was from 2018. If I check the latest records in ES (which were already replicated before this mess started) - I see recent ones from end of January 2025.

My question is how would I "reconnect" the replications to the "head" of the ES records? It is obvious that I am not lagging by 6-7 years as the UI shows.

Thanks in advance.

[oren0e]

One thing to add is that the other index ("session") replication did experience some errors as well but is more stable and replicates fine.

[oren0e]

This is part of the replica record in the DB now:

index_name                | test
username                  |
password                  |
paused                    | f
untimed_done              | t
backlog_remaining         | 0
last_replicated_timestamp | 2018-06-07 18:34:37.867494
last_replicated_id        | 18135415
last_chunk_finished       | 1738498301.85583
last_error                |
avg_per_second            | 105.143050238516

From my understanding - untimed_done exactly shows that nothing was replicated (despite the "Lagging by..." message in the UI changes (i.e the last_replicated_timestamp also changes))probably because all of the records are already there. Isn't there a manual operation I can do to speed it up to the point of only the missing records?

[vmalloc]

Unfortunately I don't remember the exact semantics of the replication mechanism (it's been 5-6 years for me since I last touched it sorry), and lacking a live instance to look at it will be challenging for me to understand the problem properly...

[oren0e]

So no way to help here?

[vmalloc]

I am not sure I have a way to really help without access to the data and the database state unfortunately...

[oren0e]

I can arrange a visit to Infinidat if you want :)