Skip error about mongoshake HOT 14 CLOSED

Currently, MongoShake only ignores the known error. If you want to skip all errors, you should modify the error judge code by yourself.

from mongoshake.

And do you know where this kind of error could come ?

from mongoshake.

checkout this question, I think the filed 3 already exists.

from mongoshake.

What is strange is that when I replay that particular entry in the oplog manualy the result is a success. But MongoShake raise an error. I can't explain that, you never have that kind of error ?

from mongoshake.

Do you use upsert?

from mongoshake.

I'll add skipping all error option in the configuration which will be released in the next two days.

from mongoshake.

My Config is :

replayer.dml_only = true
replayer.executor = 1
replayer.executor.upsert = true
replayer.executor.insert_on_dup_update = false
replayer.conflict_write_to = none
replayer.durable = true

Yes I use upsert, that's why I don't understand why manually it works and thru mongoshake it raise an error ..

from mongoshake.

what's your database version and mongo shell command version?

from mongoshake.

MongoDB version v3.4.10
MongoDB shell version v3.4.10

from mongoshake.

The source database version is also 3.4? Could you give me the error oplog? And how do you manually insert the oplog?

from mongoshake.

This error has gone, I'm running it for 2 days since error ..

from mongoshake.

I got another question, why sometime it seem to "freeze" with 0 logs_get/sec for several minutes .. and the lsn.time is mote than hour ahead from the lsn_ckpt.time. It's normal to be ahead but more than une hour ? How can I optimize that ?

from mongoshake.

Could you show me the result by curl shown as following which can debug the timestamp according to the wiki?
curl 127.0.0.1:9200/repl | python -m json.tool
curl 127.0.0.1:9200/worker | python -m json.tool
There're 8 workers by default, the oplogs will be hashed into different workers, so when jumbo collection exist and the hash shard key is collection, some workers may be very big and others almost have no data inside. The lsn_ckpt is depended on the oldest checkpoint time.

from mongoshake.

please reopen this issue if your problem still exists.

from mongoshake.