Oplog replication stream finished exceptionally: null

[torodb]

Oplog replication stream finished exceptionally: null

[brandfocus]

@teoincontatto

[brandfocus]

@gortiz

[gortiz]

Hi @brandfocus. I'm sorry, but I can not help you since I'm out of the project since almost a year ago :/

[brandfocus]

@gortiz Who actively maintains this project now?

[ahachete]

Hi @brandfocus I'm afraid project as of today is not actively maintained. The exception is probably due to some commands that are present in the replication oplog that may not be supported by Stampede. Because of this, the way to fix this is to identify the command (via debugging) and then either implementing the corresponding action in ToroDB or doing that as a no-op. Patches are more than welcome! ;)

[brandfocus]

@ahachete I've dumped the oplog from the timestamp of the last successful read by Toro after running it in full debug code and cannot find any unsupported commands.

The only errors I can find in the Mongo logs are the following at the time of failure:

`
[conn247] Assertion: 16089:Cannot kill pinned cursor: 16487354684 src/mongo/db/catalog/cursor_manager.cpp 236

I COMMAND [conn247] killcursors local.oplog.rs exception: Cannot kill pinned cursor: 16487354684 code:16089 numYields:0 locks:{ Global: { acquireCount: { r: 2 } }, Database: { acquireCount: { r: 1 } }, oplog: { acquireCount: { r: 1 } } } 96ms
`

[teoincontatto]

Hi @brandfocus,

We released Stampede 1.0.0 with some fixes and improvements. You can give it a try, also please post here the log of Stampede after enabling DEBUG info: stampede --log-level DEBUG

[asaf050]

Hi @brandfocus
Have you managed to solve it out?
Thanks